JP7689872B2 - Information processing device and information processing method - Google Patents

Description

The present invention relates to an information processing device and an information processing method.

Conventionally, systems are known that are used when a caregiver provides care to a person being assisted. Patent Document 1 discloses a method in which a sensor is placed in a living space, and information is generated about the condition of the resident living in the living space based on changes over time in the detection information acquired by the sensor.

JP 2021-18760 A

To provide an information processing device and information processing method that can appropriately utilize tacit knowledge.

One aspect of the present disclosure includes a processing unit that receives a registration request for know-how information that is used in care, the know-how information including information in which output information output to a caregiver is associated with input information for outputting the output information, and a storage unit that stores the know-how information based on the registration request, and the processing unit is related to an information processing device that determines the importance of the know-how information based on association information that associates changes in status information that indicates the status of a person being assisted with know-how information used in providing care to the person being assisted.

1 illustrates an example of an information processing system including the information processing device of the present embodiment. 1 shows an example of a server system configuration. 13 shows an example of the configuration of a terminal device. FIG. 4 is a diagram for explaining the flow of know-how information registration processing. Examples of know-how information. FIG. 11 is a diagram for explaining a process flow for associating devices and the like with know-how information. An example of the screen displayed when tagging correct data. Example of registration information. Example of registration information. FIG. 4 is a diagram for explaining the flow of search processing. Example of list information. FIG. 11 is a diagram for explaining a process flow for using know-how information. FIG. 4 is an explanatory diagram of a first similarity determination process. FIG. 4 is an explanatory diagram of a first similarity determination process. 13 is an example of a survey item for determining status information. 13 is an example of a survey item for determining status information. FIG. 11 is an explanatory diagram of a process for determining importance based on a change in state information and used know-how information. 1 is an example of a user page that presents recommended know-how information based on importance. An example of neural network configuration. Example of neural network input and output. 11 is a flowchart illustrating a learning process. 11 is a flowchart illustrating an inference process. An example of neural network input and output when estimating evaluation results for multiple survey items. An example of know-how information used in meal assistance. An example of know-how information used in excretion assistance. An example of know-how information used in transfer and mobility assistance. Examples of input data for each survey item. Examples of input data for each survey item. Examples of input data for each survey item. Examples of input data for each survey item. Examples of input data for each survey item. 6A and 6B are diagrams for explaining a process of determining whether or not sensing data is insufficient. Proposals for cases where sensing data is insufficient, and examples of items for which sensing data can be obtained. 13 is an example of association between know-how information and multiple devices. FIG. 13 is an explanatory diagram of a process for identifying a recommended supplier based on know-how information. An example of a screen showing recommended suppliers. An example of a screen presenting information about a given user. An example of a screen presenting output information regarding end-of-life care. An example of a screen on nursing care software that displays marks related to output information. An example of a screen on nursing care software that displays output information related to end-of-life care. An example of a system configuration for providing meal assistance. 13 is an example of a screen displayed on a first terminal device. Another example of an object for output information. 13 is an example of a notification screen displayed on the second terminal device. 13 is an example of a status screen displayed on a second terminal device. An example of a screen presenting information about a given user. Examples of information required to obtain appropriate device data (input information). 13 is an example of a system configuration for providing assistance in the event of a fall. 13 is an example of a screen displayed on a third terminal device. 13 is an example of a notification screen displayed on a fourth terminal device. 13 is an example of a notification screen displayed on a fourth terminal device. 13 is an example of a status screen displayed on a fourth terminal device. 13 is an example of a detailed screen displayed on a fourth terminal device.

The present embodiment will be described below with reference to the drawings. In the drawings, identical or equivalent elements are given the same reference numerals, and duplicated explanations will be omitted. Note that the present embodiment described below does not unduly limit the contents described in the claims. Furthermore, not all of the configurations described in the present embodiment are necessarily essential components of the present disclosure.

1. System Configuration Example FIG. 1 is a configuration example of an information processing system 10 including an information processing device according to the present embodiment. The information processing system 10 according to the present embodiment provides information to care staff so that appropriate care can be provided regardless of the level of expertise of the care staff, for example, in a care facility, by digitizing the "intuition" or "tacit knowledge" of the care staff for work performed by the care staff in the care facility so that the care staff can provide appropriate care regardless of the level of expertise of the care staff. In the following, for the sake of simplicity, "intuition" and "tacit knowledge" are also simply referred to as tacit knowledge. In addition, a method for accumulating and using tacit knowledge regarding care in a care facility or hospital will be described below, but the present embodiment is not limited to this. For example, in home care where care is provided outside the facility, tacit knowledge regarding the care of a helper may be accumulated and used. Furthermore, the method of the present embodiment provides tacit knowledge of experts and the like to other users, and is widely applicable to situations where tacit knowledge is used, such as schools, factories, and companies.

The information processing system 10 shown in FIG. 1 includes a server system 100, a terminal device 200, and a headset 300. In FIG. 1, two terminal devices 200-1 and 200-2 are illustrated as the terminal devices 200, and headsets 300-1 and 300-2 are illustrated as the headsets 300. However, the configuration of the information processing system 10 is not limited to that shown in FIG. 1, and various modifications such as omitting some components or adding other components are possible. For example, the number of terminal devices 200 and headsets 300 may be three or more. The headset 300 may not be used, and may be substituted by the sound input unit or sound output unit of the terminal device 200. The same applies to FIG. 2 and FIG. 3 described later, etc., in that modifications such as omissions and additions to the configuration are possible.

Hereinafter, when there is no need to distinguish between multiple terminal devices 200, they will simply be referred to as terminal devices 200. Similarly, when there is no need to distinguish between multiple headsets 300, they will simply be referred to as headsets 300.

The information processing device of this embodiment corresponds to, for example, the server system 100. However, the method of this embodiment is not limited to this, and the processing of the information processing device may be executed by distributed processing using the server system 100 and other devices. For example, the information processing device of this embodiment may include the server system 100 and a terminal device 200. Below, an example in which the information processing device is the server system 100 will be described.

The server system 100 is connected to the terminal device 200 and the headset 300, for example, via a network. The network here is, for example, a public communication network such as the Internet, but may also be a LAN (Local Area Network) or the like. For example, the terminal device 200 and the headset 300 are devices used by staff at a nursing facility or nurses at a hospital while on duty. Note that the headset 300 is not limited to a device that can be directly connected to the server system 100, and may also be a device that is connected to the server system 100 via the terminal device 200.

The terminal device 200 is not limited to a device that directly communicates with the server system 100. For example, a relay device (not shown) may be provided in a nursing facility or the like. The relay device is a device that can communicate with the server system 100 via a network. The terminal device 200 and the headset 300 may be connected to the relay device using a LAN in the nursing facility and connected to the server system 100 via the relay device. For example, it is assumed that multiple terminal devices 200 and multiple headsets 300 are used simultaneously in a nursing facility or the like. The relay device may perform a process of selecting the terminal device 200 or headset 300 to which information from the server system 100 is to be transmitted. Alternatively, the relay device may be an administrator terminal used by an administrator of the nursing facility and may operate based on an operation input by the administrator. For example, information from the server system 100 may be displayed on the display unit of the relay device, and the administrator who views the display result may select the terminal device 200 or headset 300 to which the information is to be transmitted. As described above, the information processing device of this embodiment can be modified in various ways, and for example, the above relay device may be included in the information processing device.

The server system 100 may be one server, or may include multiple servers. For example, the server system 100 may include a database server and an application server. The database server stores various data, which will be described later with reference to FIG. 2. The application server performs the processes, which will be described later with reference to FIG. 4, FIG. 6, FIG. 10, FIG. 12, etc. Note that the multiple servers here may be physical servers or virtual servers. Furthermore, if a virtual server is used, the virtual server may be provided in one physical server, or may be distributed and arranged across multiple physical servers. As described above, the specific configuration of the server system 100 in this embodiment can be modified in various ways.

The terminal device 200 is a device used to present information provided by the server system 100 to the user and to input information by the user of the terminal device 200. The user in this embodiment may be, for example, a caregiver who provides care to a person being assisted (patient, resident) in a care facility or the like. Alternatively, the user of the terminal device 200 may be a home helper who provides home care, or a nurse, physical therapist, occupational therapist, or speech-language-hearing therapist who provides care to patients in a hospital or the like. Assistance in this embodiment refers to assistance with daily activities that a person is less able to perform on their own. Assistance includes various types of care for a person being assisted, such as assistance with eating, excretion, and transfer/movement assistance. Assistance in this embodiment may also be expanded to caregiving or nursing.

In addition, the user in this embodiment is not limited to a user who provides assistance as a job, but may also be a family member of the person being assisted. Since the family member of the person being assisted may not have specialized knowledge regarding assistance, they are users who do not register tacit knowledge, as described below using Figures 4 and 6, for example, but instead exclusively use tacit knowledge registered by other users. However, there is no prohibition on the family member of the person being assisted registering tacit knowledge.

Considering that the terminal device 200 will be used in situations where care is being provided, it is, for example, a smartphone or tablet terminal that is easy to carry. However, the screen described below using FIG. 16 and other figures may be viewed in situations where the caregiver is not providing care, and the terminal device 200 may be a PC (Personal Computer) or the like.

The headset 300 also includes earphones or headphones for outputting sound, and a microphone that converts sound into an electrical signal and outputs it as voice data. The headset 300 is a device that performs a process of outputting speech by the user as voice data, and a process of presenting information from the server system 100 to the user as sound.

For example, a user such as a caregiver is provided with one terminal device 200 and one headset 300, and communicates with the server system 100 using the terminal device 200 and the headset 300. However, the method of this embodiment is not limited to this, and the headset 300 may be omitted from the device used by the user, or another wearable device may be added. The wearable device here may be a glasses-type device, a wristwatch-type device, or a device of another shape.

FIG. 2 is a block diagram showing a detailed configuration example of the server system 100. The server system 100 includes, for example, a processing unit 110, a storage unit 120, and a communication unit 130. The processing unit 110 accepts a registration request for know-how information 121, which is information used in caregiving and includes output information to be output to a caregiver and input information for outputting the output information. The storage unit 120 stores multiple pieces of know-how information 121 based on multiple registration requests.

The output information here refers to information that allows the user, who is a caregiver, to use tacit knowledge by being presented with the output information. The output information may be, for example, information for supporting the specific actions of the caregiver. The caregiver performs various types of assistance, such as meal assistance and excretion assistance. In addition, for each type of assistance, the caregiver performs specific actions to perform the assistance. For example, in meal assistance, the caregiver performs various assistance actions, such as having the person being assisted sit in a posture suitable for eating, picking up food with a spoon and bringing it to the person being assisted's mouth, etc. The output information may be assistance information that represents these assistance actions. For example, the output information may be information that represents whether or not an action should be performed, or may be information that identifies the correct timing, movement, posture, amount, etc. when performing an action.

Furthermore, an experienced caregiver may have a hunch that a particular person being assisted "seems like he or she is likely to fall." This is thought to be because an experienced caregiver has the ability to properly grasp the condition of the person being assisted and the living environment of the person being assisted, and to naturally estimate the risk of the person being assisted. The output information of this embodiment may be information that indicates the risk of the person being assisted. In a narrow sense, the risk here is the risk of an incident occurring, and the incident refers to an incident that can be suppressed by appropriate assistance. The output information here does not need to instruct the caregiver to take a specific action. For example, even if the output information is not information that instructs a specific action to prevent a fall, by properly grasping the risk of the person being assisted falling, the caregiver can decide on an action that takes the risk of falling into consideration.

In this way, the output information of this embodiment may be assistance information for instructing the caregiver to take specific actions, or may be information that provides the caregiver with knowledge that is considered useful in assistance. When the output information is the latter, the specific action content is left to the caregiver, but it is possible to encourage the execution of high-quality assistance compared to when there is at least no output information. Note that knowledge that is useful in assistance is not limited to risks. For example, in meal assistance, information presenting the type of food may be used as output information. The output information may be information that represents other knowledge used in meal assistance, or may be information used in assistance other than meal assistance.

In addition, the input information in this embodiment is information for outputting the output information, and specifically may be information representing elements that the expert considers in tacit knowledge. For example, when the output information is assistance information representing an assistance action performed by an assistant, the input information may be condition information representing a starting condition of the assistance action. In assistance, there may be situations in which the same assistance action is useful and situations in which it is not. For example, an expert provides high-quality assistance by adjusting the action according to the situation (condition), such as performing a first action when a first condition is met and performing a second action when a second condition is met. Therefore, by using condition information as input information for the know-how information 121, it becomes possible to have the assistant perform an appropriate action in an appropriate situation.

However, just as the output information is not limited to assistance information representing assistance actions, the input information is not limited to condition information representing start conditions. For example, the input information may be one or more pieces of information that serve as an indicator when determining output information. In the above example of risk, an expert can estimate the risk of the person being assisted by intuition or tacit knowledge, by comprehensively judging the physical function, cognitive function, specific behavior, etc. of the person being assisted. In this case, in order to use the expert's tacit knowledge as output information, information representing physical function, information representing cognitive function, information representing behavior, etc. may be used as input information. The same applies when information other than risk is used as output information, and the input information can widely include information that can be a judgment indicator when determining output information.

In the following explanation, an example will be described in which the input condition is condition information representing a given start condition. Also, an example will be described in which the output information is assistance information representing an assistance action that is executed when the start condition is satisfied. However, as explained above, the condition information in this specification can be expanded to input information representing things other than the start condition. Also, the assistance information in this specification can be expanded to output information representing things other than assistance actions.

The processing unit 110 of this embodiment is configured with the following hardware. The hardware can include at least one of a circuit for processing digital signals and a circuit for processing analog signals. For example, the hardware can be configured with one or more circuit devices or one or more circuit elements mounted on a circuit board. The one or more circuit devices are, for example, an IC (Integrated Circuit), an FPGA (field-programmable gate array), etc. The one or more circuit elements are, for example, a resistor, a capacitor, etc.

The processing unit 110 may also be realized by the following processor. The server system 100 of this embodiment includes a memory that stores information and a processor that operates based on the information stored in the memory. The information is, for example, a program and various data. The processor includes hardware. The processor can be various types of processors, such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a DSP (Digital Signal Processor). The memory may be a semiconductor memory such as a static random access memory (SRAM), a dynamic random access memory (DRAM), or a flash memory, or may be a register, a magnetic storage device such as a hard disk drive (HDD), or an optical storage device such as an optical disk drive. For example, the memory stores computer-readable instructions, and the processor executes the instructions to realize the functions of the processing unit 110 as processing. The instructions here may be instructions from an instruction set that constitutes a program, or may be instructions that instruct the hardware circuitry of the processor to operate.

The processing unit 110 includes, for example, a registration processing unit 111, a search processing unit 112, a similarity determination unit 113, a state determination unit 114, and an importance determination unit 115.

The registration processing unit 111 receives information corresponding to the user's tacit knowledge and performs processing to store the information in the storage unit 120. For example, the registration processing unit 111 performs processing to register the tacit knowledge as know-how information 121, as described later with reference to FIG. 4. The registration processing unit 111 may also perform processing to create and update registration information 122 that associates devices, processing algorithms, parameters, etc. with the know-how information 121 in order to make the tacit knowledge easier to use, as described later with reference to FIG. 6.

When there is a user who wishes to use know-how information 121 registered by the registration processing unit 111, the search processing unit 112 performs processing to accept a search request from the user and present the search results. Furthermore, when a user selects any of the know-how information 121 from the search results, the search processing unit 112 may create and update list information 123, which is information that associates the user with the know-how information 121. This allows each user to use know-how information 121 registered by others. More specifically, the list information 123 is a collection of one or more pieces of know-how information 121 that are being used by a given user.

The similarity determination unit 113 performs a first similarity determination process to determine the similarity between two pieces of know-how information 121. For example, when the search processing unit 112 acquires the above-mentioned search request, the search processing unit 112 may acquire the result of the first similarity determination process from the similarity determination unit 113 and determine the know-how information 121 to be presented as the search result based on the result.

The status determination unit 114 performs a process to acquire status information that indicates the status of the person being assisted. Details of the status information and the acquisition process will be described later.

The importance determination unit 115 performs processing to determine the importance of the know-how information 121 based on association information that associates changes in the condition information of the person being assisted with one or more pieces of know-how information 121 used to assist the person being assisted.

The memory unit 120 is a work area for the processing unit 110, and stores various information. The memory unit 120 can be realized by various types of memory, and the memory may be a semiconductor memory such as an SRAM, a DRAM, a ROM, or a flash memory, or may be a register, a magnetic storage device, or an optical storage device.

The storage unit 120 may store know-how information 121, registration information 122, and list information 123. The know-how information 121 is information in which condition information representing a start condition is associated with assistance information representing an assistance action to be performed when the start condition is satisfied. The condition information and assistance information are, for example, text.

The registered information 122 includes one or more pieces of know-how information 121 registered by a user. The registered information 122 is, for example, information in which at least a device (sensor) for automating the determination of the start condition and specific processing content of the start determination are associated with the know-how information 121. However, the registered information 122 may include information that is not associated with a device or the like. The list information 123 is a collection of one or more pieces of know-how information 121 that are being used by a given user. Details of each piece of information will be described later. The storage unit 120 may also store other information.

The communication unit 130 is an interface for communicating via a network, and includes, for example, an antenna, an RF (radio frequency) circuit, and a baseband circuit. The communication unit 130 may operate according to the control of the processing unit 110, or may include a processor for communication control different from the processing unit 110. The communication unit 130 is an interface for communicating according to, for example, TCP/IP (Transmission Control Protocol/Internet Protocol). However, the specific communication method can be modified in various ways.

Figure 3 is a block diagram showing a detailed configuration example of the terminal device 200. The terminal device 200 includes, for example, a processing unit 210, a storage unit 220, a communication unit 230, a display unit 240, and an operation unit 250.

The processing unit 210 is configured with hardware including at least one of a circuit for processing digital signals and a circuit for processing analog signals. The processing unit 210 may also be realized by a processor. The processor may be of various types, such as a CPU, a GPU, or a DSP. The processor executes instructions stored in the memory of the terminal device 200, whereby the functions of the processing unit 210 are realized as processing.

The memory unit 220 is the work area of the processing unit 210 and is realized by various types of memory such as SRAM, DRAM, and ROM.

The communication unit 230 is an interface for communicating via a network, and includes, for example, an antenna, an RF circuit, and a baseband circuit. The communication unit 230 communicates with the server system 100, for example, via a network.

The display unit 240 is an interface that displays various information and may be a liquid crystal display, an organic EL display, or another type of display. The operation unit 250 is an interface that accepts user operations. The operation unit 250 may be a button or the like provided on the terminal device 200. The display unit 240 and the operation unit 250 may also be a touch panel that is integrally configured.

The terminal device 200 may also include components not shown in FIG. 3, such as a light-emitting unit, a vibration unit, a sound input unit, and a sound output unit. The light-emitting unit is, for example, an LED (light emitting diode) and issues a notification by emitting light. The vibration unit is, for example, a motor and issues a notification by vibration. The sound input unit is, for example, a microphone, and the sound output unit is, for example, a speaker and issues a notification by sound. The terminal device 200 may also include various sensors, such as motion sensors such as an acceleration sensor or a gyro sensor, an imaging sensor, a GPS (Global Positioning System) sensor, etc.

4 to 9, a process for registering know-how information 121 in the storage unit 120 of the server system 100 will be described below. The following process is a process for a user, such as a caregiver or a nurse, to accumulate his or her own tacit knowledge in a form that can be used by others.

2.1 Registration process by text First, the user inputs text including a start condition such as "When xxx, do yyy" and an assistance action to be executed when the start condition is satisfied. In this way, it becomes possible to accumulate in the server system 100 situations and actions that the target user considers important in providing assistance.

Figure 4 is a diagram explaining the flow of the process of registering know-how information 121. When this process starts, the user first speaks into the microphone of the headset 300, saying the above-mentioned "If xxx, do yyy." In step S101, the headset 300 converts the user's speech into voice data and transmits the voice data to the terminal device 200. Note that, prior to the voice input, a recognition process of a given trigger word may be performed, or an operation unit provided on the headset 300 may be operated.

In step S102, the terminal device 200 performs a voice recognition process on the voice data transmitted from the headset 300. In the voice recognition process, first, an acoustic analysis is performed to extract features from the voice data. The results of the acoustic analysis are processed to identify phonemes with similar features using an acoustic model. Furthermore, a pronunciation dictionary and a language model are used to convert the phonemes into words and sentences, thereby obtaining a voice recognition result. The voice recognition result is data that represents the conversion result of converting voice data into text. Note that since a wide range of known methods can be applied to the voice recognition process of this embodiment, further detailed explanation is omitted.

In step S103, the terminal device 200 transmits text resulting from the voice recognition process to the server system 100. The data transmitted in step S103 is, for example, text such as "If you do xxx, do yyy." Alternatively, the terminal device 200 may obtain two pieces of text representing the start condition and the assistive action by detecting words such as "if...", "in the case of", and "when" in the voice recognition process. In the above example, "you did xxx" is the text representing the start condition, and "do yyy" is the text representing the assistive action.

In step S104, the registration processing unit 111 of the server system 100 performs a process of storing know-how information 121 in the storage unit 120 based on the acquired text. For example, the registration processing unit 111 stores information identifying the user who is the sender of the text. The information identifying the user may be identification information assigned to the headset 300 or identification information assigned to the terminal device 200. Alternatively, the information identifying the user may be a user ID that uniquely identifies the user. For example, the registration processing unit 111 stores information including the user ID, text representing the start condition, and text representing the assistance action as know-how information 121 in the storage unit 120.

Figure 5 is an example of know-how information 121 stored in step S104. The ID in Figure 5 is information that uniquely identifies the know-how information 121. If the know-how information 121 can be uniquely identified by a combination of the start condition and the assisting action, the ID may be omitted. The registered user is a user ID or the like that represents the user who registered the target know-how information 121. As described above, the start condition and the assisting action are text based on user input. Note that the condition information representing the start condition is not limited to text only, and may include voice data or the result of voice recognition processing. The result of voice recognition processing is, for example, the result of morpheme analysis, such as information that associates morphemes with parts of speech. The same is true for the assisting information representing the assisting action, which may include voice data or the result of voice recognition processing.

As shown in FIG. 5, the registration processing unit 111 may also perform a process of registering information other than the above as the know-how information 121. For example, the know-how information 121 may include information specifying the situation in which the assistance action is performed. For example, the know-how information 121 includes information indicating in which type of assistance the assistance action is performed, among various types of assistance such as meal assistance, excretion assistance, and transfer/movement assistance. The know-how information 121 may also include information indicating the attributes of the person being assisted, who is the subject of assistance provided by the user. The attributes here include information such as the age, sex, height, weight, medical history, and medication history of the person being assisted. The know-how information 121 may also include physical evaluation data indicating the physical evaluation of the person being assisted. The physical evaluation data includes information such as the evaluation value of ADL (Activities of Daily Living), rehabilitation history, risk of falling, and risk of bedsores. In this way, it becomes possible to store know-how information 121, which represents tacit knowledge, for what type of assistance and for what type of person being assisted.

Additional information such as the type of assistance, attributes of the person being assisted, and physical evaluation data may be input voluntarily by the user. For example, the user may utter "When xxx, do yyy" as well as words including "eating" and "tall". The user may also utter a series of utterances including additional information, such as "When a tall patient does xxx while eating, do yyy."

The server system 100 may also send questions such as "In what situations will you use this?" and "For what type of person will you use this?" to the headset 300. The user answers the questions by voice, and text representing the answer is sent to the server system 100, whereby the additional information is acquired.

In this manner, in this embodiment, tacit knowledge is accumulated in the storage unit 120 of the server system 100 as know-how information 121 including text. The user can simply tweet using the headset 300 the starting conditions and assistance actions that the user considers important while providing assistance, eliminating the need for complex input operations. This makes it easy to collect large amounts of tacit knowledge used in the field of nursing care and medical care.

Furthermore, when the collection of know-how information 121 progresses, the server system 100 may perform an analysis process of the know-how information 121. For example, the processing unit 110 may perform a process of mapping each piece of know-how information 121 onto a feature space by determining features based on each piece of know-how information 121. For example, the processing unit 110 may determine dense regions in the feature space, thereby making it possible to estimate information that is particularly important among tacit knowledge.

Note that while FIG. 4 shows an example in which the voice recognition process is performed in the terminal device 200, this is not limiting. For example, the terminal device 200 may transmit voice data received from the headset 300 to the server system 100. In this case, the processing unit 110 of the server system 100 may perform the voice recognition process. The voice recognition process may also be executed in an external voice recognition server.

The user's input may also be performed using text instead of voice. For example, the terminal device 200 may obtain text such as "When you do xxx, do yyy" by accepting a character input operation by the user. The processing after obtaining the text is similar to the example in FIG. 4.

2.2 Correlation with Devices By performing the process shown in Fig. 4, for example, know-how information 121 including a start condition of "if the face starts to become unsteady during a meal" and an assistance action of "stop serving the meal" is registered. These texts are meaningful information because they can call attention to meal assistance.

However, the method of this embodiment is not limited to accumulating tacit knowledge as text data, and further information may be associated. For example, the registration processing unit 111 associates information for automatically determining the start condition with the know-how information 121. In this way, the server system 100 can automatically determine whether the condition "the face begins to wobble" is met. As a result, it is possible to suppress the variation in judgment between users, and therefore to allow less skilled users to behave in the same way as more skilled users.

Specifically, in this embodiment, information identifying a device having a sensor that collects data, and information identifying specific processing content for sensor information from the sensor, etc. may be associated with the know-how information 121. This will be explained below with reference to Figures 6 to 8.

Figure 6 is a diagram explaining the process flow for collecting information for automation. First, in step S201, the registration processing unit 111 extracts the portion requiring interpretation from the condition information representing the start condition. Specifically, the portion requiring interpretation is the portion that is subject to automatic determination by the server system 100, and is, for example, text that represents the movement, state, and environment of the person being assisted.

For example, if the start condition is "If the person's face begins to unsteady while eating," the text "unsteady" represents the movement of the person being assisted that should be detected in determining the start condition. The "face" part is information that identifies the part of the body that is the subject of movement detection. Therefore, the registration processing unit 111 extracts the part "The person's face began to unsteady" from "If the person's face begins to unsteady while eating" as the part that requires interpretation.

Similarly, if the know-how information 121 associates the assisting action of "changing to an easier position to eat" with the starting condition of "if the person only ate a little bit of the rice from the spoon," then "did not eat" represents the direct movement of the person being assisted. In this case, "rice" represents the object to be eaten and is therefore used in the judgment. The "spoon" portion also specifies the location of the rice and can be used in the judgment. "Only a little" also provides a standard for the amount and can be used in the judgment. Therefore, the registration processing unit 111 extracts, for example, text "only ate a little bit of the rice from the spoon" as a portion that requires interpretation.

As described above, the registration processing unit 111 may identify the parts that require interpretation, for example by performing morpheme analysis in natural language processing. For example, the registration processing unit 111 first extracts words that express movements or states based on the results of the morpheme analysis as described above. Furthermore, the registration processing unit 111 may perform processing to sequentially extract noun phrases that serve as objects, adverbial phrases that modify movements or states, adjective phrases, and the like. For example, when morpheme analysis is performed in the speech recognition processing shown in step S102 of FIG. 4, the registration processing unit 111 may extract the parts that require interpretation based on the results of the speech recognition processing.

Alternatively, the storage unit 120 of the server system 100 may store in advance words that represent movements, states, etc. that need to be detected during assistance. The registration processing unit 111 may extract the parts that require interpretation based on a comparison process between these words and the text that represents the start conditions. In addition, the process of extracting the parts that require interpretation can be modified in various ways. For example, machine learning may be performed on a neural network or the like using training data of know-how information and information on the parts of that know-how information that require interpretation, and the parts that require interpretation may be automatically extracted using the learned model. Hereinafter, the neural network is abbreviated as NN. An example of an NN will be described later using FIG. 17, etc.

In step S202, the registration processing unit 111 identifies a device including a sensor required for processing based on the extraction result of the part that requires interpretation. For example, when determining that the start condition is "the face has begun to unsteady," the registration processing unit 111 determines that it is necessary to detect the movement of the face (head) of the person being assisted. For example, the registration processing unit 111 identifies a camera capable of capturing an image of the face of the person being assisted, a wearable device that can be worn on the head and includes a motion sensor, etc., as a device required for processing. For example, the storage unit 120 of the server system 100 may store in advance one or more devices capable of detecting the movement of each part of the person being assisted. In addition, for example, machine learning may be performed on a neural network or the like using text information of the start condition and training data of the devices required for processing, and the devices required for processing may be automatically identified using the learned model.

In addition, as a device that determines whether "only a small amount of rice was eaten from the spoon," a camera that can capture an image of the caregiver's hands or the mouth of the person being assisted is identified as a device that includes the sensor necessary for processing.

In step S203, the registration processing unit 111 determines whether the target user can use the identified device. The user here is a registered user who has registered know-how information 121 by performing, for example, the processing of FIG. 4.

For example, the storage unit 120 stores in advance, for each user, a list of devices that the user can use. The device list is information that identifies specific devices such as smartphones, headsets, eyeglass-type wearable devices, watch-type wearable devices, and devices that can acquire biometric information of a person being assisted. More specifically, the device list may not only store information such as a smartphone, but may also store the manufacturer of the smartphone, the product model number, etc.

Furthermore, devices available to a user are not limited to equipment worn or carried by the user, but may also be devices located in a care facility or the like. For example, the device list may include cameras located in the target user's work environment, or may include other sensor devices. The sensors included in the sensor devices can be modified in various ways, and various sensors such as temperature sensors, humidity sensors, illuminance sensors, air pressure sensors, activity meters, and odor sensors can be used.

In step S203, the registration processing unit 111 compares the device identified in step S202 with a list of devices available to registered users. In the above example, the registration processing unit 111 determines whether each device included in the device list can capture an image of the face of the person being assisted, or whether it includes a motion sensor and can be worn on the head.

If the registration processing unit 111 determines that the registered user cannot use the specified device, it skips the processing from step S204 onwards. In this case, the registered know-how information 121 is not associated with a device or the like. In other words, the know-how information 121 is used in the form of text, such as "do yyy when "do xxx", and no automatic determination of start conditions, etc. is performed.

The registration processing unit 111 may instruct the terminal device 200 via the communication unit 130 that the identified device is not available, and the terminal device 200 may display a message. If it is determined that the registered user can use the identified device, the registration processing unit 111 determines information necessary to automatically determine the start condition. For example, the registration processing unit 111 determines a first processing algorithm for the device to collect device data, a second processing algorithm for processing the collected device data, and parameters to be used in the second processing algorithm. The device data is specifically sensor information detected by a sensor possessed by the device. For example, when a device with a camera such as a smartphone is identified, the device data (sensor information) is image data captured by the camera.

In step S204, the registration processing unit 111 determines the first processing algorithm. In other words, the registration processing unit 111 determines the processing content when acquiring the sensor information.

For example, when automating the determination of the start condition that the "face is unsteady," the registration processing unit 111 needs to acquire, as sensor information, information that produces a discernible difference between cases where the "face is unsteady" and cases where the "face is not unsteady." That is, the sensor information in this case is information that represents the results of detecting the facial movement of the person being assisted, and may be, for example, a video image of about one second capturing an image of an area including the head of the person being assisted, or time-series data of about one second from a motion sensor attached to the head of the person being assisted.

For example, the storage unit 120 may store a table in which multiple first processing algorithms are associated with words that represent movements such as "unsteady." In the example of detecting whether the face is unsteady, the first processing algorithm is an algorithm that causes the camera (image sensor) to capture a moving image, divide it into one-second intervals, and output it. Alternatively, the first processing algorithm is an algorithm that causes the motion sensor to acquire acceleration data and angular velocity data in a time series, divide it into one-second intervals, and output it. The registration processing unit 111 performs a process of identifying a table based on the word extracted in step S201 and selecting one of the first processing algorithms included in the table. Although omitted in FIG. 6, the registration processing unit 111 may perform a process of displaying multiple first processing algorithms included in the identified table on the terminal device 200 and accepting a selection operation by the user.

By determining the first processing algorithm, it becomes possible to collect device data (sensor information). Next, the processing unit 110 (registration processing unit 111) performs a process of collecting multiple device data acquired by the device and a process of transmitting a request to add a correct answer tag for each of the multiple device data to the terminal device 200 used by the registered user. The correct answer tag here represents the result of the registered user's judgment as to whether each of the multiple device data is data that satisfies the start condition. In this way, it is possible to collect information that associates input data in the judgment process of the start condition with correct answer data that should be output when the input data is input. Based on these, the registration processing unit 111 can determine parameters to be used in the second processing algorithm. The parameters will be described later. According to the method of this embodiment, the judgment criteria of the registered user are reflected in the parameters, so it is possible to appropriately digitize the tacit knowledge of the registered user. Specific processing will be described below.

In step S205, the registration processing unit 111 instructs the terminal device 200 to collect sample device data (hereinafter also referred to as sample data) via the communication unit 130. For example, the registration processing unit 111 may transmit a program that executes processing corresponding to the first processing algorithm described above. In step S206, the terminal device 200 instructs the sensor to collect sample data. Note that the sensor here may be included in the terminal device 200, or may be included in a sensor device different from the terminal device 200. That is, in step S206, the terminal device 200 may perform processing to control an internal sensor, or may transmit information instructing an external device to collect data. The terminal device 200 or the sensor device starts collecting sample data by installing the program transmitted from the registration processing unit 111.

In step S207, the sensor collects sample data. In step S208, the sensor transmits the collected sample data to the terminal device 200. In step S209, the terminal device 200 transmits the sample data to the server system 100. In step S210, the registration processing unit 111 of the server system 100 stores the received sample data in the storage unit 120.

By the processing of steps S207 to S210, one sample data is stored in the storage unit 120 of the server system 100. In this embodiment, the processing of steps S207 to S210 is repeated until a predetermined number of sample data is accumulated. For example, after a registered user performs the processing shown in FIG. 4, the collection of sample data is gradually progressed as the user continues normal work. For example, if the start condition "face begins to stagger while eating" is registered, when the registered user assists with eating, the smartphone camera is controlled to be turned ON, and sample data that captures an image of the head of the person being assisted is automatically collected. Then, when the registered user continues work, including assisting with eating, for a certain period of time, the collection of a predetermined number of sample data is completed.

If it is determined that collection of a predetermined number of sample data has been completed, in step S211, the registration processing unit 111 performs processing to generate screen information for displaying the sample data. In step S212, the registration processing unit 111 transmits the screen information to the terminal device 200. In step S213, the display unit 240 of the terminal device 200 displays the sample data. Note that the screen information here may be the display screen itself, or may be information that can identify the display screen.

Figure 7 is an example of a screen displayed on the display unit 240 in step S213. For example, if the sample data is a video of about 1 second capturing an image of the head of the person being assisted, the display unit 240 displays a screen including thumbnails of each video. The display unit 240 may also display a screen that prompts the user to input whether or not each sample data satisfies the start condition. In the example of Figure 7, the display unit 240 displays the text "Please select the data for 'face starting to stagger'."

In step S214, the terminal device 200 acquires correct answer data indicating whether or not each sample data satisfies the start condition. For example, the user performs an operation to select the data "face begins to wobble" using the operation unit 250 based on the screen of FIG. 7. For example, the terminal device 200 acquires a correct answer tag indicating that the data is correct as the correct answer data corresponding to the selected sample data. The terminal device 200 also acquires an incorrect answer tag indicating that the data is incorrect as the correct answer data corresponding to sample data that has not been selected when the user operation is completed.

In step S214, the terminal device 200 may also acquire information regarding the viewpoint of the user's judgment. For example, when judging whether the "face is unsteady," the maximum amount of movement from a reference position may be used as the judgment criterion. The reference position here may be, for example, the position of the face when sitting upright in a chair or bed, or the center of the captured image. Alternatively, it is also possible to use the amount of head movement in one go as the judgment criterion, regardless of the reference position. In other words, even if the same word "unsteady" is extracted, the judgment on it may differ depending on the user.

The storage unit 120 may store a table that associates information representing multiple perspectives with a word that describes a movement, such as "unsteady." For example, the table stores two perspectives: "the maximum angle of head movement relative to a reference position in the image is greater than a threshold value α" and "the angle of head movement per one movement is greater than a threshold value β." For example, in step S212, the registration processing unit 111 may transmit a display screen that prompts the user to select one of the multiple perspectives included in the table. Then, in step S214, the terminal device 200 accepts information related to the judgment perspective along with the acceptance of the correct answer data.

In step S215, the terminal device 200 transmits the correct answer data to the server system 100. Also, when a judgment viewpoint is input as described above, the terminal device 200 transmits information related to the viewpoint to the server system 100.

In step S216, the registration processing unit 111 first performs a process of determining a second processing algorithm for inputting device data and outputting output data indicating whether or not the start condition is satisfied. For example, the registration processing unit 111 determines the second processing algorithm based on the user input regarding the viewpoint of judgment acquired in step S215. For example, when the viewpoint of "the maximum head movement angle relative to the reference position of the image is greater than the threshold value α" is selected, the second processing algorithm is an algorithm including a process of obtaining the "maximum head movement angle relative to the reference position of the image" and a process of comparing the obtained movement angle with the threshold value α. When the viewpoint of "the head movement angle per time is greater than the threshold value β" is selected, the second processing algorithm is an algorithm including a process of obtaining the "head movement angle per time" and a process of comparing the obtained movement angle with the threshold value β. In addition, it is assumed that the specific processing content differs between the case where a moving image is the target and the case where time-series acceleration data or the like is the target. Therefore, the second processing algorithm may be determined according to the type of device (type of sensor) and the content of the first processing algorithm.

However, the parameters α and β in the second processing algorithm are unknown. Therefore, in step S216, the registration processing unit 111 performs processing to calculate the parameters based on the sample data and the correct answer data. For example, the registration processing unit 111 obtains the "maximum movement angle of the head relative to the reference position of the image" for the sample data according to the second processing algorithm. The registration processing unit 111 then performs processing to obtain the most likely α such that the movement angle is greater than α for sample data tagged as correct, and is less than or equal to α for sample data tagged as incorrect.

For example, the registration processing unit 111 may classify sample data tagged with a correct answer tag and sample data tagged with an incorrect answer tag using an SVM (support vector machine). For example, the registration processing unit 111 obtains a hyperplane that separates sample data tagged with a correct answer tag from sample data tagged with an incorrect answer tag, and determines parameters such as α and β based on the hyperplane.

Note that the second processing algorithm is not limited to the above example, and a neural network may be used. For example, the storage unit 120 may store multiple neural networks with different structures as multiple second processing algorithms. For example, the storage unit 120 stores NN1, which is an NN suitable for processing with image data as input, and NN2, which is an NN suitable for processing with acceleration data or angular velocity data from a motion sensor as input. In step S216, the registration processing unit 111 performs a process of selecting one of multiple neural networks including NN1 and NN2 automatically or based on user input. Note that NN1 is, for example, a convolutional neural network (CNN). NN2 is, for example, a deep neural network (DNN).

Then, in step S216, the registration processing unit 111 may perform a learning process using a NN. For example, the registration processing unit 111 inputs sample data into the NN and performs forward calculations using the weights at that time to obtain output data. The registration processing unit 111 also obtains an objective function (e.g., an error function such as a mean squared error function) based on the output data and the correct answer data, and updates the weights to reduce the error using an error backpropagation method or the like. The registration processing unit 111 may store the NN including the weights at the end of learning in the storage unit 120 as a learned model. That is, when a NN is used, the structure of the NN corresponds to the second processing algorithm, and the weights correspond to the parameters.

In step S217, the registration processing unit 111 stores in the storage unit 120 the device used to acquire the sample data, the processing content for the device data (sensor information) of the device, and the identified parameters in association with the know-how information 121.

8 is an example of the registration information 122 stored in step S217. As shown in FIG. 8, the registration information 122 includes a user ID representing the user who associated the device, an ID representing the know-how information 121, the device, and a processing program. The user here is, for example, a registered user who registered the know-how information 121 using FIG. 4. The device included in the registration information 122 is, for example, information such as the manufacturer and model number of the device, as described above. The processing program is, for example, a set of a second processing algorithm and parameters, and may be a NN including weights. In addition, an example in which the processing program included in the registration information 122 is a second processing algorithm and parameters will be described below, but the processing program here may include the first processing algorithm described above. In this way, it is also possible to manage the first processing algorithm for acquiring device data to be processed by the second processing algorithm using the registration information 122.

By using the know-how information 121 in FIG. 5 and the registration information 122 in FIG. 8, it becomes possible to automatically determine whether the start condition is satisfied. For example, in the case of the know-how information 121 corresponding to ID1 in FIG. 5, by performing processing according to PG1 on the device data of Device1, it is possible to automatically determine whether the start condition if1 is satisfied.

As described above, the processing unit 110 (registration processing unit 111) performs a text analysis process on the condition information to identify the device used to determine the start condition represented by the condition information (e.g., steps S202 and S203), and may associate information representing the identified device with the know-how information 121 (e.g., step S217). In this way, the condition information is associated with a specific device, making it possible to determine whether the start condition is satisfied using that device.

The know-how information 121 in this embodiment can be divided into three types from the viewpoint of association with devices. The first type is know-how information 121 for which the device is determined to be usable in step S203 and the processing of step S217 has been completed. This is know-how information 121 for which the start conditions can be automatically determined because the second processing algorithm and parameters are specified in addition to the device association.

Secondly, there is know-how information 121 for which the device was determined to be unusable in step S203 and the processing from step S204 onwards was not performed. Since the know-how information 121 is used in the form of text, for example, the user himself determines whether the start condition is satisfied.

Thirdly, there is know-how information 121 for which it has been determined in step S203 that the device is usable, but the processing in step S217 has not been completed. This is know-how information 121 for which sufficient sample data has not been collected to determine parameters. For example, the registration processing unit 111 may not generate registration information 122 for this know-how information 121, and may treat it in the same way as know-how information 121 for which it has been determined in step S203 that the device is unusable. When sufficient sample data is accumulated in the future, the processing in step S217 will be completed and registration information 122 will be generated, making it possible to automatically determine the start condition. There may also be cases in which the collection of sample data is not completed even after time has passed, and registration information 122 will not be generated.

In FIG. 6, an example is shown in which the parameters found in step S216 are stored as is in step S217. However, the processing of this embodiment is not limited to this. For example, the registration processing unit 111 may use a part of the combination of sample data and correct data as validation data, and use the validation data to find the accuracy rate of the determination processing using the second processing algorithm and parameters. The registration processing unit 111 transmits the accuracy rate to the terminal device 200. The terminal device 200 presents the accuracy rate and accepts user input on whether or not to adopt the parameters. Then, when the user inputs that the parameters will be adopted, the registration processing unit 111 may perform the processing of step S217. Furthermore, when the user inputs that the parameters will not be adopted, the registration processing unit 111 may, for example, reset the parameters and resume collection of sample data.

2.3 Determining the correct action The above describes a method for automating the determination of the start condition among the start conditions and the assisting action included in the know-how information 121. However, the method of the present embodiment is not limited to this, and the process related to the assisting action may be automated. For example, if there is know-how information 121 that "changes the assisted person to a posture that makes it easier to eat" when "the person eats only a little rice from the spoon", a process for determining the correct "posture that makes it easier to eat" may be performed, or a process for issuing a warning may be performed when the posture the user has taken deviates from the correct posture. By determining the correct answer for the assisting action, it becomes possible to have the user perform the assisting action according to the registered know-how information 121 regardless of the user's level of proficiency.

The specific process flow is the same as that shown in FIG. 6. That is, the registration processing unit 111 extracts the portion requiring interpretation from the text representing the assistance action, similar to step S201. In the above example, the registration processing unit 111 extracts "posture that makes it easy to eat."

Next, similar to steps S202 and S203, the registration processing unit 111 identifies devices that include a camera that captures an image of the user and a motion sensor that detects posture as devices for detecting postures that are easy to eat, and determines whether the devices are usable.

If it is available, similar to steps S207 to S215, the registration processing unit 111 collects sample data showing the posture of the person being assisted and instructs the user to tag the collected data. For example, the sample data is a still image capturing the entire body of the user while eating, and the registration processing unit 111 performs a process to accept a selection operation of a still image that shows an easy-to-eat posture from among multiple still images.

The registration processing unit 111 determines parameters based on the assigned tag, as in step S216. The second processing algorithm representing the processing content for the device data may be a comparison process between the bending angle of a joint or the like and a threshold value, or may be other processing. Also, a neural network that receives the still image itself as input may be used. The parameters may be the threshold value or the weight of the neural network.

As in step S217, the registration processing unit 111 associates the registration information 122, which includes the device, the second processing algorithm, and the parameters used when determining the assistance action, with the know-how information 121.

Figure 9 is another example of the registration information 122. As shown in Figure 9, the registration information 122 includes a user ID representing a user who has associated the device, an ID representing the know-how information 121, a device In which is a device used for start determination, a processing program In which is a processing program used for start determination, a device Out which is a device used for determining the assistance action, and a processing program Out which is a processing program used for determining the assistance action. Like the device In, the device Out is information such as the manufacturer and model number of the device. Like the processing program In, the processing program Out is, for example, a set of a second processing algorithm and parameters, and may be a NN including weights.

For example, in the case of know-how information 121 that "stop serving food" if "the patient's face begins to become unsteady during a meal," it is easy to perform the assistance action of "stop serving food," and there is little need to find a correct action in the server system 100. Therefore, in this case, the above-described processing for assistance actions may be omitted. For example, as in the know-how information 121 of ID1 in FIG. 9, the device Out and processing program Out may have no data.

In addition, when determining whether or not an assistive action is required, the device may be determined to be usable, but the second processing algorithm and parameters may not be determined due to factors such as insufficient sample data being collected. In this case, too, there will be no data in Device Out and Processing Program Out.

3. Use of Data Through the above process, the user's tacit knowledge is accumulated as know-how information 121. Furthermore, the know-how information 121 for which the conditions are satisfied is associated with registration information 122 that identifies devices and the like for automating the determination of the start conditions and the determination of the assistance actions. A method for using the acquired know-how information 121 will be described below.

3.1 Search process If a user with low skill level can use the tacit knowledge of an expert, appropriate assistance can be provided regardless of the user's skill level. For example, each of multiple users who use the information processing system 10 selects one of the know-how information 121 stored in the storage unit 120 of the server system 100 and uses the selected know-how information 121.

Figure 10 is a diagram explaining the flow of the process in which each user selects and uses know-how information 121. First, in step S301, the user uses the headset 300 to perform a process of inputting words for a search. For example, the user speaks the start condition or the assistance action into the microphone of the headset 300.

In step S302, the terminal device 200 performs a voice recognition process to obtain text representing the start condition or text representing the assistance action. In step S303, the terminal device 200 transmits the obtained text to the server system 100 as a search key.

In step S304, the server system 100 executes a search process using the acquired search key. That is, the processing unit 110 (search processing unit 112) of the server system 100 outputs, as a search result, any one of the multiple pieces of know-how information 121 based on a search request including search information (search key) that is either text corresponding to the start condition or text corresponding to the assistance action. For example, the search processing unit 112 may output, as a search result, know-how information 121 that satisfies a condition such as the degree of match with the search key. Also, as described later with reference to Figures 13A and 13B, the search processing unit 112 may output, as a search result, know-how information 121 whose result of the first similarity determination process satisfies a given condition.

A situation in which a start condition is used as a search key is, for example, a situation in which the user is unable to determine an appropriate assistance action. For example, the user may recognize a situation, such as the person being assisted making a certain movement, or the environment in which the person being assisted lives having changed in this way, but may not know what assistance action should be performed in that situation. In this case, by performing a search process using that situation as a start condition, know-how information 121 is provided that indicates the appropriate response in that situation.

Assistance actions refer to specific actions taken by a caregiver, such as feeding the person with a spoon, talking to the person, changing posture, etc. For example, a user may be aware of the actions required to assist the person being assisted with eating and excretion, but may lack the ability to judge in what situations and at what times to perform these actions. In this case, by performing a search process based on assistance actions, know-how information 121 is provided that indicates the start conditions for performing the assistance action.

In this way, according to the method of this embodiment, by performing a search process using the start condition or the assistive action as a search key, it is possible to determine and present know-how information 121 suitable for the user from among multiple pieces of know-how information 121 that represent tacit knowledge.

The specific processing of step S304 can be modified in various ways. For example, when text representing a start condition is input as a search key, the search processing unit 112 may determine that the know-how information 121 satisfies the condition when at least a part of the text representing the start condition included in the know-how information 121 matches the search key. Furthermore, when text representing an assistive action is input as a search key, the search processing unit 112 may determine that the know-how information 121 satisfies the condition when at least a part of the text representing the assistive action included in the know-how information 121 matches the search key. Alternatively, the search processing unit 112 may determine the similarity between the texts, and determine that the condition is satisfied when the similarity is equal to or greater than a threshold value.

In step S305, the server system 100 transmits to the terminal device 200 one or more pieces of know-how information 121 that have been determined to satisfy the conditions. In step S306, the display unit 240 of the terminal device 200 displays the acquired one or more pieces of know-how information 121. In step S307, the terminal device 200 accepts a selection operation by the user using, for example, the operation unit 250. That is, the user selects the know-how information 121 that he or she wishes to use from the know-how information 121 presented as the search results.

When the know-how information 121 is associated with the registration information 122 shown in FIG. 5, a specific device needs to be selected in order to fully utilize the know-how information 121. For example, assume that the registered user who registered the target know-how information 121 uses a smartphone of manufacturer AAA and model number BBB to determine the start condition, and registration information 122 indicating this is stored. However, the user who uses the know-how information 121 does not necessarily own a smartphone of the same manufacturer AAA and model number BBB. In addition, if the camera of the smartphone is only used, a product of a different model number of the same manufacturer may be used, or a product of another manufacturer may be used. Furthermore, for example, if the smartphone of the registered user is used for the purpose of capturing an image of the face of a person being assisted, a camera installed on a nursing bed, a camera installed in a room, a camera mounted on a movable device, etc. may be used as a device used for automatic determination. Therefore, when a user other than the registered user uses the know-how information 121 registered by the registered user, a device may be selected for each user.

In step S308, the terminal device 200 accepts a device selection operation by the user. For example, the memory unit 120 of the server system 100 may hold a device list representing devices owned by the user who performed the search process, and may select and present devices similar to the devices included in the registration information 122 from the list. For example, as described above, if the device of the registered user is a smartphone, the search processing unit 112 may perform processing to display a list of smartphones and similar devices owned by the user who performed the search process on the display unit 240 of the terminal device 200. Note that if the registration information 122 is not associated with the know-how information 121, the processing of step S308 is omitted.

Next, in step S309, the terminal device 200 transmits the know-how information 121 selected by the user to the server system 100. In step S310, the server system 100 updates the list information 123 representing the know-how information 121 being used by the target user. When the processing of step S308 is performed, information on the selected device is also transmitted and added to the list information 123.

As shown in FIG. 10, the processing unit 110 can obtain requests from multiple users to use any of the multiple know-how information 121 stored in the storage unit 120. The storage unit 120 may store list information 123 including one or more pieces of know-how information 121 in use for each of the multiple users in association with each other. In this way, it becomes possible to appropriately manage the know-how information 121 that each user is using among the many pieces of know-how information 121 stored in the storage unit 120.

For example, a user with a low level of skill may increase the number of situations in which the tacit knowledge of experts can be utilized by proactively using know-how information 121. Also, when there is an excess of information and it is difficult to grasp it all, adjustments can be made such as limiting the know-how information 121 used to important information. Also, since a user with a certain level of experience can provide assistance appropriately in many situations without using know-how information 121, the amount of know-how information 121 used may be reduced compared to a beginner.

FIG. 11 is an example of list information 123. List information 123 includes information for identifying a user, information for identifying know-how information 121 being used by the user, and information for identifying a device for using the know-how information 121. In the example of FIG. 11, user a corresponding to UserIDa is using know-how information 121 of ID1 registered by a registered user corresponding to UserID1. As shown in FIG. 8, the registered user registered Device1 to automate know-how information 121 of ID1. In response to this, user a selects Device1a as a device for automating know-how information 121 of ID1, as shown in FIG. 11. That is, even if the know-how information 121 is the same, the device used may differ depending on the user, and the storage unit 120 can store multiple devices associated with one know-how information 121.

In the example of FIG. 11, user a is using know-how information 121 of ID2 registered by a registered user corresponding to UserID2. Since no registered user or device has been registered for know-how information 121 of ID2, no device is associated with it when user a uses it. Note that know-how information 121 that is not associated with a device, such as ID2 in FIG. 11, is used in the form of text, for example. For example, if know-how information 121 includes information such as "meals" as additional information, text corresponding to know-how information 121 may be notified to the user at the timing of starting meal assistance.

On the other hand, in know-how information 121 to which a device is associated, such as ID1 in Figure 11, it is possible to automate the determination of start conditions, etc.

Figure 12 is a diagram explaining the process flow for using know-how information 121 associated with a device. First, when know-how information 121 is added to list information 123, in step S401, the corresponding device starts collecting sensor information. Note that the sensor information may be collected continuously, or may be used in a specific situation related to the know-how information 121. For example, if the know-how information 121 includes information such as "excretion" as additional information, the device may start collecting sensor information when excretion assistance begins.

In step S402, the sensor transmits sensor information to the terminal device 200. Note that, if the device in this case is the terminal device 200, the processing in step S402 corresponds to the transfer of data from the sensor to the processor in the terminal device 200. In step S403, the terminal device 200 transmits the sensor information to the server system 100.

In step S404, the processing unit 110 automatically determines the start condition based on the sensor information. For example, the processing unit 110 determines the second processing algorithm and parameters based on the registration information 122 in FIG. 8. The processing unit 110 uses the sensor information as input data and performs processing according to the second processing algorithm and parameters to obtain output data indicating whether the start condition is satisfied.

If it is determined that the start condition is satisfied, in step S405, the processing unit 110 identifies an assisting action based on the know-how information 121. In step S406, the processing unit 110 transmits information representing the identified assisting action to the terminal device 200. In step S407, the terminal device 200 transmits information representing the assisting action to the headset 300. In step S408, the headset 300 announces the assisting action using a speaker. Note that the information representing the assisting action here is text, and the processing in step S408 may be a voice read-aloud process. However, as described above with reference to FIG. 9, the processing unit 110 may also find a correct answer to the assisting action and make an announcement based on the correct answer.

As described above, according to the method of this embodiment, the tacit knowledge of an experienced user is digitized, and it is possible to allow a less skilled user to provide appropriate care. For example, since a less skilled user can provide care equivalent to that of an experienced user, the reproducibility of care is improved. In addition, the variation in care skills is suppressed, and organizational management is also facilitated, so that the occurrence of incidents such as falls of the person being assisted is suppressed. As a result, for example, in a nursing home, the occurrence of vacant beds due to hospitalization and the occurrence of overtime work due to the preparation of accident reports can be suppressed. In addition, if incidents are suppressed, it is also possible to suppress users from becoming too sensitive to risks, which makes it possible to reduce stress and, as a result, to suppress the turnover rate. In addition, by enabling the improvement of user skills and the improvement of the working environment, it is possible to improve the satisfaction of the person being assisted and their family, and to improve QOL (Quality of Life).

The information processing system 10, server system 100, terminal device 200, etc. of this embodiment may realize part or most of their processing by a program. In this case, a processor such as a CPU executes the program to realize the information processing system 10, etc. of this embodiment. Specifically, a program stored in a non-transitory information storage medium is read, and the read program is executed by a processor such as a CPU. Here, the information storage medium (a medium readable by a computer) stores programs, data, etc., and its functions can be realized by an optical disk, HDD, or memory (card-type memory, ROM, etc.). The processor such as a CPU then performs various processes of this embodiment based on the program stored in the information storage medium. That is, the information storage medium stores a program for causing the computer to function as each part of this embodiment.

The technique of this embodiment can also be applied to an information processing method including processing executed in the information processing device 10. The information processing method of this embodiment accepts a registration request for know-how information 121 including information in which condition information representing a given start condition is associated with assistance information representing an assistance action to be executed when the start condition is satisfied, and outputs, as a search result, any one of the multiple pieces of know-how information 121 stored based on multiple registration requests, based on a search request including search information that specifies either the start condition or the assistance action.

3.2 Similarity Determination Between Know-how Information Furthermore, the server system 100 (similarity determination unit 113) of this embodiment may perform a process of determining the similarity between a given know-how information 121 and another know-how information 121. For example, the search processing unit 112 may determine that not only the know-how information 121 extracted using the search key in step S304 of Fig. 10, but also similar know-how information that is know-how information 121 similar to the know-how information 121 satisfies the condition.

For example, the similarity determination unit 113 may determine the similarity between two pieces of know-how information 121 based on additional information included in the know-how information 121. For example, the additional information includes words that indicate the type of assistance and the attributes of the person being assisted, as shown in FIG. 5. The similarity determination unit 113 determines that the similarity is high when the same word is included in the two pieces of know-how information 121. Alternatively, synonyms are defined for each word, and the similarity determination unit 113 may determine that the similarity between the two pieces of know-how information 121 is high when a synonym of a word included in one piece of know-how information 121 is included in the other piece of know-how information 121.

Alternatively, the similarity determination unit 113 may determine the similarity based on text mining. For example, the similarity determination unit 113 performs text mining on at least one of the text representing the start condition and the text representing the assistance action. The similarity determination unit 113 obtains tf-idf representing the importance of each word extracted by text mining. tf is the frequency of occurrence of a word, and idf represents the inverse document frequency. tf-idf is an index in which the importance of a word that appears frequently is high, and the importance of a word that appears in many documents is low. For example, the similarity determination unit 113 obtains a vector in which tf-idf is associated as a value with each word that appears in the know-how information 121. The similarity determination unit 113 obtains a vector for each of the two pieces of know-how information 121, and obtains the similarity between the two pieces of know-how information 121 based on the angle θ between the obtained two vectors. For example, the similarity is cos θ. However, various methods for determining the similarity between two documents are known, and these can be widely applied in this embodiment. Furthermore, the part that is the subject of text mining is not limited to at least one of the start condition and the assistance action, and may include additional information.

The similarity determination unit 113 may also determine similar know-how information of a given know-how information 121 from the perspective of whether or not the know-how information is frequently used together with the given know-how information 121.

For example, the multiple know-how information 121 include first know-how information, second know-how information, and third know-how information, and the processing unit 110 (similarity determination unit 113) performs a similarity determination process to determine the similarity between any two of the multiple know-how information 121 stored in the storage unit 120. At this time, the similarity determination unit 113 determines the similarity based on the number of users whose list information 123 includes the first know-how information and the second know-how information, and the number of users whose list information 123 includes the first know-how information and the third know-how information.

Figures 13A and 13B are diagrams explaining the similarity determination process. Figure 13A is an example of list information 123 related to users who are using both the first know-how information corresponding to IDa and the second know-how information corresponding to IDb. In the example of Figure 13A, 100 users corresponding to UserIDx1 to UserIDx100 are using both the first know-how information and the second know-how information.

Figure 13B is an example of list information 123 related to users who are using both the first know-how information corresponding to IDa and the third know-how information corresponding to IDc. In the example of Figure 13B, only one user corresponding to UserIDy1 is using both the first know-how information and the third know-how information.

In this case, the second know-how information is likely to be used together with the first know-how information, and the third know-how information is unlikely to be used together with the first know-how information. The similarity determination unit 113 determines that the similarity between the first know-how information and the second know-how information is higher than the similarity between the first know-how information and the third know-how information.

In this way, it becomes possible to present know-how information 121 that is useful to use together to a user as similar know-how information. For example, it is possible to present a combination of useful know-how information 121 to a user who has performed the search process described above using FIG. 10. Note that the first know-how information and the second know-how information here may be different types of assistance. For example, if the first know-how information is related to meal assistance, the second know-how information is related to excretion assistance or transfer assistance. In this way, it becomes possible to include know-how information 121 that is determined to have a low similarity from the perspective of additional information, etc., in the similar know-how information.

3.3 Determination of the Importance of Know-how Information As described above, in this embodiment, a user can easily register the know-how information 121. In addition, a search process allows users other than the registered user to use the know-how information 121. However, if it is easy to register the know-how information 121, there is a high probability that a large amount of know-how information 121 will be accumulated in the storage unit 120. In this case, the importance of each piece of know-how information 121 may be determined in order to efficiently use the large amount of know-how information 121. Below, a first importance from the perspective of the person being assisted who receives assistance using the know-how information 121 and a second importance from the perspective of the caregiver who performs the assistance will be described. In the following description, the first importance will also be simply referred to as importance.

3.3.1 Importance considering the person being assisted <Status information>
The processing unit 110 of this embodiment may determine the importance of the know-how information 121 based on association information that associates a change in status information indicating the status of the assisted person with the know-how information 121 used to assist the assisted person. In this way, the importance can be determined from the viewpoint of how the assistance provided using the know-how information 121 affected the status of the assisted person. In other words, the importance of the know-how information 121 is determined according to the degree of contribution to improving or maintaining (suppressing deterioration of) the status of the assisted person.

The condition information here may be the level of care required, which indicates the degree to which the person being assisted requires assistance. Since various methods are known for calculating the level of care required as numerical data, it becomes possible to use an index that is easy to evaluate as the condition information.

The level of care required is an index used to determine whether or not a person requires care or assistance, and the level of that level. A state requiring care is one in which, for example, constant care is required. A state requiring assistance is one in which assistance is required for daily life, such as housework and dressing, and in which preventive care services are particularly effective.

For example, the level of care required may be information that represents the results of an assessment on a seven-level scale, from level 1-2 requiring assistance to level 1-5 requiring care, determined by the method described in https://www.mhlw.go.jp/stf/seisakuunitsuite/bunya/hukushi_kaigo/kaigo_koureisha/nintei/gaiyo2.html. For example, the level of care required is determined based on the results of a survey conducted by a certified surveyor on various survey items, such as physical function/activity at home, daily living functions, cognitive function, mental/behavioral disorders, and adaptation to social life. Hereinafter, survey items will also be referred to simply as items.

14A and 14B are examples of items used to determine the level of care required. Each item is determined using one of the evaluation axes: ability, assistance, or presence or absence. Ability refers to the degree to which the person has the ability to perform the target item. For example, for "turning over in bed" shown in 1-3 of the physical functions and daily living activities, the evaluation result is one of the stages such as "can do without holding on," "can do if holding on to something," and "cannot do." Assistance refers to the degree of assistance when performing the target item is evaluated. For example, for "washing" shown in 1-10 of the physical functions and daily living activities, the evaluation result is one of the stages such as "independent," "partial assistance," and "full assistance." Presence or absence refers to the evaluation of the presence or absence of an event corresponding to the target item. In addition, each item is classified into one of five categories, ADL/daily living activities, cognitive function, behavior, social life, and medical care, based on how it affects the life of the target patient or elderly person.

These items are disclosed, for example, at https://www.mhlw.go.jp/file/06-Seisakujouhou-12300000-Roukenkyoku/0000077237.pdf, and may be used to determine the level of care required in this embodiment. Items for determining the level of care required may also include items not shown in Figures 14A and 14B, such as whether end-of-life care should be started, whether a lift/wheelchair/walker should be used, whether toilet guidance should be provided, etc.

The condition information of the person being assisted may also be an evaluation result of the person being assisted's ADL (Activities of Daily Living). The evaluation of ADL may be obtained, for example, using FIM (Functional Independence Measure). In FIM, 13 items related to motor ADL and 5 items related to cognitive ADL are each evaluated on a 7-point scale from 1 to 7 points, and ADL is evaluated with a numerical value from 18 to 126 points. Note that the method of evaluating ADL is not limited to FIM, and various methods can be applied.

However, the status information in this embodiment is not limited to the above example, and can be expanded to an index that indicates the status of the person being assisted with regard to assistance. Specifically, the status of assistance is information that specifies whether or not assistance is necessary, the type and degree of assistance required, etc.

Below, an example will be described in which the status information is the level of care required. In the following, the status information is assumed to be a number between 1 and 7. The number 1 represents level 1 of assistance required, 2 represents level 2 of assistance required, 3 represents level 1 of care required, 4 represents level 2 of care required, 5 represents level 3 of care required, 6 represents level 4 of care required, and 7 represents level 5 of care required. In other words, in the following example, the smaller the number representing the status information, the better the condition of the person being assisted.

<Flow of Importance Determination Process>
FIG. 15 is a diagram for explaining the method in this embodiment. The horizontal axis of FIG. 15 represents time, and t1 and t2 represent given timings. Also, state represents the value of the status information at each timing. The status determination unit 114 of this embodiment sets a period of a given length and acquires status information of the person being assisted for each period. The given length here is, for example, one month, but may be a period of a different length. In the example of FIG. 15, the status determination unit 114 performs a process of acquiring status information of the person being assisted at each of the timings t1 and t2.

For example, the status information may be determined by a specialist such as a doctor based on the results of an investigation by a certified investigator. The server system 100 accepts input of the status information determined by a doctor or the like. The status determination unit 114 acquires the transmitted status information. Alternatively, when the results of an investigation by an investigator are acquired, the status determination unit 114 may calculate the status information. For example, the above-mentioned https://www.mhlw.go.jp/stf/seisakuunitsuite/bunya/hukushi_kaigo/kaigo_koureisha/nintei/gaiyo2.html describes a method of estimating the reference time for nursing care certification, etc. using a tree model, and a correspondence between the reference time for nursing care certification, etc. and the level of nursing care required. The status determination unit 114 may calculate the status information from the investigation results by performing a similar process. Alternatively, the state determination unit 114 may perform a process to estimate the evaluation results for each item, as described later with reference to FIG. 18, etc. In this case, it is also possible to omit the investigation by an investigator.

In addition, in this embodiment, the processing unit 110 acquires know-how information 121 (tacit knowledge) used to assist each person being assisted. For example, the user who is the caregiver may be a family member of the person being assisted. It is highly likely that family members do not provide assistance as a job, but rather provide assistance only to the person being assisted. Therefore, it is possible to consider that the know-how information 121 included in the list information 123 of the family user has been used to assist the person being assisted.

The user here may be a caregiver or nurse working in a care facility or hospital. However, if the user is in charge of two care recipients, A and B, it is not necessarily the case that all of the know-how information 121 included in the user's list information 123 is used in common for both care recipients A and B. For example, the user may use part of the know-how information 121 included in the list information 123 to assist care recipient A, and the other part to assist care recipient B. For example, in addition to the list information 123 shown in FIG. 11, the storage unit 120 may store information that can identify the care recipient to which each piece of know-how information 121 applies. In this case, even if the user is in charge of multiple care recipients, the know-how information 121 to be used for each care recipient can be identified.

Figure 15 shows an example of three pieces of association information acquired for three individuals receiving assistance, for example, during the period from t1 to t2. For example, an individual receiving assistance whose status information value was 4 at t1 received assistance using four pieces of know-how information 121 with IDs of 1, 2, 34, and 56, respectively, during the period from t1 to t2, and the status information value at t2 became 5. Note that, in the following, to simplify the explanation, know-how information 121 with ID i will also be written as ai, where i is an integer equal to or greater than 1.

In the example of FIG. 15, similarly, for example, an assisted person whose status information value was 2 at t1 received assistance using three pieces of know-how information 121 with IDs of 1, 2, and 31, respectively, during the period from t1 to t2, and the value of the status information at t2 became 1. Similarly, an assisted person whose status information value was 2 at t1 received assistance using know-how information 121 with ID of 1 during the period from t1 to t2, and the value of the status information at t2 remained unchanged at 2.

In this way, one piece of association information is obtained by targeting one person receiving care and identifying the change in status information after one unit of time has passed and the know-how information 121 used during that one unit of time. Although three pieces of association information are illustrated in FIG. 15, more pieces of association information can be obtained by increasing the number of people receiving care and the length of the target period.

The importance determination unit 115 determines the importance of the know-how information 121 based on the association information. For example, if the condition of the person being assisted using the know-how information 121 of a1 tends to improve, the importance determination unit 115 determines the importance of a1 to be high. In the above example, the higher the value of the condition information, the more assistance (care) is needed, so that an improvement in the condition corresponds to a decrease in the numerical value. However, as can be seen from the example of FIG. 15, when assistance is provided using the know-how information 121 of a1, the condition may improve, may worsen, or may be maintained. In addition, when multiple pieces of know-how information 121 are combined, it is necessary to identify which piece of know-how information 121 contributes to the change or maintenance of the condition information. In addition, it is also possible that using a given piece of know-how information 121 in combination with other know-how information 121 contributes to the improvement or maintenance of the condition information rather than using it alone. In other words, it is difficult to know the extent to which the know-how information 121 influences the change in the condition information just by focusing on the individual know-how information 121.

The processing unit 110 (importance determination unit 115) may therefore perform processing to identify important know-how information based on regression analysis in which a value related to the state information is used as a target variable and information indicating the use/non-use of know-how information 121 is used as an explanatory variable. Specifically, the regression analysis here may be a multiple regression analysis in which information indicating the use/non-use of each of the multiple know-how information 121 is used as an explanatory variable. Important know-how information here refers to know-how information 121 whose importance is determined to be equal to or greater than a predetermined level.

The value relating to the status information is, for example, a value based on the difference between the value Initial of the status information at the start point of the target period and the value Output of the status information at the end point. As described above, an example is considered in which the better the status, the smaller the value, so the reward shown in the following formula (1) may be used as the value relating to the status information. The reward is numerical data whose value increases as the degree of improvement in the status increases, and is a positive value when the status improves, 0 when the status does not change, and a negative value when the status worsens.
reward = Initial - Output...(1)

Also, let Ai be a variable indicating the use/non-use of know-how information 121 corresponding to ai. For example, Ai is a variable that is 1 when ai is used to assist the target person being assisted, and is 0 when it is not used. In this case, multiple regression analysis is performed based on the regression equation of the following equation (2). In the following equation (2), α1 to αn are partial regression coefficients, and β is a constant term. n is a positive integer indicating the number of know-how information 121. In a narrow sense, n indicates the number of know-how information 121 that appears at least once in the multiple association information to be processed, among the know-how information 121 stored in the storage unit 120. In other words, the importance determination unit 115 may exclude know-how information 121 that does not appear even once in the association information from the target of importance determination.
reward = β + α1×A1 + α2×A2 + … + αn×An … (2)

For example, the importance determination unit 115 performs processing to find the most likely α1 to αn and β using the least squares method based on multiple pieces of association information. Since multiple regression analysis is a well-known technique, a detailed description is omitted.

The importance determination unit 115 also tests the significance of each partial regression coefficient. The significance test is, for example, a t-test. The importance determination unit 115 obtains a t-value by performing a t-test on each partial regression coefficient, and further obtains a p-value from the t-value. When the p-value is smaller than a given threshold, the importance determination unit 115 determines that the target partial regression coefficient is significant, that is, the use/non-use of know-how information 121, which is the explanatory variable corresponding to the partial regression coefficient, affects the change in the state information, which is the objective variable. The given threshold here is, for example, 0.05, but other values may be used.

In addition, in the above formula (2), the better the condition of the person being assisted, the larger the value of the objective variable, reward. Therefore, when the partial regression coefficient is positive, the target know-how information 121 may contribute to improving the condition, and when the partial regression coefficient is negative, the target know-how information 121 may contribute to worsening the condition. Therefore, when the p-value is smaller than 0.05 and the partial regression coefficient is positive, the importance determination unit 115 determines that the importance of the corresponding know-how information 121 is high.

The importance determination unit 115 may determine the importance in two stages, high and low. In the above example, know-how information 121 whose p-value is smaller than the threshold and whose partial regression coefficient is positive is of high importance, and other know-how information 121 is of low importance. However, the importance determination unit 115 may determine the importance in three or more stages depending on the value of the p-value and the value of the partial regression coefficient.

In this embodiment, as described above, it is easy to register know-how information 121 that represents tacit knowledge. For example, a large number of users each register their own tacit knowledge as know-how information 121, and the storage unit 120 can accumulate a huge amount of know-how information 121. For example, the number of pieces of know-how information 121 may reach tens of thousands to hundreds of thousands. In this regard, the importance determination unit 115 performs the above process to determine the importance of the know-how information 121.

The number of explanatory variables that significantly affect the above-mentioned objective variable and have a positive partial regression coefficient is not limited to one, and there may be multiple explanatory variables. That is, the importance determination unit 115 may identify multiple pieces of know-how information 121 as know-how information 121 with high importance. For example, the importance determination unit 115 determines that, of the n pieces of know-how information 121 corresponding to a1 to an, two pieces of know-how information 121, as and at, have high importance. s and t are integers between 1 and n, and s ≠ t.

In this case, although as and at can be used individually, it is believed that the quality of care can be improved by using a combination of the two. Therefore, when the importance determination unit 115 determines that the importance of multiple pieces of know-how information 121 is high in one multiple regression analysis, the importance determination unit 115 may store information specifying the combination in the storage unit 120.

<Attributes of the person receiving assistance>
By the above process, among the multiple know-how information 121, know-how information 121 that is considered important for improving and maintaining the condition of the person being assisted can be specified. However, if the attributes of the person being assisted differ greatly, the important assistance for the target person being assisted may also differ. The attributes here are, for example, status information at the start of the target period. For example, a person being assisted with level 1 support, whose current condition is relatively good, can often do things by himself, and assistance to prevent the transition to needing care and assistance to promote independence are effective. On the other hand, for a person being assisted with level 3 or higher support, it is considered that thorough assistance in a facility is effective in preventing the condition from worsening. In other words, important know-how information 121 may differ between a person being assisted with a relatively good condition and a person being assisted with a relatively bad condition.

The storage unit 120 may therefore store attribute information representing the attributes of the person being assisted. The processing unit 110 identifies important know-how information for assisting the person being assisted who has the given attribute, based on the association information corresponding to the person being assisted who has been determined to have the given attribute based on the attribute information. Note that the important know-how information is information that is found, for example, based on regression analysis, and specifically, multiple association information corresponding to multiple people being assisted who have been determined to have the given attribute may be used for processing.

For example, the attribute information is status information, and the importance determination unit 115 performs a process of identifying know-how information 121 that is important for a care recipient with status information of 1 (needing support 1) based on the association information acquired for a care recipient with status information of 1 (needing support 1). The importance determination unit 115 also performs a process of identifying know-how information that is important for a care recipient with status information of 2 (needing support 2) based on the association information acquired for a care recipient with status information of 2 (needing support 2). The same applies to other statuses, and the importance determination unit 115 determines the importance of the know-how information 121 for each of the care recipients with care needs 1 to 5. Note that here, an example is shown in which the status information is information of seven levels, and there are seven types of attributes represented by the attribute information. In this case, the importance determination unit 115 performs an importance determination for each of the seven types of attributes. However, the relationship between the status information and the attribute information can be modified in various ways, such as processing the support recipients with status information of 1 and 2 together as one attribute.

By taking into account the attributes of the person being assisted in this way, it becomes possible to improve the processing accuracy of the importance determination unit 115. In other words, it becomes possible to appropriately identify know-how information 121 suitable for improving and maintaining the condition for each attribute.

Note that the attributes in this embodiment are not limited to status information at the start of the period. For example, as described above as the attributes of the person being assisted in the know-how information 121 of FIG. 5, the attributes here may include information such as the age, sex, height, weight, medical history, and medication history of the person being assisted. The attributes may also include physical evaluation data that represents a physical evaluation of the person being assisted, and the physical evaluation data may include information such as rehabilitation history, risk of falling, and risk of bedsores.

<Example of processing using importance>
The processing unit 110 may also perform processing to present recommended know-how information recommended for assisting a given person being assisted, based on the know-how information 121 used for assisting a given person being assisted and the know-how information 121 determined to be of high importance by the importance determination unit 115. In this way, it becomes possible to encourage the caregiver assisting the person being assisted to use the know-how information 121 of high importance, i.e., the know-how information 121 considered to be useful for improving and maintaining the condition of the person being assisted.

More specifically, the processing unit 110 may determine the recommended know-how information based on the attribute information of the person being assisted as described above. For example, a first attribute to an m-th attribute are defined as attributes, and the importance determination unit 115 performs a process of determining know-how information 121 with a high degree of importance for each of the first attribute to the m-th attribute. m is an integer of 2 or more. Then, when it is determined that the target person being assisted has the k-th attribute, the processing unit 110 identifies the recommended know-how information based on the know-how information 121 with a high degree of importance corresponding to the k-th attribute. k is an integer of 1 or more and m or less. In this way, it becomes possible to recommend know-how information 121 that matches the attributes of the person being assisted.

The recommended know-how information is, for example, know-how information 121 that is similar to the know-how information 121 currently in use and has a high level of importance. In addition, the recommended know-how information is, for example, know-how information 121 that is effective when used in combination with the know-how information 121 currently in use. Each example will be explained below using an example of a display screen.

Figure 16 is a service usage screen of the information processing system 10 of this embodiment, and is an example of a user page on which information about a given user is displayed. The user here is a caregiver who performs care, and may be a family member of the person being assisted, or a caregiver working in a care facility or the like. The user page is displayed on the display unit 240 of the terminal device 200, for example, based on the control of the processing unit 110 of the server system 100. The processing unit 110 may generate and transmit the image shown in Figure 16. The processing unit 110 may also transmit information for image generation to the terminal device 200, and the processing unit 210 of the terminal device 200 may generate the image.

As shown in FIG. 16, the user page includes an area RE1 that displays know-how information 121 being used by the target user, and an area RE2 that displays registered know-how information 121. In RE1, for example, the know-how information 121 included in the list information 123 described above with reference to FIG. 11 is displayed. In RE2, for example, the know-how information 121 registered by the target user is displayed. By displaying in this manner, the user can easily grasp the registration and usage status of the know-how information 121.

The user page may also include a display for proposing replacement of the know-how information 121 being used. For example, the processing unit 110 identifies know-how information 121 that is similar to the know-how information 121 that has been determined to have high importance. The importance is determined by the importance determination unit 115 as described above. The similar know-how information 121 is determined by the similarity determination unit 113 as described above. For ease of explanation, the know-how information 121 that has been determined to have high importance by the importance determination unit 115 is referred to as important know-how information. Furthermore, the know-how information that has been determined by the similarity determination unit 113 to be similar to the important know-how information is referred to as know-how information to be replaced.

Then, when the know-how information 121 being used by the target user includes know-how information to be replaced, the processing unit 110 proposes replacing the know-how information to be replaced with important know-how information. For example, the importance determination unit 115 determines by multiple regression analysis that the importance of the know-how information 121 corresponding to a34 is high. Furthermore, the similarity determination unit 113 determines that the similarity between a34 and a5 is high. In this case, a34 is important know-how information, and a5 is know-how information to be replaced.

As shown in FIG. 16, the target user is currently using know-how information 121 corresponding to a5. In FIG. 16, a5 is written as "if5-then5" using the start condition and assistance action. The same is true for the other know-how information 121. Since a5 and a34 are similar, they can be used, for example, for the same type of assistance and in similar situations. For example, a5 and a34 are both know-how information for appropriately determining the pace of eating in meal assistance. Therefore, even if the know-how information 121 currently in use is replaced with similar know-how information 121, it is considered that the user will be able to use the know-how information 121 in similar situations as before the replacement.

And a34 is determined to be more important than a5. That is, by replacing a5 with a34, the user can use the know-how information 121 in a similar situation and provide assistance that is more useful for improving and maintaining the condition of the person being assisted.

In the example of FIG. 16, "if5-then5" corresponding to the know-how information 121 currently in use is displayed in the RE1 area, while "if34-then34", which is important know-how information recommended for replacement, is displayed as Replaced tacit knowledge. By suggesting replacement with know-how information 121 that is similar to the know-how information 121 currently in use and has a high level of importance, it becomes possible to change the assistance provided by the user to one of higher quality that contributes more to the condition of the person being assisted.

The user page may also include a display suggesting know-how information 121 that is recommended for use in combination with the know-how information 121 being used.

As described above, when there are multiple pieces of know-how information 121 that are determined to be of high importance, the importance determination unit 115 stores information that identifies the combination in the storage unit 120. When the know-how information 121 that the user is using is included in the combination, the processing unit 110 performs processing to suggest to the user that the other know-how information 121 included in the combination be added.

For example, the importance determination unit 115 determines that the four pieces of know-how information 121, a6, a7, a10, and a34, are highly important, and stores information about this combination in the storage unit 120. In the example of FIG. 16, the user is currently using the know-how information 121 of a6. Therefore, the processing unit 110 proposes adding a7, a10, and a34 and combining them with a6, rather than using a6 alone. Note that in the example of FIG. 16, a34 is displayed in RE1 as a replacement, so the know-how information 121 to be proposed for addition may be the two pieces of know-how information a7 and a10.

For example, as shown in FIG. 16, the user page may include an RE3 that displays know-how information 121 that is recommended for use. In the RE3, for example, know-how information 121 that is recommended to be combined with the know-how information 121 currently in use is displayed as "Recommend tacit knowledge." In the above example, "if7-then7" and "if10-then10" corresponding to a7 and a10 are displayed in the RE3. In this way, it is possible to suggest the combined use of multiple pieces of know-how information 121 that are considered to be useful for improving and maintaining the condition of the person being assisted, thereby further improving the quality of assistance.

As shown in FIG. 16, the area shown in RE1 may display the know-how information 121 currently in use as well as the devices associated with that know-how information 121. The devices displayed here are the devices included in the list information 123 as shown in FIG. 8 or FIG. 9. In this way, information on the devices used in the know-how information 121 currently in use can be presented to the user in an easy-to-understand manner.

As shown in FIG. 16, the area shown in RE2 may display information indicating the registration status of devices related to the know-how information 121 registered by the target user. For example, for know-how information 121 to which the device in FIG. 6 is not associated (No in step S203), an object including the text "No device" indicating this is displayed.

In addition, if a device is associated (Yes in step S203), but sufficient sample data has not been collected (the loop of steps S207-S210 is ongoing), and the know-how information 121 is not yet ready for the user to add correct answer data, an object containing the text "Not ready" to that effect is displayed.

In addition, since sufficient sample data has been collected (the loop of steps S207-S210 ends), an object containing the text "ready" indicating that the know-how information 121 is ready for the user to add correct answer data is displayed in the know-how information 121. For example, the user can select an object that is displayed as "ready," which starts the processing from step S211 in FIG. 6 onward.

Although not shown in FIG. 16, for know-how information 121 for which the user has added correct answer data, determined the second processing algorithm and parameters (step S216), and created the registration information 122 (step S217), an object including the text "completed" indicating this may be displayed. As described above, by displaying not only information identifying the know-how information 121 but also information associated with the know-how information 121, it is possible to clearly present the usage status and registration status of the know-how information 121 by the target user.

Although not shown in FIG. 16, the above-mentioned importance may be used to determine the priority of association with devices. By preferentially associating know-how information 121 with a high importance with devices, it becomes easier to use know-how information 121 that is considered useful for improving and maintaining the condition of the person being assisted.

For example, the processing unit 110 may promote the collection of sample data by displaying a screen that prompts the user to actively perform the target assistance for know-how information 121 that is highly important but "Not ready." The processing unit 110 may also promote association with a device by displaying a screen that prompts the user to perform the processing from step S211 onward for know-how information 121 that is highly important but "ready."

<Modifications in importance judgment and presentation of the results>
In the above, an example has been described in which the judgment result of the importance judgment unit 115 is binary data indicating whether or not the information is important. In the above description, the know-how information 121 judged to be important in one regression analysis becomes important know-how information, and when multiple pieces of important know-how information are detected in one regression analysis, it is judged that it is useful to use all of them in combination. However, whether or not the know-how information is important and whether or not it is useful when combined may be judged using different conditions.

For example, the importance determination unit 115 may determine the importance in three or more stages as described above. For example, the importance determination unit 115 determines know-how information 121 whose importance is equal to or greater than a given first threshold th1 as important know-how information. Furthermore, when there are multiple pieces of know-how information 121 whose importance is equal to or greater than a second threshold th that is greater than the first threshold th1, the importance determination unit 115 may determine that it is useful to combine and use these pieces of know-how information 121. In other words, the importance determination unit 115 may determine that it is useful to combine some pieces of know-how information 121 that are particularly important among the important know-how information.

For example, consider a case where the importance determination unit 115 determines that the four pieces of know-how information 121, a6, a7, a10, and a34, are highly important, and further determines that the combination of a6 and a7 is useful. In this case, a6, a7, a10, and a34 become candidates for replacement. For example, as shown in FIG. 16, if a5 used by the user is similar to a34, information representing a34 is displayed in the Replaced tacit knowledge item.

In the above example, a6 and a7 are candidates for know-how information 121 that is recommended to be used in combination. For example, as shown in FIG. 16, if the user is using a6, a7 is presented as the recommended know-how information 121. In this way, it becomes possible to flexibly determine the information to be displayed on RE1 and the information to be displayed on RE3 based on different conditions.

The importance determination unit 115 may also repeatedly execute the regression analysis in a chronological order. For example, the importance determination unit 115 performs a regression analysis at a timing tx based on the association information acquired in the period from t1 to tx, and performs a regression analysis at a timing ty based on the association information acquired in the period from tx to ty. For example, x and y are integers that satisfy x<y, and ty represents a timing later than tx. However, the specific method can be modified in various ways, such as performing a regression analysis at a timing ty based on the association information acquired in the period from t1 to ty.

When the results of repeated regression analyses are obtained in this manner, the importance determination unit 115 may make a determination by integrating the results of multiple regression analyses.

For example, the importance determination unit 115 determines that know-how information 121 whose importance has been determined to be high a predetermined percentage or more of the times in multiple regression analyses is important know-how information. For example, the importance determination unit 115 determines that a1 is important know-how information when the number of times that a1 is determined to be important in multiple regression analyses is equal to or greater than a given threshold value th3.

The importance determination unit 115 also determines that a plurality of pieces of know-how information 121 whose number of times of being simultaneously determined to be high in importance is equal to or greater than a given threshold value th4 are know-how information 121 that are useful to use in combination. For example, if a1 is determined to be important in a given regression analysis, and a2 is determined to be important in the same regression analysis, the importance determination unit 115 counts up the count value for the combination (a1, a2). The importance determination unit 115 performs a count-up process for all combinations based on the results of multiple regression analyses. For example, if the importance of (a1, a2, a3) is determined to be high in a given regression analysis, the importance determination unit 115 may count up the count values for the combinations (a1, a2), (a1, a3), (a2, a3), and (a1, a2, a3). The importance determination unit 115 identifies combinations whose final count value is equal to or greater than th4, and presents the above-mentioned RE3 based on the identified combination. Therefore, for example, in the judgment using th3 above, even if a1 is greater than or equal to th3 and a2 is greater than or equal to th3, if the number of times they appear together is small, the combination of (a1, a2) will not be judged to be useful.

In this way, it is possible to flexibly determine what know-how information is important and what know-how information 121 is useful to use in combination.

The importance determination unit 115 may also limit the know-how information 121 that is often used in combination in advance. For example, the importance determination unit 115 identifies a combination of know-how information 121 that is frequently used based on a plurality of pieces of association information. For example, it is assumed that the number of pieces of association information that include all of the four pieces of know-how information 121 (a1, a2, a3, a4) among the plurality of pieces of association information is equal to or greater than a given threshold th5. In this case, the importance determination unit 115 may determine the importance of the four pieces of know-how information 121 (a1, a2, a3, a4) based on a plurality of pieces of association information that include the four pieces of know-how information 121 (a1, a2, a3, a4). Then, when a plurality of pieces of know-how information 121 with an importance equal to or greater than a given threshold th6 are detected among (a1, a2, a3, a4), the importance determination unit 115 determines that the plurality of pieces of know-how information 121 are know-how information 121 that are useful to be used in combination. For example, if the importance of a1 and a3 is equal to or greater than th6, and the importance of a2 and a4 is less than th6, the importance determination unit 115 determines that the combination of (a1, a3) is useful.

In this way, by determining the importance of multiple pieces of know-how information 121 that are frequently used in combination, it is possible to identify know-how information 121 that is considered to be particularly important among the combinations. Note that although the combination (a1, a2, a3, a4) is shown here as an example, similar processing is possible for other combinations.

In addition, the method of this embodiment broadly includes determining the importance of know-how information 121 based on changes in state information and the know-how information 121 used, and the specific method is not limited to the above example.

3.3.2 Importance Based on Selection by Caregiver The importance determination unit 115 may determine the second importance based on the degree of use or popularity by users of each piece of know-how information 121. While the above-mentioned importance is information that takes into account the state of the person being assisted, the second importance is information that is based on the judgment of the caregiver providing the care.

For example, the importance determination unit 115 may count the number of times each piece of know-how information 121 has been downloaded for use. The number of downloads represents how many users have determined that the target know-how information 121 is useful. Thus, the importance determination unit 115 determines that the second importance of the target know-how information 121 is higher the greater the number of downloads.

The importance determination unit 115 may also count the number of users currently using each piece of know-how information 121. The number of users also indicates how many users have determined that the target know-how information 121 is useful. Therefore, the importance determination unit 115 determines that the second importance of the target know-how information 121 is higher the greater the number of users.

In this embodiment, each user may be able to evaluate the know-how information 121 that he or she has used. Various forms of evaluation are possible, but for example, each user may assign a score to the know-how information 121. The importance determination unit 115 may obtain a statistic (such as an average value) of the scores assigned to the target know-how information 121, and determine the second importance based on the statistic.

Furthermore, the number of downloads, the number of users, and the score representing the evaluation are not limited to being used individually, and two or more of them may be combined. For example, the importance determination unit 115 may perform a process of determining the second importance based on a given function that inputs two or more of the number of downloads, the number of users, and the score representing the evaluation.

The processing unit 110 may also perform processing to present second recommended know-how information recommended for assisting a given person being assisted, based on one or more pieces of know-how information 121 used for assisting a given person being assisted and know-how information 121 determined by the importance determination unit 115 to have a high second importance. In this way, it is possible to encourage the use of know-how information 121 with a high second importance, i.e., know-how information 121 that the caregiver considers useful. The second recommended know-how information is, for example, know-how information 121 that is similar to the know-how information 121 being used and has a high second importance.

3.4 Determination of the Status of the Person Being Assisted As described above, in the method of this embodiment, status information of the person being assisted may be used to determine the importance of the know-how information 121. The status information is obtained based on, for example, the results of an investigation by an investigator. However, in this embodiment, part or all of the investigation by the investigator may be automated.

The information processing device of this embodiment may perform a process of estimating the evaluation results of each item in the survey. Below, machine learning will be described as a specific example of a method for estimating the evaluation results. However, the method of this embodiment is not limited to using machine learning, and various modifications are possible. Also, below, an example will be described in which NN is used as machine learning, but other machine learning methods such as SVM may be used, or a method that is an extension of NN or SVM may be used.

Figure 17 shows an example of the basic structure of a neural network. Each circle in Figure 17 is called a node or neuron. In the example of Figure 17, the neural network has an input layer, two or more intermediate layers, and an output layer. The input layer is I, the intermediate layers are H1 and Hn, and the output layer is O. In the example of Figure 17, the input layer has two nodes, the intermediate layers have five nodes, and the output layer has one node. However, the number of intermediate layers and the number of nodes in each layer can be modified in various ways. Figure 17 shows an example in which each node in a given layer is connected to all nodes in the next layer, but this configuration can also be modified in various ways.

The input layer accepts input values and outputs them to the intermediate layer H1. In the example of FIG. 17, the input layer I accepts two types of input values. Each node in the input layer may perform some processing on the input value and output the processed value.

In a NN, weights are set between two connected nodes. W1 in Figure 17 is the weight between the input layer I and the first hidden layer H1. W1 represents the set of weights between a given node included in the input layer and a given node included in the first hidden layer. For example, W1 in Figure 17 is information that includes 10 weights.

At each node in the first hidden layer H1, the outputs of the nodes in the input layer I connected to that node are weighted and summed using weight W1, and a bias is then added. Furthermore, at each node, the output of that node is found by applying an activation function, which is a nonlinear function, to the sum result. The activation function may be a ReLU function, a sigmoid function, or another function.

The same is true for subsequent layers. That is, in a given layer, the output of the previous layer is weighted and added using the weight W, a bias is added, and then the activation function is applied to determine the output to the next layer. The output of the output layer is the output of the NN.

As can be seen from the above explanation, in order to obtain desired output data from input data using a NN, it is necessary to set appropriate weights and biases. In learning, training data is prepared in which given input data is associated with correct answer data that represents correct output data for that input data. The NN learning process is a process of determining the most likely weights based on the training data. Note that various learning methods such as backpropagation are known for the NN learning process. In this embodiment, these learning methods can be widely applied, so detailed explanations are omitted. In addition, the NN is not limited to the configuration shown in FIG. 17, and a CNN or RNN (recurrent neural network) may be used.

Figure 18 is a diagram illustrating input data and output data of a NN for determining the presence or absence of paralysis, etc., used to estimate an evaluation result regarding "presence or absence of paralysis, etc.," which is an example of a survey item. The input data here may include output data of a sensor placed in the living environment of the person being assisted. The output data of the sensor is referred to as sensing data. The input data may also include information regarding one or more pieces of know-how information 121 used in assisting the person being assisted.

For example, when estimating an evaluation result regarding the presence or absence of paralysis, the sensing data includes at least one of data detecting the muscle mass of the person being assisted, data detecting myoelectric potential, and data capturing an image of the target area. Muscle mass may be detected, for example, using a weight scale (body composition scale). Small, portable body composition scales have also become known in recent years, and the specific sensor shape can be modified in various ways. Myoelectric potential is detected by fixing a sensor having multiple electrodes to the body surface of the person being assisted. Data capturing an image of the target area is acquired by a camera (image sensor) capable of capturing an image of the person being assisted. The camera here may be fixed to the room or bed of the person being assisted, or may be mounted on the terminal device 200 or headset 300 used by the caregiver.

The input data also includes information indicating whether or not know-how information 121 on "how to change diapers when there is contracture or paralysis" has been used. This know-how information 121 is information used when changing the diaper of a person being assisted who has contracture or paralysis in the limbs, and for example, the assistance action includes the correct action for changing a diaper. When there is contracture or paralysis, the movement of the limbs of the person being assisted is restricted, so a correct action different from that when there is no contracture or paralysis is used. When know-how information 121 on "how to change diapers when there is contracture or paralysis" is used, there is a high probability that the person being assisted has contracture or paralysis. Therefore, whether or not this know-how information 121 has been used is input data that is useful when estimating the evaluation result of the presence or absence of paralysis, etc. The value of the input data here is binary data that, for example, becomes a first value when the target know-how information 121 is being used, and becomes a second value when it is not being used.

The input data also includes information indicating whether or not know-how information 121 on "how to change position when there is contracture or paralysis" has been used. This know-how information 121 is information used when changing the position of an assisted person who has contracture or paralysis in the limbs, and for example, the assistance action includes the correct action for changing position. In this case, too, it is considered that the correct action is different from when there is no contracture or paralysis. When know-how information 121 on "how to change position when there is contracture or paralysis" is used, there is a high probability that the assisted person in question has contracture or paralysis. Therefore, whether or not this know-how information 121 has been used is useful input data when estimating the evaluation results of the presence or absence of paralysis, etc.

The input data also includes information indicating whether or not know-how information 121 on "how to do exercises and how to move the body to prevent contracture and paralysis" has been used. The assisting action in this case may be the correct exercises or body movements performed by the person being assisted, or the correct movements of the assistant when having the person perform the exercises, etc. When know-how information 121 on "how to do exercises and how to move the body to prevent contracture and paralysis" is being used, there is a high probability that the person being assisted is at risk of developing paralysis, etc. Therefore, whether or not this know-how information 121 has been used is useful input data when estimating the evaluation results of the presence or absence of paralysis, etc.

The evaluation result of the presence or absence of paralysis, etc. is, for example, the result of selecting all the applicable numbers for "1. None," "2. Left upper limb," "3. Right upper limb," "4. Left lower limb," "5. Right lower limb," and "6. Other." Note that "6. Other" refers to the absence of limbs, etc.

The output data is therefore six pieces of data representing the likelihood that "1. None" will be selected, "2. Left upper limb" will be selected, "3. Right upper limb" will be selected, "4. Left lower limb" will be selected, "5. Right lower limb" will be selected, and "6. Other" will be selected. For example, each of the six pieces of data is numerical data between 0 and 1, and the closer the value is to 1, the more likely the target number should be selected.

The configuration of the NN for determining the presence or absence of paralysis, etc. is not limited to that shown in FIG. 18, and various modifications of the input data and output data are possible. For example, some of the sensing data in the input data may be omitted, and other sensing data may be added to the input data. In addition, with regard to the know-how information 121, input data relating to some of the know-how information 121 may be omitted, and input data relating to other know-how information 121 may be added.

In the learning stage, training data for creating a NN for determining the presence or absence of paralysis, etc. is obtained by matching the above input data with correct answer data for multiple individuals being assisted. For example, input data is obtained based on sensing data from sensors placed in the living environment of the individuals being assisted, or know-how information 121 used for the individuals being assisted. The know-how information 121 is identified, for example, from list information 123 of caregivers who assist the individual being assisted.

The correct answer data may also be provided by an expert with specialized knowledge, such as an investigator. The expert will conduct an investigation, for example, following the certified investigator text shown in the above URL, and select all of the above six numbers that apply. For example, if the expert selects only "1. No" and does not select the other five, the correct answer data will be data in which the value corresponding to "1. No" is 1 and the values corresponding to the other five are 0.

For example, the processing unit 110 of the server system 100 acquires training data and performs machine learning based on the training data to create a neural network for determining the presence or absence of paralysis, etc. Note that the machine learning may be performed by a device other than the server system 100.

Figure 19 is a flowchart explaining the learning process for generating a NN for determining the presence or absence of paralysis, etc. When this process is started, first, in step S501, the processing unit 110 acquires input data for learning. The input data here is as described above, and includes, for example, sensing data and information indicating whether or not know-how information 121 has been used. Note that the input data may include information indicating the results of communication by the person being assisted. For example, the input data may include the output of a communication robot. Details of the communication robot will be described later.

In addition, in step S502, the processing unit 110 acquires the correct answer data associated with the input data. For example, the processing unit 110 executes the processes of steps S501 and S502 by reading out one of the data sets of the training data acquired in the learning stage.

In step S503, the processing unit 110 performs a process of updating the weights of the NN. Specifically, the processing unit 110 inputs the input data acquired in step S501 to the NN for determining the presence or absence of paralysis, etc., and acquires output data by performing forward calculations using the weights at that stage. The processing unit 110 determines an objective function based on the output data and the correct answer data. The objective function here is, for example, an error function based on the difference between the output data and the correct answer data, or a cross-entropy function based on the distribution of the output data and the distribution of the correct answer data.

The processing unit 110 updates the weights, for example, so that the error function decreases. The above-mentioned backpropagation method and other methods are known as weight update methods, and these methods can be widely applied in this embodiment.

In step S504, the processing unit 110 determines whether or not to end the learning process. For example, multiple data sets included in the training data may be divided into learning data and validation data. The processing unit 110 may end the learning process when the process of updating the weights has been performed using all of the learning data, or may end the learning process when the accuracy rate based on the validation data exceeds a given threshold.

If the learning process is not to be terminated, the processing unit 110 returns to step S501 and continues the process. That is, the processing unit 110 reads a new data set from the training data and performs a process of updating the weights based on the new data set.

When the learning process is terminated, the processing unit 110 stores the NN for determining the presence or absence of paralysis, etc. at that stage in the storage unit 120 as a learned model. The learned model includes an algorithm for performing forward calculations and a weighting coefficient. Note that FIG. 19 is an example of the learning process, and the method of this embodiment is not limited to this. For example, methods such as batch learning are also widely known in machine learning, and these methods can be widely applied in this embodiment.

Figure 20 is a flowchart explaining the processing of the status determination unit 114 in the inference stage. When this processing starts, first in step S601, the status determination unit 114 determines whether or not the current timing is the timing to obtain status information of the person being assisted. For example, as described above with reference to Figure 15, if the status information is obtained at a predetermined interval such as one month, the status determination unit 114 determines whether or not the predetermined interval has elapsed since the previous processing.

If it is determined that it is not the time to request status information, the status determination unit 114 ends the process without performing steps S602 and onward.

If it is determined that it is time to obtain status information, in step S602, the status determination unit 114 acquires input data related to the person being assisted that is the processing target. For example, the memory unit 120 acquires and stores sensor information collected by a group of sensors placed in the living environment of the person being assisted. The status determination unit 114 performs a process of reading out data from the collected data that is related to the target person being assisted and that is to be used as input data. For example, the status determination unit 114 reads out the output of a body composition scale, the output of a sensor that detects myoelectricity, the output of a camera that has captured an image of the target body part, etc.

The state determination unit 114 also identifies know-how information 121 that has been used for the person being assisted, based on the list information 123 of the caregivers who are assisting the person being assisted. Specifically, the state determination unit 114 determines whether or not the three pieces of know-how information 121, "how to change a diaper when there is contracture or paralysis," "how to change posture when there is contracture or paralysis," and "how to do exercises and move the body to prevent contracture or paralysis," have been used for the person being assisted.

In step S603, the state determination unit 114 reads out the NN for determining the presence or absence of paralysis, etc. from the storage unit 120. The input data acquired in step S602 is then input to the NN for determining the presence or absence of paralysis, etc., and output data is obtained by performing forward calculations. The output data of the NN for determining the presence or absence of paralysis, etc. is, for example, six numerical data representing the likelihood that each of the six numbers will be selected, as described above. The state determination unit 114 determines that the target number will be selected, for example, when the numerical value is equal to or greater than a threshold value th that satisfies 0<th<1. For example, when the numerical data corresponding to "1. No" is equal to or greater than th and the other five numerical data are less than th, the state determination unit 114 selects "1. No" as the evaluation result regarding the presence or absence of paralysis, etc., and outputs an inferred result that the other five are not selected.

In this way, the method of this embodiment makes it possible to automatically estimate the evaluation result for the survey item of the presence or absence of paralysis, etc. Note that the above description has been given using the specific item of the presence or absence of paralysis, etc. as an example, but the ability to estimate the evaluation result based on the use or absence of sensing data and know-how information 121 is similar for other items.

Figure 21 shows an example of the configuration of a NN when estimating evaluation results for p survey items. Figure 21 shows an example in which one NN is created for one item. Each NN acquires, as input data, sensing data, information indicating whether or not know-how information 121 was used, and information indicating the results of communication using a communication robot. However, each NN does not need to accept all three types of input, and some may be omitted.

In particular, as shown in FIG. 21, a filtering process may be performed to extract a portion of the know-how information 121 according to the target item with respect to whether or not it has been used. As described above, the amount of know-how information 121 can be enormous. If filtering is not performed, the majority of the input data will be determined by whether or not the know-how information 121 has been used, and there is a risk that the influence of the sensing data and the results of the interview by the communication robot on the output will be extremely reduced. In this regard, filtering can be performed to limit the amount of know-how information 121 that is input data, making it possible to accurately estimate the evaluation results.

The sensors that output the sensing data are not limited to the body composition scale, electromyography sensor, and imaging sensor (camera) described above, but various sensors can be used, such as motion sensors such as acceleration sensors and angular velocity sensors, excretion detection sensors such as pressure sensors and odor sensors, position sensors such as GPS (Global Positioning System), sleep detection sensors that detect pulse and heart rate, temperature sensors, humidity sensors, illuminance sensors, and air pressure sensors. Each NN does not need to accept all of these as inputs, and may acquire only a portion of the sensing data, for example, as described below using Figures 23A to 23E.

As described above, various information is registered by multiple users in the know-how information 121. Figures 22A to 22C are examples of the know-how information 121. Figure 22A is an example of the know-how information 121 used in meal assistance to assist a person receiving care to eat. For example, in meal assistance, the caregiver grasps the characteristics of the person receiving care and explains them to the person receiving care in an easy-to-understand manner, thereby performing an assistance action to make the meal run smoothly. For example, in the case of a person receiving care who has a characteristic of low chewing ability, if the caregiver understands this, it is possible to take measures to prevent aspiration, and it is also useful to guide the person receiving care such as "The rice has been softened, so chew it well." Number 2 in Figure 22A is know-how information 121 for "communicating (making the caregiver understand) the characteristics of the user" to the caregiver, and for example, when a predetermined start condition is met, the caregiver is made to perform an assistance action of obtaining and viewing data representing the characteristics of the person receiving care. As described above, the assistance behavior may include an action in which the caregiver communicates the characteristics of the person being assisted to the person being assisted. The same is true for other know-how information 121, and the know-how information 121 shown in FIG. 22A includes information for supporting various actions of the caregiver in meal assistance.

Figure 22B is an example of know-how information 121 used in excretion assistance to assist the excretion of a person being assisted. Note that excretion assistance may be performed in a toilet or using a diaper. Numbers 66-72 represent know-how information 121 for excretion assistance in a toilet, and Numbers 73-75 represent know-how information 121 for excretion assistance using a diaper.

Figure 22C is an example of know-how information 121 used in transfer assistance and movement assistance to assist the person being assisted in transferring or moving. Note that the presence or absence of equipment or the type of equipment used for transfer and movement assistance varies depending on the condition of the person being assisted and the availability of lifts, etc. In the example of Figure 22C, Numbers 92-103 represent know-how information 121 for assistance using a wheelchair, Numbers 104-107 represent know-how information 121 for assistance using a cane, and Numbers 108-112 represent know-how information 121 for assistance using a lift.

Furthermore, the know-how information 121 is not limited to this, and know-how information 121 used in situations other than eating, excretion, and transfers/movements may be used in the processing. Furthermore, the "How to change diapers when there is contracture or paralysis" described above using FIG. 18 is know-how information 121 used in excretion assistance, and is information for dealing with more specific situations than the example shown in FIG. 22B. In this way, the know-how information 121 used in eating, excretion, and transfers/movements is not limited to FIGS. 22A to 22C, and various modifications are possible.

A communication robot is a robot for communicating with a person receiving care. The communication robot may be, for example, a humanoid robot with two arms, and may be a device capable of voice recognition and voice synthesis. When using such a communication robot, while having a conversation with the person receiving care, it is possible to ask the person receiving care whether he or she can make the same movement by, for example, bending his or her arms. However, the communication robot is not limited to being humanoid, and may be one that carries out a conversation by performing voice recognition and voice synthesis. In this case, the communication robot may be realized by a device such as a PC, and may be capable of carrying out a conversation with an avatar displayed on a display. Furthermore, communication with the person receiving care is not limited to a conversation using voice, and may be one that uses text. For example, at least one of the speech by the communication robot and the response by the person receiving care may be performed using text. The results of communication using a communication robot are useful mainly as input data when evaluating cognitive function.

Figures 23A to 23E show examples of evaluation content indicating what type of evaluation is performed for each item, and specific input data when estimating the evaluation result for that item. The input data is listed separately for the sensing target, tacit knowledge (know-how information 121), and communication robot. In the column related to the communication robot, a circle is written if it is used, and it is left blank if it is not used. Figure 23A shows items related to physical functions and daily living activities. Figure 23B shows items related to daily living functions. Figure 23C shows items related to cognitive functions. Figure 23D shows items related to mental and behavioral disorders. Figure 23E shows items related to adaptation to social life. Numbers such as 1-1 correspond to the numbers written in Figure 14A or Figure 14B.

As shown in FIG. 23A, many of the physical functions and daily living activities are used to evaluate whether the person being assisted can perform a specific movement, and sensing data such as the movement of specific parts of the body during the movement, contact with specific locations, and load distribution are used as input data. For some items, the use or non-use of know-how information 121 is also used as input data. On the other hand, the output of the communication robot is only used for some items such as hearing.

As shown in FIG. 23B, the daily living functions relate to eating, excretion, transfers/movements, cleanliness, etc., and therefore many items use sensing data such as whether the person being assisted is receiving assistance. Sensing data may be, for example, captured images. In addition, the information shown in FIG. 22A to FIG. 22C can be used to determine whether know-how information 121 is being used. Number in FIG. 23B refers to the number in FIG. 22A to FIG. 22C. The output of the communication robot is not used in the example of FIG. 23B.

As shown in FIG. 23C, the behavior of the person being assisted is important in estimating the evaluation results for cognitive function. Therefore, the output of the communication robot serves as input data for many items. The state determination unit 114 may also acquire and use responses to questions as sensing data using an image sensor, microphone, etc. Regarding the use or non-use of know-how information 121, it is not used in the example of FIG. 23C.

As shown in FIG. 23D, the words and actions of the person being assisted are also important in estimating the evaluation results for items related to mental and behavioral disorders. Therefore, the output of the communication robot serves as input data for many items. The state determination unit 114 may also acquire and use responses to questions as sensing data using an image sensor, microphone, etc. Detection results regarding how many times and how frequently a given behavior was performed within a certain period of time may also be used as sensing data. In the example of FIG. 23D, know-how information 121 is not used.

As shown in FIG. 23E, many items related to adaptation to social life use sensing data that indicates the detection results of how many times and how frequently a certain behavior was performed in a certain period of time. For taking medicine, the use or non-use of know-how information 121 is used. There may also be items that use the output of a communication robot, such as maladaptation to groups.

As shown in Fig. 21 and Fig. 23A to Fig. 23E, the method of this embodiment makes it possible to estimate the evaluation results for each item used when determining the condition information of the person being assisted from the use or non-use of sensing data and know-how information 121. In particular, by using the use or non-use of know-how information 121, it becomes possible to take into account the assistance being provided to the person being assisted, thereby improving the estimation accuracy.

For example, the processing unit 110 (status determination unit 114) may perform a process of calculating status information based on know-how information used to assist the person being assisted and output data from sensors placed in the living environment of the person being assisted. That is, in addition to the process of estimating the evaluation results in the survey items, the status determination unit 114 may calculate status information based on the estimation results. For example, if the status information is a level of care required, the status determination unit 114 obtains information that indicates any of levels 1 to 5 of care required as status information. In this way, it becomes possible to automatically calculate the status information. Therefore, for example, it becomes possible to easily realize a process of determining the importance of know-how information 121 based on a change in the status information.

However, the method of this embodiment is not limited to automatically calculating status information. For example, the estimated results for each item obtained by the method of this embodiment may be provided to an investigator. In this case, the investigator visits the person being assisted, as in the conventional method. However, since the investigator only needs to confirm whether the estimated results are appropriate, the burden on the investigator can be reduced compared to investigating each item from scratch, and the investigation time can be shortened.

The processing unit 110 may also perform a primary judgment regarding the condition information based on the estimation results for each item, and provide the results of the primary judgment to a specialist such as a doctor. In this case, the doctor or other specialist determines the final result of the condition information based on the results of the primary judgment. Since an investigation by an investigator is omitted, it becomes easier to obtain the condition information.

<Proposal for action>
As explained above, sensor output data may be used as input data for estimating the evaluation results of a survey item. However, a certain amount of sensing data may be required to make a judgment using the sensing data.

For example, as shown in FIG. 23A, the state determination unit 114 may use sensing data indicating fluctuations in body pressure and load when determining the evaluation result of "turning over" 1-3. In this case, for example, if the load on the bed is greatly reduced, it is considered that the person being assisted has grabbed onto something or has been assisted by a caregiver, and the state determination unit 114 estimates that the ability may be low.

However, being able to turn over on one's own is not limited to the fact that the load does not decrease in all of the sensing data. For example, if the load does not decrease significantly in sensing data that is equal to or exceeds a threshold value representing a predetermined percentage of the sensing data corresponding to multiple turns, it may be determined that the patient is able to turn over on one's own. In this case, it is important to consider the percentage of sensing data that meets a given condition among the sensing data acquired during the period from the previous state determination to the current state determination, so it is desirable to have a relatively large number of sensing data. For example, if only one piece of sensing data is acquired when turning over, either the load has decreased in all of the sensing data or the load has not decreased in all of the sensing data will be obtained, which is an extreme result, and there is a risk of reducing the accuracy of the estimation result regarding turning over.

The number of sensing data is also important for other sensing data, and using a small amount of sensing data as input data may reduce the estimation accuracy of the evaluation results.

Therefore, when the processing unit 110 determines that the amount of sensor output data is insufficient in a given determination period, the processing unit 110 may output an instruction to the person being assisted or the caregiver assisting the person being assisted to take a specific action. In this way, the opportunities to obtain sensing data increase, making it possible to increase the input data in the state determination unit 114 and thereby improve the estimation accuracy.

Figure 24 is a diagram explaining the process of determining whether the amount of sensing data is insufficient. The horizontal axis of Figure 24 represents time, and the vertical axis represents the amount of sensing data. Also, t1 and t2 represent the timing for acquiring status information, similar to Figure 15.

For example, the state determination unit 114 estimates the evaluation result based on the sensing data in the period from t1 to t2 in order to obtain state information at t2 in FIG. 24. The state determination unit 114 determines in advance the number of pieces of sensing data required for each piece of sensing data that is to be input data for each item, for example. At a given timing between t1 and t2, the state determination unit 114 estimates the number of pieces of sensing data to be obtained at time t2 based on the number of pieces of sensing data obtained up to that point. For example, the state determination unit 114 may assume that the number of pieces of sensing data increases linearly as shown in FIG. 24, or may obtain an approximation function that expresses the transition in the number of pieces of sensing data from t1 and estimate the number of pieces of sensing data at time t2 based on the approximation function. If the estimated value of the number of pieces of sensing data at time t2 is less than the number of pieces of sensing data required, the state determination unit 114 determines that there is a shortage of sensing data.

For example, if the state determination unit 114 determines that there is insufficient sensing data regarding turning over, it may output a suggestion to the person being assisted to turn over. Alternatively, the state determination unit 114 may output a suggestion to the caregiver to have the person being assisted turn over. In this way, it is possible to obtain sensing data before the actual timing of t2, thereby improving the estimation accuracy of the evaluation result at t2.

Note that the instruction output by the state determination unit 114 is not limited to direct suggestions such as turning over or getting up. For example, the state determination unit 114 may output a suggestion to perform a series of actions related to multiple evaluation items.

FIG. 25 is a diagram in which the contents of suggestions made by the status determination unit 114 correspond to survey items for which sensing data can be acquired by the suggestions. For example, the status determination unit 114 may make a suggestion such as "Let's go shopping on XX day until XX." In order for a person being assisted to go shopping at a specified location, preparations such as brushing teeth, washing face, styling hair, cutting nails, and changing clothes are necessary. In addition, during the preparation stage and after going out, actions such as standing on both feet or one foot, walking, and standing up are performed. In other words, by making shopping suggestions, it becomes possible to acquire sensing data related to the multiple items shown in FIG. 25 all at once. In other words, it becomes possible to efficiently collect a large amount of sensing data without making individual suggestions.

In addition, as shown in FIG. 25, the state determination unit 114 may make suggestions for cleaning or preparing to go out. If it is difficult to get ready to go out, the state determination unit 114 may make suggestions for washing the mouth, styling the hair, and putting on and taking off clothes. In this case, too, since carrying out each suggestion requires actions such as standing up or walking, it is possible to efficiently collect sensing data.

3.5 GPO (Group Purchasing Organization)
When medical facilities purchase medical equipment, they may use a GPO (Group Purchasing Organization). GPOs are in the business of specializing in price negotiations with manufacturers and other distributors, and provide members with a service that reduces unit prices by committing to purchasing large lots. Even when the minimum purchase lot desired by a manufacturer is large, by using a GPO, member medical facilities can keep costs down while purchasing only the necessary amount of high-priced products. For example, in the United States, many medical facilities are members of GPOs, and purchase various medical equipment through the GPO.

GPOs provide purchasing conditions (contracts), and when members use those contracts, they pay a portion of the purchase price as a commission to the GPO. The contents of the contracts vary, including those that set prices for each manufacturer and those that set discounts according to purchase volume.

Computer systems and methods suitable for GPO are described in U.S. patent application Ser. No. 15/783,992, filed on October 13, 2017, entitled "COMPTER-BASED SYSTEMS SPECIFICALLY CONFIGURED TO MANAGE SOFTWARE OBJECTS THAT ARE INTERRELATED VIA TRIGGER CONDITIONS AND METHODS OF USE THEREOF" and "METHODS AND SYSTEMS FOR PROVIDING IMPROVED MECHANISM FOR UPDATED HEALTHCARE" No. 16/985,609, filed August 5, 2020, entitled "METHOD FOR USE INFORMATION SYSTEMS," which are incorporated herein by reference in their entireties.

FIG. 6 of U.S. Patent Application No. 15/783,992 discloses an example of a screen for a buyer to create an RFP (Request for Proposal) by inputting contract parameters, discount conditions, etc. In this example, an RFP is created based on a product category specification using a product classification code such as the UNSPSC (United Nations Standard Products and Services Code), and the created RFP is sent to one or more suppliers.

When a supplier responds with a specific product based on the RFP, the buyer is presented with a screen corresponding to FIG. 22, for example. FIG. 22 is a screen that serves as an interface when selecting a product. In FIG. 22, the buyer can select a product according to his/her needs while referring to multiple products.

FIG. 10 of U.S. Patent Application No. 16/985,609 also shows a screen for evaluating the effect of replacing a given product with another product.

As can be seen from these descriptions, it is important for GPOs to propose appropriate products in accordance with buyers' requests. The method of this embodiment may be used for consulting to GPOs, specifically for supporting product proposals by GPOs.

For example, when multiple devices are associated with a given know-how information 121, the processing unit 110 may perform a process of presenting a device other than the first device among the multiple devices as an alternative device to the first device among the multiple devices.

Figure 26 is an example of the know-how information 121, registration information 122, and list information 123 described above. As described above, when a registered user performs the process shown in Figure 6, a device for determining either the start condition of the know-how information 121 or the assisting action is associated with the know-how information 121. Also, when each user performs the process shown in Figure 10 (particularly step S308), a device for determining either the start condition of the know-how information 121 or the assisting action is associated with the know-how information 121. As a result, as shown in Figure 26, one or more devices associated with the given know-how information 121 can be specified. Note that, as shown in Figure 26, by using the know-how information 121 itself, more detailed information such as specific start conditions and assisting actions can be associated with the device.

As described above, Device 1 included in the registration information 122 is a device designated by a registered user to determine start conditions, etc. Device 1a included in the list information 123 is a device designated by another user to use the registered user's tacit knowledge. In other words, multiple devices associated with a given know-how information 121 are all devices for determining the same start conditions, etc., and are therefore highly likely to be similar. The same is true for Device 1b.

Therefore, for example, if a buyer is considering replacing Device 1, the processing unit 110 performs processing to propose Device 1a and Device 1b as alternative devices. In this way, it becomes possible to identify and present products that meet the user's requirements from a perspective other than product classification codes such as UNSPSC.

Furthermore, when a buyer is considering replacing a device used in a given know-how information 121, for example, the processing unit 110 may propose one or more devices associated with similar know-how information similar to the know-how information 121 as alternative devices. The similar know-how information is determined based on the first similarity determination process as described above. The know-how information 121 and the similar know-how information have high similarity, for example, between texts expressing start conditions and in the type of assistance used. Therefore, the know-how information 121 and the similar know-how information are likely to be used in similar situations, and the device associated with the similar know-how information is also considered to be similar to the device to be replaced. By using similar know-how information in this way, it is possible to increase the number of devices that can be presented and support a wide range of proposals.

The method of this embodiment does not need to be fixed to a method of proposing a device using know-how information 121. For example, the processing unit 110 may be able to switch between a process of determining an alternative device using a code such as UNSPSC and a process of determining an alternative device using know-how information 121. For example, the processing unit 110 determines whether to use a code or know-how information 121 based on a user input.

The processing unit 110 of this embodiment may also perform a process of identifying, from the plurality of know-how information 121, fifth know-how information that has a high similarity to the fourth know-how information to which no device is associated and to which a device is associated. The processing unit 110 performs a process of determining a supplier of a device associated with the fifth know-how information as a supplier of a device for determining the start condition of the fourth know-how information.

Figure 27 is a diagram explaining the process of identifying a supplier based on the fourth know-how information and the fifth know-how information. For example, the know-how information 121 of ID 43 corresponds to the fourth know-how information. The know-how information 121 of ID 43 does not have corresponding registration information 122, and is not associated with a device. The similarity determination unit 113 searches for similar know-how information similar to the know-how information 121 of ID 43 based on the first similarity determination process. For example, the know-how information 121 of ID 10 is similar know-how information and corresponds to the fifth know-how information.

Here, as shown in FIG. 27, there is corresponding registration information 122 for know-how information 121 of ID10, and Device10 is associated as the device. The storage unit 120 also stores information that associates the device with the supplier that supplies the device. For example, Supplier10 is associated with Device10.

In this case, the processing unit 110 performs processing to propose Supplier 10 as a supplier of the device to be used in the know-how information 121 of ID 43. As described above, the know-how information 121 of ID 43 and the know-how information 121 of ID 10 are similar, so there is a high probability that a device similar to Device 10 can be used to automatically determine the know-how information 121 of ID 43. In other words, the device used to automatically determine the know-how information 121 of ID 43 has a high affinity with Supplier 10 and may be developed and provided by Supplier 10.

As described above, since no device is associated with the know-how information 121 of ID 43, there is a possibility that devices suitable for automatically determining start conditions and assistance actions are not widely available on the market. However, since it has been registered as know-how information 121, it represents the tacit knowledge of some user and may be useful in assistance situations. In this regard, the method of this embodiment makes it possible to present information for automating the processing of know-how information 121 that could not be handled by existing devices. As a result, it becomes possible to develop a new sensing device market, etc.

28 is an example of a screen presenting recommended suppliers. In the example of FIG. 28, the ID number of know-how information 121 to which no device is associated and the recommended supplier associated with the know-how information are displayed. The Category in FIG. 28 indicates a category determined based on a product classification code such as UNSPSC. For example, in the example of FIG. 28, Supplier 1 and Supplier 2 are presented as suppliers of the device corresponding to the know-how information 121 of ID 11. In addition, when there are multiple devices associated with the fifth know-how information as in FIG. 26, or when multiple pieces of fifth know-how information similar to one piece of fourth know-how information are selected, there may be multiple recommended suppliers. Similarly, Supplier 10 is presented as a supplier of the device corresponding to the know-how information 121 of ID 43. The same applies to the other know-how information 121. In this way, it is possible to present the recommended suppliers for each piece of know-how information 121 in an easy-to-understand manner.

At this time, the processing unit 110 may present a ranking (second importance) of each piece of know-how information 121. As described above, the index for determining the ranking may be the number of downloads, the number of users using the information, an evaluation value, or a combination of these. For example, the know-how information 121 of ID 11 has a high ranking, and if a device that automates this determination is supplied, it is thought that many users would like to use it. In this way, the ranking serves as a material to encourage suppliers to supply new devices, so it is useful to use it in processing. For example, FIG. 28 shows a screen that displays know-how information 121 that satisfies a given condition, out of multiple pieces of know-how information 121 that are not associated with a device, in order of ranking.

The processing unit 110 may also present the importance of each piece of know-how information 121, taking into account the condition of the person being assisted. The importance here corresponds to the care to improve quality in FIG. 28, and in FIG. 28, the importance is displayed in three levels, A to C. As described above, the importance taking into account the condition of the person being assisted is determined, for example, by multiple regression analysis based on changes in condition information and whether or not know-how information has been used. For example, the know-how information 121 of ID 11 has a high ranking, and also has a high degree of contribution to improving and maintaining the condition of the person being assisted. In other words, since it is clear that the importance is high from the perspective of both the caregiver and the person being assisted, this serves as material to encourage suppliers to supply new devices.

On the other hand, there may be know-how information 121 such as ID 43 that is highly rated by the caregiver but is determined to be of low importance when the condition information of the person being assisted is taken into account. Conversely, there may be know-how information 121 such as ID 44 that is relatively low rated by the caregiver but is determined to be of high importance when the condition information of the person being assisted is taken into account. By displaying the importance from two perspectives as shown in FIG. 28, the amount of information regarding the degree of importance of the know-how information 121 increases, making it possible for the viewer (e.g., GPO) to make an appropriate judgment.

Here, an example is shown in which two importance levels are displayed on a screen that presents recommended suppliers to the GPO. However, the situations in which the importance level and the second importance level are used are not limited to this. For example, in the search process described above with reference to FIG. 10, two importance levels may be displayed when displaying search results. In this way, it is possible to present two importance levels to various targets, such as caregivers working in nursing facilities or hospitals, managers who supervise the caregivers, home helpers, and people receiving care. Alternatively, the search processing unit 112 may perform a search process based on the importance levels. In addition, two importance levels may be used when determining the priority order when associating devices, etc. with know-how information 121.

Furthermore, the screen on which the two degrees of importance are presented is not limited to that shown in FIG. 28. For example, the importance of given know-how information 121 may be illustrated on a plane in which the number of downloads or the number of users is the first axis and the importance based on the status information of the person being assisted is the second axis. Alternatively, the processing unit 110 may perform processing to calculate an overall importance based on the two degrees of importance and present the overall importance. In addition, various modifications of the specific content to be presented are possible.

4. Modifications In the above embodiment, multiple specific examples of the know-how information 121 are shown, but many other examples are possible, some of which are listed below. For example, the know-how information 121 may include (1) information for selecting an appropriate food form for each person being assisted, such as a patient, (2) information suggesting whether end-of-life care should be started after a specified period for each person being assisted, such as a patient, and information suggesting the timing for changing the content of care after the start of end-of-life care, (3) information regarding the timing for the caregiver to stop providing food, (4) information regarding the caregiver's response when the person being assisted swallows food while eating, and (5) information for detecting situations where there is a high risk of falling.

(1) Know-how information 121 for selecting an appropriate food type for each person being assisted, such as a patient, receives five types of information as input information: the results of a doctor's or other medical diagnosis regarding chewing and swallowing ability, the requests of the person being assisted or his/her family, whether the person being assisted appears to be unable to chew the food provided to them, whether the person being assisted appears to be swallowing food (choking, coughing, etc.) while eating, and whether the person being assisted expresses negative emotional information (indexed values of discomfort, disgust, sadness, surprise, fear, etc.) while eating. In addition, know-how information 121 for selecting an appropriate food type for each person being assisted, such as a patient, outputs the food type (for example, regular food, chopped food, extra chopped food, soft vegetables, blended food, thickened food, or jelly food) as output information.

Here, the results of the doctor's diagnosis may be, for example, the number of times a repetitive saliva swallowing test (a test to see how many times saliva can be swallowed in succession after continuously swallowing for 30 seconds) or the results of the doctor's judgment may be classified and tagged for each classification and entered as information. The requests of the person being assisted or their family may be classified and tagged for each classification and entered as information. For example, a request such as "I would like them to eat regular food if possible" and a request such as "I would like them to eat with safety as a priority" may be tagged differently. Information regarding whether the person being assisted is not chewing the food provided to them, whether they are swallowing (choking, coughing, etc.) while eating, and whether they are expressing negative emotional information is entered by extracting the appearance of not chewing or swallowing from a video of the person being assisted eating taken with a camera.

(2) In the know-how information 121 that suggests whether end-of-life care should be started for each patient or other person receiving care after a predetermined period of time, five types of information are input as input information: the amount or ratio of each type of meal (for example, main dish or side dish, or each ingredient such as meat or fish), the amount of water intake, the timing of intake, information on illness, and weight (or BMI). In the know-how information 121 that suggests whether end-of-life care should be started for each patient or other person receiving care after a predetermined period of time and whether it is time to change the content of care after starting end-of-life care, information indicating whether end-of-life care should be started after a predetermined period of time and whether it is time to change the content of care after starting end-of-life care is output as output information. End-of-life care refers to care for a person receiving care who is considered to be likely to die in the near future. End-of-life care differs from normal care in that it places emphasis on relieving physical and mental pain, and supporting the person receiving care to live a dignified life. Furthermore, even while providing end-of-life care, the condition of the person receiving care may change over time, which may lead to changes in the type of care appropriate for the patient. In other words, by presenting the timing for starting end-of-life care and the timing for changing the content of care during end-of-life care, it becomes possible to provide appropriate care to the person receiving care until the very end. For example, experienced caregivers have tacit knowledge to estimate the timing and content of care required for end-of-life care from various perspectives, such as the amount of food eaten, and by digitizing this tacit knowledge, other caregivers will also be able to provide appropriate end-of-life care.

The system for providing the care information related to (1) and (2) to the caregiver includes a terminal device 200 and a server system 100, as shown in FIG. 1, for example. The terminal device 200 here is, for example, a PC. As described above, the information processing system 10 of this embodiment can be realized in various ways, and the server system 100 may be omitted. The terminal device 200 acquires input information including the above five types of information. The terminal device 200 may automatically acquire the input information using a sensor, or may acquire the input information based on the input operation of the caregiver. In addition, in a care facility or the like, care software may be used separately from the information processing system 10 according to this embodiment. The care software is software for storing attributes and care history of the person being assisted. Conventionally, various types of care software have been used, and the care software here can be widely applied to them. The information processing system 10 may acquire input information including the above five types of information from the care software.

When the terminal device 100 receives an instruction to start analysis based on the input information, it sends each piece of input information to the server system 200, and the server system 200 outputs the analysis results to the terminal device 100. Figure 29A is an example of My Page in this case. The My Page in Figure 29A is a screen that presents information about a given user, similar to the user page shown in Figure 16, and displays know-how information 121 that the target user is using and know-how information 121 that the user has registered.

In the example of FIG. 29A, the target user is using know-how information 121 to estimate whether it is the timing to start end-of-life care, which corresponds to (2) above. The My Page shown in FIG. 29A includes an object for uploading in the area corresponding to the know-how information 121. When a selection operation of the object is performed, the terminal device 200 uploads the input information related to the target person being assisted to the server system 100.

The server system 100 obtains an analysis result, for example, by inputting the uploaded input information into a trained model for end-of-life care that has been created in advance. The server system 100 transmits the analysis result to the terminal device 200. The terminal device 200 presents information representing the analysis result to the user.

Figure 29B is an example of a screen that displays the analysis result, and is a screen that is displayed on the display unit of the terminal device 200, for example. The analysis result is, for example, as described above, a judgment result of whether or not end-of-life care should be started after a predetermined period of time, or a judgment result of whether or not it is time to change the content of care after the start of end-of-life care. For example, as shown in Figure 29B, the analysis result may include a time series change in the feature amount calculated based on the input information and a judgment result of whether or not end-of-life care should be started after a predetermined period of time. The feature amount here may be information that is judged to be important among the input information, such as a moving average of the amount of food eaten, or information that is calculated based on the above five pieces of input information. For example, when an NN is used, the feature amount may be the output of a given intermediate layer or output layer. In Figure 29B, the actual value of the feature amount up to a predetermined timing (for example, October) and the estimated value of the feature amount thereafter (for example, after November) are presented, and the start timing of end-of-life care determined using the estimated value is displayed. In this way, it is possible to appropriately present information regarding end-of-life care to the user.

The frequency with which it is necessary to determine the timing of starting end-of-life care, etc., is sufficiently low compared to the frequency with which determinations are made regarding daily assistance such as eating, excretion, and transfers. Therefore, as described above, by using a user operation on the terminal device 200 as a trigger for uploading and analyzing, it is possible to reduce the communication load and processing load. However, the processing of the know-how information 121 regarding end-of-life care is not limited to being triggered by a user operation, and may be automatically executed when the target input information is collected.

Furthermore, in the information related to (1) and (2) above, information equivalent to the input information may be managed using nursing software. Therefore, as described above, some or all of the input information may be acquired via the nursing software. In this case, the information processing system 10 according to this embodiment may be linked to the nursing software. For example, the analysis results by the server system 100 may be displayed on the display screen of the nursing software.

Figure 30A is an example of a display screen of the care software. For example, the care software displays a screen in which the contents of care performed for a given person being assisted are arranged in chronological order. However, the display screen of the care software is not limited to this, and various modifications are possible. For example, each time any input information is input to the care software, the input information may be sent to the server system 100 and automatically analyzed. The care software may display, for example, a mark indicating the analysis result in a predetermined field on the display screen. In the example of Figure 30A, a mark including an exclamation mark is displayed in the display field of the intake amount or intake ratio for each type of food used as one of the input information. Figure 30B is an example of a display screen when, for example, an operation to select the mark in Figure 30A is performed. In the example shown in Figure 30B, the same analysis result as in Figure 29B is popped up on the display screen of the care software. In this way, by linking the information processing system 10 according to this embodiment with the care software, for example, the care record and the analysis result can be displayed together, making it possible to present appropriate information to the user. Note that the display method of the analysis result on the care software is not limited to Figure 30B, and various modifications are possible.

(3) In the know-how information 121 regarding the timing for the caregiver to stop providing food, three types of information are input as input information: "If the person being assisted continues to be unable to chew the food provided to them," "If the person being assisted frequently swallows food (such as choking or coughing) while eating," and "If the person being assisted appears sleepy while eating," and when any one of the three types of information is met, the output information is "Stop providing food."

(4) In know-how information 121 regarding the response of a caregiver when an assisted person aspirates while eating, "If the assisted person aspirates and loses their posture" is input as input information, and when this condition is met, "Please check their posture" is output as output information. "If the assisted person aspirates and looks sleepy" is input as input information, and when this condition is met, "They look sleepy. Please call out to them" is output as output information. "If the assisted person aspirates and does not look sleepy or lose their posture" or "If the assisted person aspirates and the food texture is not appropriate" is input as input information, and when this condition is met, "Please check the food (e.g. food texture)" is output as output information. "If the assisted person's aspirating (choking, coughing, etc.) increases while eating" is input as input information, and when this condition is met, "Stop providing food" is output as output information.

Here, each piece of input information related to the know-how information 121 in (3) and (4) above is extracted from output data such as a video captured by a camera and/or a waveform from a wearable device that measures swallowing (for example, as described in U.S. Patent Application No. 16/276,768, filed on February 15, 2019, entitled "Swallowing action measurement device and swallowing action support system," which is incorporated herein by reference in its entirety).

As shown in FIG. 31, the system for providing the caregiver with assistance information related to (3) and (4) above includes a wearable device 400 for measuring swallowing, a first terminal device 200A, a second terminal device 200B, and a server system 100. The person being assisted wears the wearable device 400 around his/her neck. The first terminal device 200A is placed, for example, on a table where the person being assisted eats, and has a function of capturing images of the person being assisted eating. The first terminal device 200A is, for example, a camera device or a smartphone with an app installed. The second terminal device 200B is a terminal carried by the caregiver, and has a function of receiving notifications via the app. The first terminal device 200A communicates with the wearable device 400 to exchange data. In addition, FIG. 31 shows an example in which the server system 100, the first terminal device 200A, and the second terminal device 200B are connected via a network NW such as a LAN or the Internet, and the second terminal device 200B and the wearable device 400 are directly connected using short-range wireless communication or the like. However, the specific connection mode of each device can be modified in various ways. In the embodiment shown below, an example will be described in which the first terminal device 200A captures an image of one person being assisted, but this is not limited thereto, and the first terminal device 200A may capture images of multiple people being assisted and output output information for each person being assisted.

The first terminal device 200A transmits an image of the person being assisted and output data of the wearable device 400 to the server system 100. Based on the image and output data, the server system 100 executes the process related to (3) or (4) above, and requests output information such as "Stop providing meals."

On the app of the first terminal device 200A, not only are the data of the wearable device 400 and the captured image displayed, but output information is also displayed. Figure 32A is an example of a screen displayed on the display unit of the first terminal device 200A. The display unit here corresponds to, for example, the display unit 240 of Figure 3. On the screen shown in Figure 32B, an object OB1 indicating data measured by the wearable device 400 is superimposed on a partial area of the captured image IM. The information contained in the object OB1 is, for example, a graph showing the time series change in the measured value by the wearable device 400.

In addition, an object OB2 showing output information acquired from the server system 100 is superimposed on another part of the captured image IM. The output information includes, for example, as shown in (4) above, outputs such as "Please check your posture", "You look sleepy. Please speak to the person", "Please check your meal", and "Please stop eating". The output information may also be capable of outputting information such as "Choice not detected", which indicates that output is unnecessary for any of the above. For example, in the object OB2 shown in FIG. 32A, the text "Choice not detected" is displayed together with a first mark including a check mark indicating a normal state. FIG. 32B is an example of other information displayed as the object OB2. For example, the object OB2 may be an object that displays text such as "Please check your posture", "You look sleepy. Please speak to the person", and "Please check your meal" together with a second mark including an exclamation mark. The object OB2 may also be an object that displays text such as "Please stop eating" together with a third mark including an exclamation mark. In the example of FIG. 32B, the second mark is a mark with an exclamation mark surrounded by a circle, and the third mark is a mark with an exclamation mark surrounded by a triangle.

"No choking detected" is information presented when no abnormality is detected in the person being assisted, and is of low urgency. The other four are of relatively high urgency because some abnormality has been detected. For example, "Please check your posture," "Looks sleepy. Please call out to the person," and "Please check the meal" correspond to situations in which attention is required when providing food. Furthermore, the information "Please stop eating" is particularly of high urgency because it corresponds to a situation in which choking occurs frequently. As described above with reference to Figures 32A and 32B, by using the first mark to the third mark together with the corresponding text, it is possible to appropriately convey the condition of the person being assisted to the caregiver. Furthermore, the presentation method according to the urgency is not limited to different marks and texts, and the color of the object OB2 may be changed. For example, the object OB2 including the first mark may be displayed in green, the object OB2 including the second mark may be displayed in yellow, and the object OB2 including the third mark may be displayed in red.

By displaying this information on the first terminal device 200A, the caregiver (caregiver A in FIG. 31) who is providing meal assistance near the person being assisted can properly grasp the condition of the person being assisted. Specifically, the caregiver can not only visually check the person being assisted, but also view the sensor output (object OB1) and the output information from the server system 100 (object OB2), enabling the caregiver to provide appropriate assistance.

The terminal device 200 may also include lighting (light-emitting unit) such as an LED. The terminal device 200 may control the lighting in response to the output information. Specifically, when it is OK to continue providing food to the person being assisted (when normal), the lighting shows green, when care is required when providing food to the person being assisted, the lighting shows yellow, and when food provision is to be stopped, the lighting shows red. As a result, the caregiver (caregiver A in FIG. 31) who is providing food assistance to the person being assisted can recognize the notification based on the output information presented using the lighting of the first terminal device 200A. Note that, although an example in which the server system 100 calculates the output information has been shown above, this is not limiting, and the first terminal device 200A may calculate the output information. In this case, the first terminal device 200A performs display processing of the output information calculated by itself.

The output information is not limited to that presented on the first terminal device 200A. For example, the output information may be presented on the second terminal device 200B used by a caregiver (caregiver B in FIG. 31) who is in a position where the first terminal device 200A cannot be directly seen. For example, caregiver B is a person in charge of assisting the assisted person with meals, but the assisted person is in relatively good condition and is likely to be able to eat independently. In this case, caregiver B may set the first terminal device 200A in a predetermined position and have the assisted person start eating, and then perform other assistance with a higher priority. In this case, by presenting the output information on the second terminal device 200B carried by caregiver B, caregiver B can return to assisting the assisted person with meals as necessary. Alternatively, caregiver B does not have to be the person in charge of the original meal assistance. For example, if caregiver A, who is in charge, is busy helping another person being assisted, caregiver B can provide appropriate help by presenting output information on the second terminal device 200B carried by caregiver B.

Figures 33A and 33B are examples of screens displayed on the second terminal device 200B. Figure 33A is an example of a lock screen (notification screen) of the second terminal device 200B. For example, as shown in Figure 33A, the second terminal device 200B may display objects OB21-OB24 corresponding to output information. Objects OB21-OB24 all correspond to object OB2 in Figures 32A and 32B. Object OB21 corresponds to "Please stop eating" out of the above-mentioned output information. Object OB22 corresponds to "Please check your eating." Object OB23 corresponds to "You look sleepy. Please speak to me." Object OB24 corresponds to "Please check your posture."

Notifications corresponding to objects OB21-OB24 are not limited to being all notifications. For example, output information to be notified may be selected according to the user's level of proficiency. For example, a setting may be made to notify all output information to alert the person being assisted as much as possible for a user with a low level of proficiency, and notifications may be limited to output information with a high level of urgency for a user with a high level of proficiency. In the above example, "Please stop eating" is more urgent than the other three, so for example, when targeting a user with a high level of proficiency, object OB21 may be the display target, and objects OB22-OB24 may not be the display targets.

The second terminal device 200B may also display a status screen showing detailed information about the output information. FIG. 33B is an example of a status screen. For example, the status screen includes the following items: monitor type, user or device, status, and response.

The monitor type is information that identifies the situation (assistance) to be monitored, which in the above example is "eating." The user or device is information that identifies the target person being assisted. This may be the name of the person being assisted, or the ID of the device used to assist the target person being assisted. The device here may be the wearable device 400, the first terminal device 200A, or another device. The status represents output information. For example, "choking not detected" corresponds to normal. "Check your posture," "Looks sleepy. Please speak to him/her," and "Check your eating" correspond to warnings. "Please stop eating" corresponds to stop.

The response is information that indicates the response status by the caregiver. If the status is normal, no action by the caregiver is required. If the status is caution, the caregiver does not necessarily need to take action, and for example, only the output information may be notified. In this case, the presence or absence of notification and information that identifies the user who is the subject of the notification may be displayed as the response. If the status is canceled, it is highly likely that the caregiver needs to take action. Therefore, information that identifies whether the caregiver has responded or not is displayed as the response. If the caregiver has not responded, an object indicating that the caregiver will respond may be displayed as shown in FIG. 33B. For example, if a user with the user name "Sato" performs this operation, the information that indicates the response is updated to information that indicates "Response completed by Sato". This can prevent missed responses in an emergency and prevent multiple caregivers from responding to a specific person being assisted in the same way.

By displaying the status screen shown in FIG. 33B, it is possible to properly grasp the eating status of the person being assisted and to immediately determine whether or not the caregiver needs to take action. Note that, although the above describes a form of notification using the display unit of the second terminal device 200B, the form of notification is not limited to this, and the caregiver may be notified by voice via the headset 300, for example.

Using the know-how information 121 of (3) and (4) above as an example, an example of a screen displayed when using given know-how information 121 will be described. As shown in step S203 of FIG. 6, when registering the know-how information 121, a device may be associated with the know-how information 121. As shown in step S216 of FIG. 6, the know-how information 121 may be associated with processing content for device data. In the case of the know-how information 121 of (3) and (4) above, the device is a camera (terminal device 200) or a wearable device 400. The processing content may be a process for detecting the posture, etc. of the person being assisted from a captured image, or a process for detecting the amplitude, frequency, etc. of a waveform output from the wearable device 400.

However, even if the know-how information 121 is similar, the required device data may differ depending on the know-how information 121. For example, in the know-how information 121 shown in (3) and (4) above or similar know-how information 121, the posture of the person being assisted is detected based on an image captured by a camera. The device data in this case may be image data capable of detecting the posture of the person being assisted, and may be data captured from the front, data captured from the side of the profile, or data captured from an intermediate direction (diagonal direction). For example, the content of the device data differs depending on the direction from which the experienced user who registers the know-how information 121 thinks the person being assisted should be observed. Alternatively, the device data may differ depending on the equipment of the care facility or the like to which the registered user belongs, as the position at which the camera can be installed differs.

That is, the know-how information 121 shown in (3) and (4) above is not limited to one, and may include multiple pieces of know-how information 121 with different device data registered by multiple registered users. In addition, one registered user may register multiple pieces of know-how information 121 with different device data in order to provide the same or similar assistance.

When using such know-how information 121, it is important to match the device data. For example, when using know-how information 121 that performs image processing on device data obtained by imaging a person receiving care from the front to determine whether or not to stop eating, if device data obtained by imaging a person receiving care from the side or at an angle is used as input information, the accuracy of the processing may be reduced. This is because the processing content for the device data in the know-how information 121 is intended to target data obtained by imaging a person receiving care from the front. Therefore, the know-how information 121 may include information for acquiring appropriate device data. For example, when the device is a camera, the information for acquiring appropriate device data is information for specifying the imaging direction when imaging the subject. The information for acquiring appropriate device data may also be information for specifying the imaging range, such as whether the imaging range includes the entire body of the person receiving care, whether it is a bust-up, or whether it is sufficient to include the face, or may be other information for determining the characteristics of the captured image. Furthermore, when the device used is other than a camera, the know-how information 121 may also include information for acquiring appropriate device data.

Figure 34A is another example of My Page. As in Figure 29A, My Page includes know-how information 121 in use and registered know-how information 121. In the example of Figure 34A, the user is using know-how information 121 corresponding to (3) or (4) above. My Page also includes a camera start object that instructs the start of the camera as an object for starting the use of the know-how information 121.

34B is an example of a display screen when an operation is performed on the camera start object shown in FIG. 34A. The screen shown in FIG. 34B includes information for acquiring appropriate device data. The information for acquiring appropriate device data is, for example, a guide object for indicating the direction of the face of the person being assisted in the image. The guide object in FIG. 34B includes an object OB5 representing an outline and an object OB6 which is an illustration of a face. However, either one of the objects OB5 and OB6 may be omitted. Also, information different from either of the objects OB5 and OB6 may be used as information indicating the appropriate direction of the face. For example, text such as "Please take an image of the person being assisted from the front" may be displayed, or the text may be output as a voice using the terminal device 200 or the headset 300. In addition, various modifications are possible for the method of presenting information for acquiring appropriate device data.

For example, the know-how information 121 is associated with information for acquiring appropriate device data based on the sample data accumulated by the processing of steps S207-S210 in FIG. 6. For example, when the server system 100 acquires multiple images as sample data, it determines an average face direction based on the multiple images. The server system 100 may determine objects OB5 and OB6 as information for acquiring appropriate device data based on the determined average face direction, and associate the determined objects OB5, OB6, etc. with the know-how information 121. When the know-how information 121 is used, the terminal device 200 presents information including objects OB5 and OB6 based on the information associated with the know-how information 121, as illustrated in FIG. 34B.

In this way, for example, a user who uses know-how information 121 can appropriately adjust the camera placement, etc., depending on the know-how information 121 being used. Therefore, it becomes possible to obtain input information suitable for the know-how information 121 being used.

Alternatively, information on device data may be presented in the search process of know-how information 121 shown in FIG. 10. For example, in step S309 of FIG. 10, know-how information 121 using data captured from the front, know-how information 121 using data captured from the side, and know-how information 121 using data captured from an oblique angle are presented as different know-how information 121. At that time, for example, sample data acquired in step S210 of FIG. 6 and information for acquiring appropriate device data may be displayed in association with the know-how information 121. When determining the know-how information 121 to be used, the user who performed the search can consider a method of acquiring device data, such as selecting know-how information 121 using data suitable for the user's environment. For example, if a camera capable of capturing an image of a person being assisted from the front is already installed in the user's environment, it is possible to adopt know-how information 121 using data captured from the front in order to reuse the camera, or it is possible to adopt know-how information 121 using data captured from an oblique angle in order to utilize highly rated know-how information 121. In the latter case, for example, the user can make a decision such as, "Since the existing camera is not compatible, I will use the camera on the terminal device 200, which is easier to adjust the position."

Here, the know-how information 121 of (3) and (4) are used as examples to explain a method of associating information for acquiring appropriate device data with the know-how information 121, and a method of presenting information for acquiring appropriate device data when using the know-how information 121. However, these methods are broadly applicable to the know-how information 121 according to this embodiment, and are not limited to the know-how information 121 of (3) and (4).

(5) In the know-how information 121 for detecting situations where there is a high risk of falling, three types of information are input as input information: "If the person cannot walk on his/her own but starts walking without assistance," "If the person has difficulty understanding the situation he/she is in and his/her physical functions and spatial awareness have deteriorated and he/she cannot recognize steps," and "If he/she does not know where the toilet is," and when any one of the conditions of the three types of information is met, the output information is that "This is a situation where there is a high risk of falling."

The above three types of information are extracted, for example, from information from foot pressure sensors (for example, pressure sensors are provided in multiple locations on the insole, and the time series changes in each pressure sensor and the time series changes in the center of gravity position) and video captured by a camera. Only when it is determined that the scene is at high risk of falling, for example, is the output information notified to the caregiver via headset 300 together with information about the person being assisted who is determined to be at high risk of falling. The caregivers who are notified may only be those who are near or in the vicinity of the person being assisted who is determined to be at high risk of falling.

As shown in FIG. 35, the system that provides the caregiver with the above-mentioned (5) assistance information (output information) includes a camera 500, a third terminal device 200C, a fourth terminal device 200D, a server system 100, and a foot pressure sensor 600. The camera 500 is, for example, a wide-area camera, and is installed on the ceiling of the room or on a wall near the ceiling. There may be multiple cameras 500, and FIG. 35 shows two cameras, a camera 500-1 installed on the ceiling and a camera 500-2 installed on the wall. The third terminal device 200C is, for example, placed at a station, and has a function of displaying the state of the person being assisted captured by the camera 500. The fourth terminal device 200D is a terminal device used by a given caregiver. Note that FIG. 35 shows an example in which the server system 100, the third terminal device 200C, the fourth terminal device 200D, the camera 500, and the foot pressure sensor 600 are connected via a network NW such as a LAN or the Internet. However, the specific connection methods for each device can be modified in various ways.

The camera 500 transmits an image of the person being assisted to the server system 100. The foot pressure sensor 600 transmits the sensing results to the server system 100. Based on the acquired information, the server system 100 executes the process related to (5) above and obtains output information indicating the risk of falling.

36 is an example of a screen displayed on the display unit of the third terminal device 200C. The third terminal device 200C displays information on the foot pressure sensor 600 of the assisted person and information indicating the degree of risk in association with the captured image of the assisted person. In the example of FIG. 36, two assisted people are captured on the screen, and objects OB31 and OB32 indicating information on the foot pressure sensor 600 and objects OB41 and OB42 indicating the degree of risk are displayed near each assisted person. The information indicating the degree of risk corresponds to the output information of (5) above. The number of assisted people captured on the image may be one or three or more. The information indicating the information on the foot pressure sensor 600 and the information indicating the degree of risk may be displayed for all assisted people captured in the image, or may be displayed for some assisted people. For example, a known face recognition technology may be used as a means for identifying the assisted people.

For example, the cameras 500 are placed in various locations in a care facility, etc., and the foot pressure sensors 600 are attached to the person being assisted who is to be monitored for falls. The third terminal device 200C is placed in a station in the care facility, etc., and used by a manager, etc. Therefore, by displaying the image shown in FIG. 36 on the third terminal device 200C, it becomes possible to properly grasp the risk of falls within the facility.

Furthermore, the presentation of the output information is not limited to that performed by the third terminal device 200C. For example, the output information may be presented by the fourth terminal device 200D. Figures 37A-37D are examples of screens displayed on the fourth terminal device 200D. Figures 37A and 37B are examples of the lock screen (notification screen) of the fourth terminal device 200D. For example, as shown in Figure 37A, the fourth terminal device 200D may display information indicating that a specific person being assisted is about to fall, or that the specific person being assisted should be monitored to prevent falls, based on the output information in (5) above.

It is also important to determine the level of risk of falling at a given timing, but from the viewpoint of preventing a given assisted person from being injured by falling, it is also important to observe in a chronological manner how the assisted person is walking. In other words, in assistance for preventing falls, chronological information including the history of past fall risks is important. Therefore, the fourth terminal device 200D may notify the chronological output information of the target assisted person as shown in FIG. 37B. For example, when an operation to select "respond" included in a given notification is performed on the screen shown in FIG. 37A, a display as shown in FIG. 37B may be possible. For example, in FIG. 37B, a notification of monitoring is executed at 17:27, but output information calculated for the target assisted person after the notification is also notified in the same object. For example, it is determined that the assisted person to be monitored is likely to fall at 17:28 and 17:29, and output information indicating this is displayed together with the output information notified at 17:27.

The fourth terminal device 200D may also display a status screen showing detailed processing information related to the output information. For example, the status screen shown in FIG. 33B above includes information related to a fall. The status screen related to a fall includes items such as location, time, person who detected it, status, response, and details.

"Location" indicates the location where the person being assisted is likely to fall. For example, the server system 100 stores information in advance that associates the ID of the camera 500 with the installation location, and identifies the location based on this information. "Time" is information indicating the timing at which it was determined that there was a high possibility of falling. "Detector" is information indicating the person being assisted who was determined to have a high possibility of falling, for example, the name of the person being assisted. However, "detector" may also be other information that can be used to identify the person being assisted, such as the ID of the foot pressure sensor 600. "Status" is information indicating whether or not the assistant has responded, and includes information such as "responded" and "not responded." "Response" is an area in which an object is displayed to indicate that the assistant will assist the person being assisted. For example, an object containing the text "respond" is displayed in the response column. "Details" is detailed information indicating the state of the person being assisted who was determined to have a high possibility of falling, and may be, for example, link information for transitioning to a screen described later using FIG. 37D. By displaying such a status screen, it is possible to properly inform the user when, where, and who is in danger of falling, and what measures are being taken in response.

Figure 37C is another example of a status screen. As shown in Figure 37C, the status screen may include time-series information regarding falls of a given assisted person. In this way, it is possible to clearly present the progress of the fall risk of the target assisted person and the caregiver's response at each timing. In the example of Figure 37C, the caregiver did not respond to the fall risk at 17:23, for example, because it was determined that there was no problem after visual inspection. In addition, the fall risks that occurred to the same assisted person at 17:27, 17:28, and 17:29 were responded to by a caregiver with the user name Sato, for example.

Considering the risk of fractures and other injuries to the person being assisted, it is important to prevent falls from occurring in the first place. Therefore, when safety-conscious control is performed, a notification may be sent to indicate that a fall is imminent even in a situation where a fall does not actually occur. In other words, even in a situation where the server system 100 determines that there is a high possibility of a fall, there may be cases where it is truly necessary to take action such as supporting the person being assisted, and cases where simply watching over them is sufficient.

Therefore, the user of the fourth terminal device 200D may view detailed information when it is determined that there is a high possibility of falling. For example, when a selection operation is performed on an object that includes the text "View" displayed in the details column of FIG. 37C, the fourth terminal device 200D may display detailed information and accept feedback based on the detailed information.

Figure 37D is an example of a screen that displays detailed information. As shown in Figure 37D, the detailed information may be a video image of the person being assisted. The video image here may be the captured video image itself, or may be information including an object (corresponding to OB31, etc.) showing information from the foot pressure sensor 600, or an object (corresponding to OB41, etc.) showing the degree of risk, as shown in Figure 36. For example, when the server system 100 determines output information indicating the risk of falling based on a 10-second video image, the video image here is a 10-second video image when it is determined that the risk of falling is high.

After viewing the video, the user of the fourth terminal device 200D inputs in FIG. 37D whether the scene was one in which the person being assisted was really likely to fall, or one in which there was no particular problem and no intervention by the caregiver was required. The input information may be stored in the care facility, or may be fed back to the server system 100 and used to update the trained model that determines the risk of falling.

Note that although the above describes a form of notification using the display unit of the fourth terminal device 200D, the form of notification is not limited to this, and the caregiver may be notified by voice via the headset 300, for example.

In the above, an example has been given in which a status screen is displayed on each terminal device 200 to check the status of each person being assisted for information related to (3) to (5), but a status screen may also be displayed for information related to (1) and (2) above. Furthermore, the know-how information 121 to be displayed on the status screen is not limited to (1) to (5), and devices such as the terminal device 200 may display a status screen related to other information.

Although the present embodiment has been described in detail above, those skilled in the art will readily understand that many modifications are possible that do not substantially deviate from the novelties and effects of the present embodiment. Therefore, all such modifications are intended to be included within the scope of the present disclosure. For example, a term described at least once in the specification or drawings together with a different term having a broader or similar meaning may be replaced with that different term anywhere in the specification or drawings. All combinations of the present embodiment and modifications are also included within the scope of the present disclosure. Furthermore, the configurations and operations of the information processing system, information processing device, server system, and terminal device are not limited to those described in the present embodiment, and various modifications are possible.

10...information processing system, 100...server system, 110...processing unit, 111...registration processing unit, 112...search processing unit, 113...similarity determination unit, 114...state determination unit, 115...importance determination unit, 120...storage unit, 121...know-how information, 122...registration information, 123...list information, 130...communication unit, 200, 200-1, 200-2...terminal device, 200A...first terminal device, 200B...second terminal device, 200C...third terminal device, 2 00D...fourth terminal device, 210...processing unit, 220...storage unit, 230...communication unit, 240...display unit, 250...operation unit, 300, 300-1, 300-2...headset, 400...wearable device, 500, 500-1, 500-2...camera, 600...foot pressure sensor, NW...network, RE1, RE2, RE3...area, OB1, OB2, OB21-OB24, OB31, OB32, OB41, OB42...objects

Claims (7)

a processing unit that receives a request to register know-how information including information used in care, the know-how information including output information to be output to a caregiver and input information for outputting the output information in association with each other;
a storage unit that stores the know-how information based on the registration request;
Including,
The processing unit includes:
Acquire status information that is calculated based on information acquired from an external sensor or is input from the outside and indicates the status of the person being assisted;
Acquire association information that associates a change in the status information based on the status information acquired at a first timing and the status information acquired at a second timing different from the first timing with the know-how information used in assisting the person being assisted between the first timing and the second timing; and
The status information includes at least one of a level of care required, which indicates the degree to which the person being assisted requires care, and an evaluation result of an ADL (Activities of Daily Living) of the person being assisted,
The processing unit includes:
An information processing device that determines the importance of the know-how information so that the importance of the know-how information associated with a change in the status information when the degree of improvement in at least one of the evaluation results of the care level required or the ADL between the first timing and the second timing is high is higher than the importance of the know-how information associated with a change in the status information when the degree of improvement in at least one of the evaluation results of the care level required or the ADL is low. In claim 1,
The storage unit is
storing attribute information representing attributes of the person being assisted;
The processing unit includes:
An information processing device that identifies important know-how information, which is know-how information whose importance in providing assistance to an assisted person having a given attribute is determined to be a predetermined level or higher, based on the association information corresponding to the assisted person who is determined to have the given attribute based on the attribute information. In claim 2,
The processing unit includes:
The information processing device identifies, as the important know-how information, information whose partial regression coefficient is significant based on a regression analysis and a significance test in which a value related to the state information is a response variable and information indicating use/non-use of the know-how information is an explanatory variable. In claim 2 or 3,
The processing unit includes:
When a given person being assisted is determined to have the given attribute, the information processing device presents the know-how information suitable for the given attribute as recommended know-how information recommended for assisting the given person being assisted, based on the know-how information being used in assisting the given person being assisted and the important know-how information. In any one of claims 1 to 4,
The processing unit includes:
An information processing device that calculates the status information based on the know-how information used in assisting the person being assisted and output data from sensors arranged in the living environment of the person being assisted. In claim 5,
The processing unit includes:
An information processing device that, when it is determined that the amount of output data from the sensor is insufficient within a given determination period, outputs instructions to the person being assisted or a caregiver assisting the person being assisted to take a specific action. The computer
receiving a request for registration of know-how information including information used in care, the know-how information including output information to be output to a caregiver and input information for outputting the output information being associated with each other;
storing the know-how information based on the registration request;
Acquire status information that is calculated based on information acquired from an external sensor or is input from the outside and indicates the status of the person being assisted;
Acquire association information that associates a change in the status information based on the status information acquired at a first timing and the status information acquired at a second timing different from the first timing with the know-how information used in assisting the person being assisted between the first timing and the second timing; and
The status information includes at least one of a level of care required, which indicates the degree to which the person being assisted requires care, and an evaluation result of an ADL (Activities of Daily Living) of the person being assisted,
determining the importance of the know-how information so that the importance of the know-how information associated with the change in the condition information when the degree of improvement in at least one of the evaluation results of the care level required or the ADL between the first timing and the second timing is high is higher than the importance of the know-how information associated with the change in the condition information when the degree of improvement in at least one of the evaluation results of the care level required or the ADL is low;
An information processing method for performing the above.