JP7633200B2 - BEHAVIOR ASSESSMENT SERVER AND BEHAVIOR ASSESSMENT METHOD - Google Patents

Description

The present invention relates to a behavior evaluation server and a behavior evaluation method.

International Publication No. WO 2018/100797 (Patent Document 1) is a background technology of this technical field. This publication describes an information processing method including: "acquiring, by an acquisition unit, the results of a physical fitness test and a cognitive function test, and lifestyle information and a social participation score; learning, by a learning unit, a correlation between the physical fitness value and the cognitive function value based on the measurement results and the lifestyle information and the social participation score; and predicting, by a prediction unit, the user's future physical fitness value and cognitive function value based on the user's lifestyle information and social participation score using the learned correlation" (see abstract).

International Publication No. 2018/100797

The technology described in Patent Document 1 predicts a user's future physical strength and cognitive function scores based on a social participation score, but because a user's social participation can be explained by various aspects such as the user's level of activity and interactions, there is a risk that the social participation may not be evaluated accurately. Therefore, one aspect of the present invention evaluates a user's social participation with high accuracy based on the user's level of activity and interactions.

In order to solve the above-mentioned problems, one aspect of the present invention employs the following configuration. The behavior evaluation server includes a processor and a memory, and the memory holds a time series of a first user's location and a time series of the amount of activity of the first user. The processor generates a time series of the first user's behavior based on the time series of the location and the time series of the amount of activity, and calculates a first index indicating an activity ratio indicating the activity level based on the generated time series of behavior, and a second index indicating interaction diversity based on the generated time series of behavior.

According to one aspect of the present invention, it is possible to evaluate a user's social participation with a high degree of accuracy based on the user's activity level and interactions.

Problems, configurations, and advantages other than those described above will become clear from the description of the embodiments below.

1 is a block diagram showing a configuration example of a behavior evaluation system according to a first embodiment. FIG. 2 is a block diagram showing an example of the configuration of a computer constituting each of an evaluation server and a client in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of account information according to the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of intervention service information in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of position information in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of activity amount information in the first embodiment. FIG. 4 is a diagram showing an example of a data configuration of action purpose information in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of map information in the first embodiment. 4 is a diagram illustrating an example of a data configuration of timeline information in the first embodiment. FIG. FIG. 11 is a diagram illustrating an example of a data configuration of activity ratio information in the first embodiment. FIG. 11 is a diagram illustrating an example of a data configuration of exchange diversity information in the first embodiment. FIG. 4 is a diagram showing an example of a data configuration of exercise information in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of behavior evaluation information in the first embodiment. 10 is a diagram showing an example of a data configuration of nursing care risk rank determination information in Example 1. FIG. 13 is a diagram showing an example of a data configuration of total score rank determination information in the first embodiment. FIG. 11 is a diagram illustrating an example of a data configuration of user aggregated evaluation information in the first embodiment. FIG. FIG. 4 is a diagram illustrating an example of a data configuration of behavioral tendency information in the first embodiment. FIG. 4 is a diagram illustrating an example of a data configuration of reference parts information in the first embodiment. 4 is an example of a reference template in the first embodiment. FIG. 11 is a diagram illustrating an example of a data configuration of inverse simulation result information in the first embodiment. 11 is a diagram illustrating an example of a data configuration of intervention service usage history information in the first embodiment. FIG. FIG. 4 is a diagram illustrating an example of a data configuration of learning data in the first embodiment. 4 is a flowchart showing an example of an evaluation process in the first embodiment. 11 is a flowchart illustrating an example of a timeline tallying process according to the first embodiment. 13 is a flowchart showing an example of an activity ratio calculation process in the first embodiment. 11 is a flowchart illustrating an example of an exchange diversity calculation process according to the first embodiment. 11 is a flowchart illustrating an example of a motion calculation process in the first embodiment. 11 is a flowchart showing an example of a behavior evaluation process in which a care-needing risk is used in the first embodiment. 11 is a flowchart showing an example of a behavior evaluation process in which a standard score is used in the first embodiment. 11 is a flowchart illustrating an example of an advice generating process according to the first embodiment. 11 is a flowchart showing an example of an advice generation process in which a care-needing risk is used in the first embodiment. 11 is a flowchart illustrating an example of an advice generating process in which a standard score is used in the first embodiment. 11 is a flowchart illustrating an example of a recommendation process in the first embodiment. 11 is a flowchart showing an example of a nursing care need risk prediction model update process in Example 1. FIG. 13 is a diagram showing an example of an evaluation result display screen in Example 1. FIG. 13 is a diagram showing an example of an evaluation result display screen in Example 1. FIG. 13 is a diagram showing an example of an evaluation result display screen in Example 1. FIG. 13 is a diagram showing an example of a recommendation result display screen in the first embodiment.

The following describes in detail an embodiment of the present invention with reference to the drawings. In this embodiment, the same components are generally designated by the same reference numerals, and repeated explanations are omitted. It should be noted that this embodiment is merely one example for realizing the present invention, and does not limit the technical scope of the present invention.

Figure 1 is a block diagram showing an example configuration of a behavior evaluation system. The behavior evaluation system includes an evaluation server 100 and one or more clients 200 connected via a network such as the Internet. The evaluation server 100 evaluates the behavior of a user having a client 200. The client 200 is a device that measures the user's location and activity level, such as a smartphone or a wearable device.

The evaluation server 100 includes, for example, a timeline aggregation unit 111, an activity ratio calculation unit 112, an interaction diversity calculation unit 113, an exercise calculation unit 114, a care need risk prediction unit 115, a behavior evaluation unit 116, an advice generation unit 117, a recommendation processing unit 118, a processing model update unit 119, and a standard score calculation unit 120, all of which are functional units.

The timeline aggregation unit 111 aggregates a timeline of the user's activities. The activity ratio calculation unit 112 calculates an activity ratio indicating the frequency with which the user performs an activity. In this embodiment, the number of times the user goes out is calculated as an example of an activity ratio, but the activity ratio may be calculated using other indicators (for example, the number of times the user goes out for each activity purpose, such as sports hobbies or cultural hobbies) in addition to or instead of the number of times the user goes out.

The interaction diversity calculation unit 113 calculates interaction diversity, which indicates the diversity of participating organizations and participants in the user's activities. In this embodiment, as an example of interaction diversity, the total number of stay locations of the user and the number of stay locations for each purpose of behavior are calculated, but in addition to or instead of the number of stay locations, the number of people who come into contact with others at the stay locations (obtained, for example, by short-range wireless communication between clients 200), as well as the frequency of contact, age, gender, and proportion of residential area, etc. may be calculated as interaction diversity. Unless otherwise specified, hereinafter, the "number of stay locations" refers to the total number of stay locations of a certain user regardless of the purpose of behavior.

The exercise calculation unit 114 calculates exercise indicating the amount of exercise due to the user's actions. In this embodiment, the average number of steps per day is calculated as an example of exercise, but in addition to or instead of the average number of steps, calories burned, heart rate, walking distance, walking time, travel distance, and travel time may be calculated as exercise. Unless otherwise specified, hereinafter, "average number of steps" refers to the average number of steps taken by a certain user per day. It can be said that activity ratio, interaction diversity, and exercise are all indicators that indicate the activity level of the user's actions, are indicators that measure the user's social participation, and are indicators that contribute to improving the risk of needing nursing care.

The nursing care risk prediction unit 115 predicts the user's nursing care risk (one example of an index that comprehensively evaluates the user's behavior) based on a nursing care risk prediction model. The nursing care risk is the probability that the user will be certified as needing assistance or nursing care within a specified period (e.g., within the next three years). In addition, instead of the nursing care risk, the expected amount of out-of-pocket nursing care expenses that the user will have to pay within a specified period may be used.

The behavior evaluation unit 116 evaluates the user's behavior based on the user's activity rate, interaction diversity, and exercise. The advice generation unit 117 generates advice for the user based on the user's activity rate, interaction diversity, and exercise comparison tendency with people of the same sex and age, and self-tendency. The recommendation processing unit 118 extracts intervention services to be recommended to the user.

The processing model update unit 119 updates the nursing care risk prediction model. The standard score calculation unit 120 calculates, for each user, a standard score indicating the position of each of the activity ratio, interaction diversity, and exercise among all users, and an overall score (an example of an index for overall evaluation of a user's behavior) indicating the position of the overall evaluation of the activity ratio, interaction diversity, and exercise among all users.

The evaluation server 100 holds, for example, account information 131, intervention service information 132, location information 133, activity amount information 134, behavioral purpose information 135, map information 136, timeline information 137, activity ratio information 138, interaction diversity information 139, exercise information 140, behavioral evaluation information 141, nursing care risk rank determination information 142, overall score rank determination information 143, user aggregated evaluation information 144, behavioral tendency information 145, reference part information 146, reference template 147, inverse simulation result information 148, intervention service usage history information 149, and learning data 150, all of which are information. Note that the client 200 may hold some or all of the information held by the evaluation server 100.

Account information 131 includes information about the account, such as the user's attributes. Intervention service information 132 is master information about services (hereinafter simply referred to as intervention services) that can intervene with the user and activate the user's activity. Location information 133 indicates a time series of the user's location. Activity amount information 134 indicates a time series of the user's activity amount. Behavioral purpose information 135 indicates keywords related to the behavioral purpose.

Map information 136 is an example of a POI (Point Of Interest) dictionary, and indicates the location of spots. Timeline information 137 indicates a time series of user behavior. Activity ratio information 138 indicates an activity ratio calculated by the activity ratio calculation unit 112. Exchange diversity information 139 indicates exchange diversity calculated by the exchange diversity calculation unit 113. Movement information 140 indicates movement calculated by the movement calculation unit 114.

The behavior evaluation information 141 indicates the behavior evaluation result of the user evaluated by the behavior evaluation unit 116. The care-needing risk rank determination information 142 is information for determining the care-needing risk rank from the care-needing risk. The total score rank determination information 143 is information for determining the total score rank from the total score.

User aggregated evaluation information 144 indicates statistics on behavioral evaluation results for all users. Behavioral tendency information 145 indicates the user's tendency to compare with others of the same sex and age regarding the user's risk of needing nursing care and activities (i.e., comparison tendency among a group of users with the same attributes) and self-tendencies. Reference parts information 146 indicates parts to be inserted into reference templates 147 for the user's tendency to compare with others of the same sex and age regarding the user's risk of needing nursing care and activities and self-tendencies.

The reference template 147 is a template for generating individual advice to be presented to the user. The inverse simulation result information 148 indicates the improvement values and amounts of activity ratio, interaction diversity, and exercise required to improve the user's nursing care risk rank or standard score rank. The intervention service usage history information 149 indicates the user's usage history of the intervention service. The learning data 150 is learning data for generating a nursing care risk prediction model.

The client 200 includes, for example, an information transmission/reception unit 211, which is a functional unit. The information transmission/reception unit 211 transmits and receives information to and from the evaluation server 100.

In this embodiment, the behavior evaluation process and the advice generation process are executed based on at least one of the care need risk (absolute evaluation) and the standard score (relative evaluation). In other words, either the care need risk or the standard score need not be calculated (both may be calculated). When the care need risk is not calculated, the assessment server 100 does not need to include the care need risk prediction unit 115, and the information held by the assessment server 100 does not need to include items related to the care need risk. When the standard score is not calculated, the assessment server 100 does not need to include the standard score calculation unit 120, and the information held by the assessment server 100 does not need to include items related to the standard score and the total score.

Figure 2 is a block diagram showing an example of the configuration of the computers that make up the evaluation server 100 and the client 200. The computer 300 is composed of a computer having, for example, a CPU (Central Processing Unit) 101, a memory 102, an auxiliary storage device 103, an input device 104, an output device 105, and a communication device 106.

The CPU 101 includes a processor and executes programs stored in the memory 102. The memory 102 includes a ROM (Read Only Memory), which is a non-volatile storage element, and a RAM (Random Access Memory), which is a volatile storage element. The ROM stores immutable programs (e.g., a BIOS (Basic Input/Output System)). The RAM is a high-speed, volatile storage element such as a DRAM (Dynamic Random Access Memory), and temporarily stores programs executed by the CPU 101 and data used when the programs are executed.

The auxiliary storage device 103 is, for example, a large-capacity, non-volatile storage device such as a magnetic storage device (HDD (Hard Disk Drive)) or a flash memory (SSD (Solid State Drive)), and stores the programs executed by the CPU 101 and data used when the programs are executed. In other words, the programs are read from the auxiliary storage device 103, loaded into the memory 102, and executed by the CPU 101.

The input device 104 is a device, such as a keyboard or mouse, that receives input from an operator. The output device 105 is a device, such as a display device or printer, that outputs the results of program execution in a format that can be viewed by the operator.

The communication device 106 is a network interface device that controls communication with other devices according to a specific protocol. The communication device 106 also includes a serial interface such as a Universal Serial Bus (USB).

A part or all of the programs executed by the CPU 101 may be provided to the computer 300 via a network from a removable medium (CD-ROM, flash memory, etc.) that is a non-transitory storage medium, or from an external computer equipped with a non-transitory storage device, and stored in the non-volatile auxiliary storage device 103 that is a non-transitory storage medium. For this reason, the computer 300 may have an interface for reading data from removable media.

Each of the evaluation server 100 and the client 200 is a computer system configured on one physical computer or on multiple logically or physically configured computers, and may operate in separate threads on the same computer, or may operate on a virtual computer constructed on multiple physical computer resources.

Each functional unit shown in FIG. 1 is included in the CPU 101 of the computer 300 constituting the device having the functional unit. For example, the CPU 101 included in the computer 300 constituting the evaluation server 100 functions as a timeline tallying unit 111 by operating according to a timeline tallying program loaded into a memory 102 included in the computer 300, and functions as an activity ratio calculation unit 112 by operating according to an activity ratio calculation program loaded into the memory 102.

In addition, some or all of the functions of the functional units included in the CPU 101 of the computer 300 constituting each device included in the behavior evaluation system may be realized by hardware such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array).

The information held by each device shown in FIG. 1 is stored in the auxiliary storage device 103 of the computer 300 constituting the device. Note that some or all of the information stored in the auxiliary storage device 103 of the computer 300 constituting each device included in the behavioral evaluation system may be stored in the memory 102 of the computer 300, or may be stored in a database connected to the device and external to the behavioral evaluation system.

In this embodiment, the information used by the behavior evaluation system does not depend on the data structure and may be expressed in any data structure. In this embodiment, the information is expressed in table format, but the information can be stored in a data structure appropriately selected from, for example, a list, a database, or a queue.

Although not shown in the figure, the computer 300 constituting the client 200 is equipped with a sensor (e.g., GPS (Global Positioning System), etc.) that measures the user's location information, and a sensor that measures the user's activity level, such as the number of steps, heart rate, or calories burned.

Figure 3 is a diagram showing an example of the data configuration of account information 131. Account information 131 includes, for example, a user ID column 1311, a name column 1312, a date of birth column 1313, a gender column 1314, and an address column 1315. The user ID column 1311 holds a user ID that identifies a user. The name column 1312 holds information indicating the user's name.

The date of birth column 1313, the gender column 1314, and the address column 1315 each hold information indicating the user's date of birth, gender, and address, each of which is an example of the user's attribute information.

Figure 4 is a diagram showing an example of the data configuration of the intervention service information 132. The intervention service information 132 holds, for example, an intervention service ID column 1321, a name column 1322, a summary column 1323, a type column 1324, an event frequency column 1325, an event location column 1326, and a target amount of exercise column 1327.

The intervention service ID column 1321 holds an intervention service ID that identifies the intervention service. The name column 1322, summary column 1323, and type column 1324 hold information indicating the name, summary, and type of the intervention service, respectively.

The frequency of events column 1325 holds information indicating how often the intervention service is held, and the frequency of events indicates the expected improvement in activity ratio. The location of events column 1326 holds information indicating the location and number of locations where the intervention service is held, and the number of locations indicates the expected improvement in interaction diversity. The guideline amount of exercise column 1327 holds information indicating the guideline amount of exercise per session when the user participates in the intervention service, and the guideline amount of exercise indicates the expected improvement in exercise.

Figure 5 is a diagram showing an example of the data configuration of the location information 133. The location information 133 includes, for example, a user ID column 1331, a timestamp column 1332, a point column 1333, a stay/movement category column 1334, and a means of transportation column 1335. The user ID column 1331 holds a user ID. The timestamp column 1332 holds information indicating the time when the user was at the location. Note that the sampling interval of each user's location indicated by the timestamp column 1332 (i.e., the sampling interval at which each client 200 acquires location information) may be constant or may not be constant depending on the implementation of the client 200 (for example, location information may not be acquired when the battery of the client 200 is in power saving mode).

The points column 1333 holds information indicating the location where the user was at the time. The stay/movement category column 1334 holds a stay/movement category indicating whether the user was staying at the location at the time or moving around the location. The means of transportation column 1335 holds information indicating the means of transportation when the stay/movement category is "movement" (when the stay/movement category is "stay", "N/A" (Not/Applicable) is stored).

Figure 6 is a diagram showing an example of the data configuration of activity amount information 134. Activity amount information 134 includes, for example, a user ID column 1341, a start time column 1342, an end time column 1343, and a step count column 1344. The user ID column 1341 holds a user ID. The start time column 1342 and the end time column 1343 hold information indicating the start time and end time, respectively, of the measurement of the activity amount indicated by the record.

The step count column 1344 holds information indicating the number of steps taken in the action indicated by the record. Note that the number of steps is an example of an activity amount, and the activity amount information 134 may store information indicating another activity amount (e.g., heart rate or calories burned) instead of or in addition to the number of steps.

Figure 7 is a diagram showing an example of the data configuration of the action purpose information 135. The action purpose information 135 includes, for example, an action purpose column 1351 and a keyword column 1352. The action purpose column 1351 holds information indicating the action purpose. The keyword column 1352 holds information indicating keywords related to the action purpose.

Figure 8 is a diagram showing an example of the data configuration of map information 136. Map information 136 includes, for example, a spot ID column 1361, a spot type column 1362, a spot name column 1363, and a coordinate column 1364. The spot ID column 1361 holds a spot ID that identifies a spot. The spot type column 1362 holds information indicating the type of spot. The spot name column 1363 holds information indicating the name of the spot. The coordinate column 1364 holds information indicating the coordinates of a representative position of the spot.

Figure 9 is a diagram showing an example of the data configuration of the timeline information 137. The timeline information 137 includes, for example, a user ID column 1371, a start time column 1372, an end time column 1373, a coordinate column 1374, a spot type column 1375, a spot name column 1376, an activity purpose column 1377, a stay/travel category column 1378, a means of transportation column 1379, and a number of steps column 13710.

The user ID column 1371 holds a user ID. The start time column 1372 and end time column 1373 hold information indicating the start time and end time, respectively, of the activity indicated by the record. The coordinate column 1374, spot type column 1375, spot name column 1376, and activity purpose column 1377 hold information indicating the coordinates where the user stayed, the type of spot, the name of the spot, and the activity purpose of the spot, respectively, when the stay/movement category is "stay" (however, when the stay/movement category is "movement", "N/A" is stored in these columns).

The stay/movement category column 1378 holds the stay/movement category. The means of transportation column 1379 holds information indicating the means of transportation when the stay/movement category is "movement" (when the stay/movement category is "stay", "N/A" is stored). The number of steps column 13710 holds information indicating the number of steps in the behavior indicated by the record. Note that the timeline information 137 may store information indicating other amounts of activity (e.g., heart rate or calories burned) instead of or in addition to the number of steps.

The user's behavior at a spot indicated by a record in timeline information 137 in which the stay/movement category is "stay" is included in the activity. On the other hand, the user's behavior (movement by means of transportation) indicated by a record in timeline information 137 in which the stay/movement category is "movement" may or may not be included in the activity (hereinafter, movement is not considered to be included in the activity).

Figure 10 is a diagram showing an example of the data configuration of activity ratio information 138. Activity ratio information 138 includes, for example, a user ID column 1381, a counting year/month/week column 1382, and a number of times going out column 1383. The user ID column 1381 holds a user ID. The counting year/month/week column 1382 holds information indicating the week for which the number of times going out is counted. Note that in the example of Figure 10, the number of times going out for each user is counted on a weekly basis, but the time unit of counting is not limited to weeks. This also applies to the time unit of counting for other information (however, it is desirable that the time unit of counting be consistent for all information). The number of times going out column 1383 holds information indicating the number of times the user went out in that week.

FIG. 11 is a diagram showing an example of the data configuration of the interaction diversity information 139. The interaction diversity information 139 includes, for example, a user ID column 1391, a collection year/month/week column 1392, a number of places visited column 1393, an athletic hobby column 1394, a humanities hobby column 1395, an employment column 1396, a volunteer column 1397, and an other column 1398.

The user ID column 1391 holds a user ID. The counting year/month/week column 1392 holds information indicating the week for which the interaction diversity is counted. The number of locations visited column 1393 holds information indicating the total number of locations visited by the user for the user's activities in that week. The athletic hobbies column 1394, liberal arts hobbies column 1395, work column 1396, volunteer column 1397, and other column 1398 hold the number of locations visited by the user for the user's activities in that week for each activity purpose.

FIG. 12 is a diagram showing an example of the data configuration of exercise information 140. Exercise information 140 includes, for example, a user ID column 1401, a counting year/month/week column 1402, and an average number of steps column 1403. The user ID column 1401 holds a user ID. The counting year/month/week column 1402 holds information indicating the week for which interaction diversity is counted. The average number of steps column 1403 holds information indicating the average number of steps per day for the user's activity in that week.

FIG. 13 is a diagram showing an example of the data configuration of the behavior evaluation information 141. The behavior evaluation information 141 holds, for example, an aggregation year/month/week column 1411, a user ID column 1412, a care-needing risk column 1413, a care-needing risk rank column 1414, a number of outings deviation value column 1415, a number of places stayed deviation value column 1416, an average number of steps deviation value column 1417, an overall score column 1418, and an overall score rank column 1419.

The aggregation year/month/week column 1411 holds information indicating the week for which behavioral evaluations are aggregated. The user ID column 1412 holds a user ID. The care risk column 1413 holds information indicating the care risk of the user. The care risk rank column 1414 holds information indicating the rank (class) of the care risk of the user indicated by the record among all users.

The deviation value of the number of times going out column 1415, the deviation value of the number of places stayed column 1416, and the deviation value of the average number of steps column 1417 respectively hold information indicating the deviation value of the number of times the user went out in the week, the deviation value of the number of places stayed in the week, and the deviation value of the average number of steps (per day) in the week for the entire user population.

The total score column 1418 holds information indicating the total score. The total score is calculated, for example, from a standard deviation (for example, a weighted average of the standard deviations) of the number of times the user went out in the week, the number of places the user stayed, and the average number of steps taken (per day) for the week. The total score rank column 1419 holds information indicating the rank (class) of the total score of the user indicated by the record among all users.

Figure 14 is a diagram showing an example of the data configuration of the care need risk rank determination information 142. The care need risk rank determination information 142 includes, for example, a care need risk section column 1421 and a care need risk rank column 1422. The care need risk rank can be determined from the care need risk value based on the correspondence between the section indicated by the care need risk section column 1421 and the care need risk rank indicated by the care need risk rank column 1422.

Figure 15 is a diagram showing an example of the data configuration of the total score rank determination information 143. The total score rank determination information 143 includes, for example, a total score interval column 1431 and a total score rank column 1432. The total score rank can be determined from the total score value based on the correspondence between the interval indicated by the total score interval column 1431 and the total score rank indicated by the total score rank column 1432.

In the examples of Figures 14 and 15, the nursing care risk rank and the total score rank are defined as four ranks, A to D (A being the best rank and D being the worst rank), but the number of ranks is not limited to four (it is sufficient if there are multiple ranks). Furthermore, the thresholds (upper and lower limits) of each section for determining the rank are not limited to the examples of Figures 14 and 15.

Figure 16 is a diagram showing an example of the data configuration of user aggregated evaluation information 144. User aggregated evaluation information 144 includes, for example, an aggregated year/month/week column 1441, a gender column 1442, an age class column 1443, a number of times going out column 1444, a number of places stayed column 1445, an average number of steps column 1446, and a care-needing risk column 1447. The aggregated year/month/week column 1441 holds information indicating the week for which the user aggregated evaluation is to be aggregated. The gender column 1442 holds information indicating the gender. The age class column 1443 holds information indicating the age class.

In the example of FIG. 16, users are grouped by gender and age class, but other attribute information may be used to group users in addition to or in addition to gender and age class. Also, in the example of FIG. 16, age classes are grouped in 5-year increments, but they may be grouped in any age increment.

The number of times going out column 1444, the number of places stayed column 1445, the average number of steps column 1446, and the care need risk column 1447 each hold statistical information on the number of times going out, the number of places stayed, the average number of steps, and the care need risk of a user of the gender and age group in the relevant week (for example, statistical quantities such as the average, minimum, and maximum of the number of times going out, the number of places stayed, the average number of steps, and the care need risk of a user of the gender and age group in the relevant week, as well as statistical quantities for classifying into the first quartile, second quartile, third quartile, etc.).

Figure 17 is a diagram showing an example of the data configuration of behavioral tendency information 145. The behavioral tendency information 145 includes, for example, a user ID column 1451, a counting year/month/week column 1452, a care need risk column 1453, a number of times going out column 1454, a number of places stayed column 1455, and an average number of steps column 1456. The user ID column 1451 holds the user ID, and the counting year/month/week column 1452 holds information indicating the week for which the behavioral tendency is being counted.

The nursing care risk column 1453, the number of outings column 1454, the number of places stayed column 1455, and the average number of steps column 1456 each hold information showing the comparative tendency of the user's nursing care risk, number of outings, number of places stayed, and average number of steps for the week in question with those of the same sex and age (same age class), and the user's own tendency (for example, the comparative tendency with the week prior to the week in question).

FIG. 18 is a diagram showing an example of the data configuration of the reference parts information 146. The reference parts information 146 includes, for example, a trend column 1461 and a parts column 1462. The trend column 1461 holds columns showing whether the same-sex and same-age tendencies of the need for care risk, number of outings, number of places stayed (total), and average number of steps are good or bad (A is the best risk of needing care and D is the worst, and the higher the number of outings, number of places stayed (total), and average number of steps, the better), as well as columns showing whether the self-term tendencies of the need for care risk, number of outings, number of places stayed (total), and average number of steps are improving or worsening.

The parts column 1462 holds information indicating parts to be inserted into the text of the reference template 147 for the risk of needing nursing care, number of times going out, number of places stayed (total), and average number of steps by same-sex and same-age tendencies and by personal tendencies.

Figure 19 is an example of a reference template 147. The reference template 147 shows the text to be displayed as individual advice. Individual advice is generated by inserting parts extracted from the reference part information 146 into each of (1) to (8) of the reference template 147.

FIG. 20 is a diagram showing an example of the data configuration of the inverse simulation result information 148. The inverse simulation result information 148 includes, for example, a user ID column 1481, a counting year/month/week column 1482, an improvement value of the number of outings column 1483, an improvement range of the number of outings column 1484, an improvement value of the number of places of stay column 1485, an improvement range of the number of places of stay column 1486, an improvement value of the average number of steps column 1487, and an improvement range of the average number of steps column 1488.

The user ID column 1481 holds a user ID. The aggregation year/month/week column 1482 holds information indicating the week for which the inverse simulation is aggregated. The improvement value of the number of outings column 1483, the improvement range of the number of outings column 1484, the improvement value of the number of places stayed column 1485, the improvement range of the number of places stayed column 1486, the improvement value of the average number of steps column 1487, and the improvement range of the average number of steps column 1488 hold information indicating the improvement value and improvement range of the number of outings, number of places stayed, and average number of steps for the user to improve their care risk rank for the week.

Figure 21 is a diagram showing an example of the data configuration of the intervention service usage history information 149. The intervention service usage history information 149 includes, for example, a user ID column 1491 and an intervention service reference ID column 1492. The user ID column 1491 holds a user ID. The intervention service reference ID column 1492 holds a flag indicating the usage history of the user of the intervention service indicated by each intervention service ID.

For example, if a user has used a certain intervention service, a "1" is stored in the column corresponding to the intervention service ID of the user; if a certain intervention service has been recommended to the user but has not been used, a "0" is stored in the column corresponding to the intervention service ID of the user; and if a certain intervention service has not been recommended to the user and has not been used, a "-" is stored in the column corresponding to the intervention service ID of the user. Note that the intervention service reference ID column 1492 may store the number of times the user has used the intervention service.

Fig. 22 is a diagram showing an example of the data configuration of the learning data 150. The learning data 150 includes, for example, a user ID field 1501, an age field 1502, a gender field 1503, a residential area field 1504, a number of times going out field 1505, a number of places stayed field 1506, an athletic hobby field 1507, a cultural hobby field 1508, an employment field 1509, a volunteer field 15010, an other field 15011, an average number of steps field 15012, and a field showing whether or not the user has been certified as needing nursing care within the past T years 15013.

The user ID column 1501 holds the user ID of the user for training the nursing care risk prediction model. The age column 1502, gender column 1503, and residential area column 1504 hold information indicating the attributes of the user (age, gender, and residential area) in the reference week of nursing care risk prediction model training. The number of times going out column 1505, number of places stayed column 1506, and average number of steps column 15012 hold information indicating the number of times the user went out (activity ratio), number of places stayed (interaction diversity), and average number of steps (amount of exercise) in the reference week of nursing care risk prediction model training, respectively.

The athletic hobby column 1507, cultural hobby column 1508, work column 1509, volunteer column 15010, and other column 15011 hold the number of locations the user stayed at for their activities in the reference week of nursing care risk prediction model learning, for each purpose of activity. The nursing care certification within T years column 15013 holds information indicating whether the user was certified as needing nursing care within T years (e.g., 3 years) from the reference week of nursing care risk prediction model learning.

In the example of FIG. 22, the information indicated in the age column 1502, gender column 1503, residential area column 1504, number of times going out column 1505, number of places stayed column 1506, athletic hobbies column 1507, cultural hobbies column 1508, employment column 1509, volunteer column 15010, other column 15011, and average number of steps column 15012 are values substituted for independent variables in the risk prediction model, and the information indicated in the column for whether or not nursing care is required within T years 15013 is a value substituted for dependent variables in the risk prediction model.

The evaluation process by the evaluation server 100 will be described below. Before the evaluation process starts, the following information is registered in advance: account information 131, intervention service information 132, location information 133, activity amount information 134, behavioral purpose information 135, map information 136, nursing care risk rank determination information 142, total score rank determination information 143, reference part information 146, and reference template 147.

For example, the account information 131 is registered according to input from the user to each client 200, and the location information 133 and activity amount information 134 are registered according to information measured by each client 200. Also, for example, the intervention service information 132, map information 136, nursing care risk rank assessment information 142, total score rank assessment information 143, reference part information 146, and reference template 147 are registered based on input from the administrator of the evaluation server 100, for example.

Figure 23 is a flowchart showing an example of the evaluation process. The process described below is executed, for example, for each user registered in the account information 131. The timeline calculation unit 111 executes the timeline calculation process (S100).

The activity proportion calculation unit 112 executes an activity proportion calculation process (S200). The exchange diversity calculation unit 113 executes an exchange diversity calculation process (S300). The motion calculation unit 114 executes a motion calculation process (S400). Note that the order in which the activity proportion calculation process of step S200, the exchange diversity calculation process of step S300, and the motion calculation process of step S400 are executed does not matter, and they may be executed in parallel.

The behavior evaluation unit 116 executes behavior evaluation processing (S500). The advice generation unit 117 executes advice generation processing (S600). The recommendation processing unit 118 executes recommendation processing (S700), and the evaluation processing ends. Note that the recommendation processing unit 118 may receive a recommendation request from the client 200 and execute recommendation processing only for the user who uses the client 200. Note that details of each process described in FIG. 23 will be described later using FIG. 24 to FIG. 30.

FIG. 24 is a flowchart showing an example of the timeline tallying process in step S100. The timeline tallying unit 111 refers to the location information 133 to identify the time at which the stay/movement segment of the target user of the timeline tallying process changes. The timeline tallying unit 111 treats a time period in which the target user's stay/movement segment has consecutive "stays" as one stay segment, and a time period in which the target user's stay/movement segment has consecutive "movements" as one movement segment. The timeline tallying unit 111 determines the identified time as the end time of the stay segment or movement segment that ends at the identified time, and the start time of the stay segment or movement segment that starts from the identified time (S101). In other words, the time identified in step S101 is the time at which the movement segment and the stay segment switch.

The timeline aggregation unit 111 refers to the location information 133 to identify a point in the target user's section of stay (e.g., the average position of all points in the section of stay, or a point at the middle time of the section of stay, etc.), and refers to the map information 136 to extract the spot type and spot name of the spot having the coordinates closest to the identified point (S102). Note that the timeline aggregation unit 111 may receive the spot type and spot name from the client 200 (e.g., the client 200 accepts input from the user), in which case the processing of step S102 is omitted.

The timeline aggregation unit 111 extracts a behavior purpose from the behavior purpose information 135 using the extracted spot type and spot name as a search key (S103). Specifically, the timeline aggregation unit 111 identifies a behavior purpose in the behavior purpose information 135 that corresponds to the keyword included in the extracted spot type and spot name.

The timeline counting unit 111 acquires the number of steps for each stay segment and each movement segment of the target user from the activity amount information 134 (S104). The timeline counting unit 111 registers the start time and end time identified in step S101, the spot type and spot name identified in step S102, the purpose of the action identified in step S103, the number of steps identified in step S104, the point (coordinates) of the spot, the stay movement/category ("stay" for a stay segment, "movement" for a movement segment), and the means of transportation indicated by the location information 133 for each stay segment and movement segment identified in step S101 in the timeline information 137, and ends the timeline counting process.

In addition, each client 200 may generate a timeline for each user. In this case, the evaluation server 100 registers the timeline of each user received from each client 200 in the timeline information 137, and the timeline aggregation process of step S100 is omitted.

Figure 25 is a flowchart showing an example of the activity ratio calculation process in step S200. The activity ratio calculation unit 112 sets the aggregation year, month, and week as the aggregation period (S201). The activity ratio calculation unit 112 refers to the timeline information 137 and counts the number of times the target user of the activity ratio calculation process goes out during the aggregation period (S202).

Specifically, for example, the activity ratio calculation unit 112 regards the target user's departure from home as going out. Note that the activity ratio calculation unit 112, for example, refers to the location information 133 and determines the most frequent place of stay of the target user during a specified time period (e.g., midnight to 4 am) within a specified period (e.g., within the last month) as the home coordinates of the target user. In addition, the client 200 may transmit the home coordinates input by the target user to the evaluation server 100, so that the activity ratio calculation unit 112 obtains the home coordinates of the target user.

In addition, when the target user goes out multiple times on the same day, it is preferable for the activity ratio calculation unit 112 to count each outing as one in order to prevent the user from staying confined to their home. On the other hand, the activity ratio calculation unit 112 may count the total number of times the target user goes out (i.e., if the target user goes out multiple times on the same day, the multiple outings are counted).

The activity ratio calculation unit 112 registers the number of times the target user went out for the aggregated year, month, and week counted in step S202 in the activity ratio information 138 (S203), and ends the activity ratio calculation process.

Figure 26 is a flowchart showing an example of the interaction diversity calculation process in step S300. The interaction diversity calculation unit 113 sets the aggregation year, month, and week as the aggregation period (S301). The interaction diversity calculation unit 113 refers to the timeline information 137 and counts the number of spots (total number of places of stay) where the target user of the interaction diversity calculation process stayed during the aggregation period, and the number of each behavioral purpose corresponding to the spots (number of places of stay for each behavioral purpose) (S302).

Note that, if the target user stays at the same spot multiple times during the collection period, it is desirable for the interaction diversity calculation unit 113 to count it as one stay location (because it is highly likely that the "diversity" of interactions does not improve whether the user stays there once or multiple times). Furthermore, the interaction diversity calculation unit 113 may count the target user's stay at the same spot multiple times during the collection period as multiple stay locations.

The interaction diversity calculation unit 113 registers the total number of places of stay for the target user during the aggregation year, month, and week counted in step S303 and the number of places of stay for each behavioral purpose in the interaction diversity information 139 (S303), and terminates the interaction diversity calculation process.

Figure 27 is a flowchart showing an example of the exercise calculation process in step S400. The exercise calculation unit 114 sets the aggregation year, month, and week as the aggregation period (S401). The exercise calculation unit 114 refers to the timeline information 137, counts the number of steps of the target user of the exercise calculation process during the aggregation period, and calculates the average number of steps per day for the target user during the aggregation period (S402).

In addition, in step S402, the exercise calculation unit 114 may count only the number of steps for which the stay/movement category of the target user in the collection period is "stay", or may also count the number of steps for which the stay/movement category is "movement" (i.e., exercise during at least a "stay" is counted).

The exercise calculation unit 114 registers the average number of steps per day for the target user during the collection period calculated in step S402 in the exercise information 140 (S403), and ends the exercise calculation process.

Figure 28A is a flowchart showing an example of the behavior evaluation process in step S500. The behavior evaluation in Figure 28A uses the care-needing risk. The care-needing risk prediction unit 115 calculates the care-needing risk of the target user of the behavior evaluation process using a care-needing risk prediction model (S501).

The nursing care risk prediction model is stored in advance, for example, in the memory 102 or the auxiliary storage device 103 of the evaluation server 100. The nursing care risk y is expressed, for example, as y = f (user attribute, number of outings, number of places stayed, average number of steps) (f is the nursing care risk prediction model). Examples of user attributes included in the independent variables of f include age, gender, and/or residential area. Also, examples of the number of places stayed included in the independent variables of f include the number of places stayed for each purpose of activity. For example, the nursing care risk is higher with age, higher for women than for men, and tends to be lower with the number of outings, number of places stayed, and average number of steps.

The nursing care risk prediction model may be expressed by a multiple regression model. For example, the nursing care risk y is expressed as y = ( _b0 , _b1 , ..., _bn ) (age, sex, residential area, number of times going out, number of places to stay for exercise, number of places to stay for cultural and exercise, ..., average number of steps) ^T + ε.

The behavior evaluation unit 116 refers to the care risk rank determination information 142 to determine the care risk rank corresponding to the care risk of the target user, and registers the care risk and care risk rank of the target user for the aggregation period in the behavior evaluation information 141 (S502).

The behavior evaluation unit 116 calculates statistics (average, minimum, maximum, and quartiles) of each indicator (number of times going out, number of places stayed, and average number of steps) for the combination of age class and gender to which the target user belongs during the target period, registers the calculated statistics in the user aggregated evaluation information 144 (S503), and terminates the behavior evaluation process in which the risk of needing care is used.

Figure 28B is a flowchart showing an example of the behavior evaluation process in step S500. Standard scores are used for the behavior evaluation in Figure 28B. The standard score calculation unit 120 calculates the standard score (S511).

Specifically, for example, the standard score calculation unit 120 calculates the standard scores for the activity ratio, interaction diversity, and exercise, respectively, based on the standard scores for the number of times the target user went out, the number of places they stayed, and the average number of steps taken during the target period when the entire user population is taken as the total population. Note that the standard score calculation unit 120 may use other quantities (e.g., z-scores) that indicate the position in the overall distribution instead of the standard scores when calculating the standard scores.

The standard score calculation unit 120 calculates the total score (S512). Specifically, for example, the standard score calculation unit 120 calculates the weighted average (e.g., weights are given in advance) of the standard scores (deviation values) calculated in step S511 as the total score.

The behavior evaluation unit 116 refers to the overall score rank determination information 143 to determine the overall score rank corresponding to the overall score of the target user, and registers the standard scores, overall word count, and overall score rank of the target user for the aggregation period in the behavior evaluation information 141 (S513).

The behavior evaluation unit 116 calculates statistics (average, minimum, maximum, and quartiles) for each indicator (number of times going out, number of places stayed, and average number of steps) for the combination of age class and gender to which the target user belongs during the target period, registers the calculated statistics in the user aggregated evaluation information 144 (S514), and terminates the behavior evaluation process in which the standard score is used.

Note that both the behavior evaluation processes in FIG. 28A and FIG. 28B may be executed, or only one of the behavior evaluation processes in FIG. 28A or FIG. 28B may be executed. When both the behavior evaluation processes in FIG. 28A and FIG. 28B are executed, either step S503 or step S514 is omitted.

Figures 29A, 29B, and 29C are flowcharts showing an example of advice generation processing. Figure 29B is a continuation of the processing in Figure 29A. Since the risk of needing care is used to generate advice in Figure 29B, the processing in Figure 28A must be executed in order for the processing in Figure 29B to be executed. Figure 29C is a continuation of the processing in Figure 29A. Since the standard score is used to generate advice in Figure 29C, the processing in Figure 28B must be executed in order for the processing in Figure 29C to be executed. In addition, both the behavior evaluation processing in Figure 29A and Figure 29B may be executed, or the behavior evaluation processing in either Figure 29A or Figure 29B may be executed.

The advice generation unit 117 calculates the behavioral tendencies of the target user of the advice generation process (comparison tendencies with people of the same sex and age and self-tendencies for each of the risk of needing care, number of times going out, number of places stayed, and average number of steps) and registers the behavioral tendencies in the behavioral tendency information 145 (S601).

For example, in step S601, the advice generation unit 117 determines whether the risk of needing care, the number of times going out, the number of places stayed, and the average number of steps for the target user in the aggregation period indicated by the behavior evaluation information 141 are equal to or greater than the risk of needing care, the number of times going out, the number of places stayed, and the average number of steps for the same sex and age (same age group) of the target user in the aggregation period indicated by the user aggregation evaluation information 144.

Specifically, for example, if the advice generation unit 117 determines that the risk of needing care of the target user for the aggregation period indicated by the behavior evaluation information 141 is equal to or higher than the risk of needing care of the target user of the same sex and age (same age class) for the aggregation period indicated by the user aggregate evaluation information 144, it determines that the target user's tendency to compare the risk of needing care with those of the same sex and age is "bad", and if it determines that this is not the case, it determines that the target user's tendency to compare the risk of needing care with those of the same sex and age is "good".

For example, when the advice generating unit 117 determines that the number of times the target user goes out during the aggregation period indicated by the behavior evaluation information 141 is equal to or greater than the number of times the target user goes out during the aggregation period of the same sex and age (same age class) indicated by the user aggregation evaluation information 144, it determines that the target user's tendency to compare the number of times he or she goes out with others of the same sex and age is "good", and when it determines that this is not the case, it determines that the target user's tendency to compare the number of times he or she goes out with others of the same sex and age is "bad". The method of calculating the tendency to compare the number of places stayed and the average number of steps with others of the same sex and age is the same as the method of calculating the tendency to compare the number of times he or she goes out with others of the same sex and age.

In addition, instead of comparing the target user with the average value of the same sex and age, the advice generation unit 117 may use each deviation value of the target user for the aggregation period indicated by the behavior evaluation information 141 to determine the tendency to compare with the same sex and age.

For example, if the advice generation unit 117 determines that the target user's risk of needing care for the current aggregation period indicated by the behavior evaluation information 141 is equal to or higher than the target user's risk of needing care for the previous aggregation period, it determines that the target user's personal tendency of needing care risk is "worsening," and if it determines that this is not the case, it determines that the target user's personal tendency of needing care risk is "improving."

For example, if the advice generation unit 117 determines that the number of times the target user went out during the aggregation period indicated by the behavior evaluation information 141 is equal to or greater than the number of times the target user went out during the previous aggregation period, it determines that the target user's self-tendency for the number of times they went out is "improved," and if it determines that this is not the case, it determines that the target user's self-tendency for the number of times they went out is "deteriorated." The method of calculating the self-tendency for the number of places stayed and the average number of steps is the same as the method of calculating the self-tendency for the number of times they went out.

In addition, instead of comparing the indicators of the target user's current collection period with those of the previous collection period, the advice generation unit 117 may use the slope of the regression coefficient of the indicators for a specified period including the current collection period to determine whether the self-tendency has improved or worsened.

The advice generating unit 117 extracts parts from the reference parts information 146 using the behavioral tendency of the target user calculated in step S601 as a key (S602). Specifically, in the example of the reference parts information 146 in FIG. 18, if the target user's tendency to compare the risk of needing nursing care with those of the same sex and age is "good", "relatively active" is extracted as part (1), and "can lower the risk of needing nursing care" is extracted as part (2). In this way, the advice generating unit 117 extracts parts from the reference parts information 146 that correspond to the same sex and age tendency and the self tendency of the target user's risk of needing nursing care, number of times going out, number of places stayed, and average number of steps.

The advice generation unit 117 generates an individual advice text by inserting each of the extracted parts into the corresponding position in the text of the reference template 147 (S603). Note that if a care-needing risk prediction has not been performed (if the behavior evaluation process in FIG. 28A has not been performed), the process regarding the care-needing risk is omitted in steps S601 to S603.

The processing of FIG. 29B will be described. Following step S603, the advice generating unit 117 calculates the difference between the upper limit value of the care risk section that is one rank higher than the care risk rank to which the target user belongs and the current care risk value of the target user (S611).

Specifically, for example, according to the care need risk rank determination information 142 in FIG. 14, if the care need risk of the target user is 30, the care need risk rank is B, and the upper limit of the care need risk section for rank A, which is one rank better than care need risk rank B, is 20, so the difference calculated in step S611 is 30-20=10. Note that if the care need risk rank of the target user is A (the best rank), the processing in FIG. 29B may be omitted.

The care need risk prediction unit 115 executes the following steps S612, S613, and S614 for each indicator of the target user (number of times going out, number of places stayed, and average number of steps).

The nursing care risk prediction unit 115 sets the current values of the indicators other than the target indicator in the loop to the nursing care risk prediction model, and performs an inverse simulation to calculate the improvement value (improved value) of the target indicator and the improvement range (from the current value) required to obtain a nursing care risk rank that is one level better (S612).

Specifically, for example, if the target indicator in the loop is the number of times going out, and the risk of needing care is y = f (user attributes, number of times going out, number of places stayed, average number of steps), the care risk prediction unit 115 sets the current values of the user attributes, number of places stayed, and average number of steps, which are indicators other than the target indicator in the loop, in the care risk prediction model, and identifies the improvement value and improvement range of the number of times going out to obtain a care risk that is one rank better than the current care risk rank.

The care-needing risk prediction unit 115 rounds up the improvement value and improvement range according to the data type of the target index within the range of the domain of the target index (S613). For example, if the current value of the target index, the number of times going out, is 3, the improvement value calculated in step S613 is 4.8 times (i.e., the improvement range is 4.8-3=1.8 times), the domain of the number of times going out is 0 to 7 inclusive, and the data type of the number of times going out is an integer, the care-needing risk prediction unit 115 rounds up the improvement value to 5 times (i.e., the improvement range is 2 times). Note that the domain and data type of each index (number of times going out, number of places stayed, and average number of steps) are predetermined.

The nursing care risk prediction unit 115 stores the improvement value (rounded up if necessary) and improvement range of the target index for the target user in the aggregation period in the inverse simulation result information 148 (S614).

If the improvement value of the index calculated in step S614 exceeds the domain, the advice display screen described below may display, for example, the upper limit of the domain instead of the improvement value, or a message such as "The rank will not change, but if you make improvements, you can reduce the risk of needing care. Let's start with what you can do."

Note that while the care need risk prediction unit 115 performs an inverse simulation for each target indicator in step S613, it may also perform an inverse simulation by combining multiple target indicators (e.g., by mathematical programming).

The process of FIG. 29C will now be described. Following step S603, the advice generating unit 117 calculates the difference between the lower limit of the total score interval that is one rank higher than the total score rank to which the target user belongs and the current value of the total score of the target user (S621).

Specifically, for example, according to the total score rank determination information 143 in FIG. 15, if the target user's total score rank is 55, then the total score rank is B, and the lower limit of the total score interval for rank A, which is one rank higher than total score rank B, is 60, so the difference calculated in step S621 is 60-55=5. Note that if the target user's total score rank is A (the best rank), the process in FIG. 29C may be omitted.

The standard score calculation unit 120 executes the processes of steps S622, S623, and S624 below for each indicator of the target user (number of times going out, number of places stayed, and average number of steps).

The standard score calculation unit 120 sets the current values of the indicators other than the target indicator in the loop to the calculation formula for the overall score, and performs an inverse simulation to calculate the improvement value (improved value) of the target indicator and the improvement range (from the current value) required to obtain a one-level better overall score rank (S622).

Specifically, for example, if the target indicator in the loop is the number of times going out, and the total score is given as a weighted average of the number of times going out, the number of places stayed, and the average number of steps, the standard score calculation unit 120 sets the current values of the number of places stayed and the average number of steps, which are indicators other than the target indicator in the loop, in a calculation formula for the weighted average, calculates the standard deviation of the number of times going out to obtain a care risk rank that is one level better than the current care risk rank, and identifies the improvement value and improvement range of the number of times going out that is required to obtain the calculated standard deviation.

The standard score calculation unit 120 rounds up the improvement value and improvement range according to the data type of the target index within the range of the domain of the target index (S623). The standard score calculation unit 120 stores the improvement value and improvement range (rounded up as necessary) of the target index for the target user in the aggregation period in the inverse simulation result information 148 (S624).

If the improvement value of the index calculated in step S624 exceeds the domain, the advice display screen described below may display, for example, the upper limit of the domain instead of the improvement value, or a message such as "The rank will not change, but if you make improvements, you can reduce the risk of needing care. Let's start with what you can do."

Note that while the standard score calculation unit 120 performs an inverse simulation for each target indicator in step S622, it may also perform an inverse simulation by combining multiple target indicators (e.g., by using mathematical programming).

Figure 30 is a flowchart showing an example of the recommendation process of step S700. The recommendation processing unit 118 identifies the behavioral performance estimate of the target user (usage history of each intervention service) by matching the venue indicated by the intervention service information 132 with the location of the target user of the recommendation process indicated by the location information 133, and stores the intervention service recommendation history for the target user indicated by the intervention service usage history information 149 in the identified behavioral performance estimate (S701).

In addition, in step S701, the client 200 may transmit the usage record value of the intervention service input by the target user to the evaluation server 100, and the recommendation processing unit 118 may store the received usage record value of the intervention service in the intervention service usage history information 149.

The recommendation processing unit 118 receives information indicating the improvement index selected by the user from the client 200, identifies an intervention service of a social participation type related to the selected improvement index from the intervention service information 132 (for example, by referring to the event frequency column 1325, the event location column 1326, and the target exercise amount column 1327), and filters the identified intervention service, for example, by collaborative filtering (filtering that identifies intervention services used by users with a similar intervention service usage history to the target user and that have not been used by the target user) (S702).

The recommendation processing unit 118 refers to the intervention service usage history information 149 from the intervention services filtered in step S702 to identify an intervention service that has not been recommended to the target user in the past, and determines the identified intervention service as a recommended candidate intervention service (S703).

The recommendation processing unit 118 refers to the account information 131 and filters the intervention services of the recommendation candidates by user attributes (e.g., age group, gender, and residential area) (identifying, from among the intervention services of the recommendation candidates, intervention services used by users whose user attributes are similar to those of the target user) (S704).

The recommendation processing unit 118 extracts the expected improvement value of the improvement index due to the intervention service extracted by the filtering in step S704 from the intervention service information 132 (S705), and ends the recommendation processing.

Figure 31 is a flowchart showing an example of a nursing care need risk prediction model update process. Before the nursing care need risk prediction model update process starts, the learning data 150 is registered in advance. The nursing care need risk prediction model update process may be executed periodically, may be executed according to instructions from a user of the evaluation server 100, or may be executed when there is a high possibility that a divergence has occurred between the data trends at the time of learning and the time of inference, such as every time the number of records in the learning data 150 increases by a specified number.

The values of the independent variables of the learning data 150 may be accumulated whenever the account information 131 is updated, the activity ratio calculation process, the exchange diversity calculation process, and the movement information calculation process are executed. The value of the dependent variable of the learning data 150, the nursing care certification within T years column 15013, is, for example, "yes" if the evaluation server 100 receives a notification from the client 200 indicating that a nursing care certification has been issued within T years, and is "no" if the evaluation server 100 does not receive such a notification within T years.

The care-needing risk prediction unit 115 re-learns the care-needing risk prediction model using the latest learning data 150 (S801). Specifically, for example, the care-needing risk prediction unit 115 learns the regression coefficients and error terms of the care-needing risk prediction model using the values of the independent variables and the dependent variables in the latest learning data 150. The care-needing risk prediction unit 115 deploys the re-learned care-needing risk prediction model and terminates the care-needing risk prediction model update process.

FIGS. 32A, 32B, and 32C are diagrams showing an example of an evaluation result display screen. FIG. 33 is a diagram showing an example of a recommendation result display screen. The evaluation result display screen and the recommendation result display screen may be displayed, for example, on the output device 105 of the evaluation server 100, or may be displayed on the output device 105 of the client 200 based on information sent by the evaluation server 100 to the client 200.

The evaluation result display screen includes, for example, an activity status display area 1001 for the period to be collected, a history of the number of outings displayed 1002, a history of the number of places stayed displayed 1003, a history of the average number of steps displayed 1004, an individual advice text display area 1005, an improvement index display area 1006, and a recommendation button 1007.

The activity status display area 1001 for the period to be counted displays, for example, the target user's care risk rank or total score rank ("You Recently B" in FIG. 32A) indicated by the behavioral assessment information 141, the number of times the target user went out indicated by the activity ratio information 138, the total number of places the target user stayed indicated by the interaction diversity information 139, and the average number of steps taken by the target user indicated by the exercise information 140, during the period to be counted (for example, the most recent year, month, and week).

In addition, the activity status display area 1001 for the period to be tabulated displays a radar chart that compares, for example, the number of times going out indicated by the activity ratio information 138, the total number of places stayed indicated by the interaction diversity information 139, and the average number of steps indicated by the exercise information 140 with the average number of times going out, the total number of places stayed, and the average number of steps of the target user of the same sex and age indicated by the user tabulated evaluation information 144. Note that the apex of each triangle that constitutes the radar chart indicates the quartile of the average number of times going out, the total number of places stayed, and the average number of steps of the target user of the same sex and age indicated by the user tabulated evaluation information 144.

The outing frequency history display area 1002 displays, for example, the history of the number of times the target user went out in each of the current tally period and multiple past tally periods, as indicated by the activity ratio information 138. The stay location number history display area 1003 displays, for example, the total number of stay locations and the number of stay locations for each behavioral purpose, as indicated by the interaction diversity information 139, as indicated by the current tally period and multiple past tally periods. The average step count history display area 1004 displays, for example, the target user's average step count history in each of the current tally period and multiple past tally periods, as indicated by the exercise information 140.

The individual advice text generated in step S603 is displayed in the individual advice text display area 1005. The improvement index display area 1006 displays the improvement values and improvement ranges for the number of times going out, the number of places stayed, and the average number of steps indicated by the inverse simulation result information 148. When the recommendation button 1007 is selected, the display transitions to a recommendation result display screen 1008.

The recommendation result display screen 1008 displays information indicating the intervention service extracted in step S704 and information indicating the expected improvement index due to the intervention service extracted in step S705.

As described above, the behavior assessment system of this embodiment calculates an activity ratio and interaction diversity indicating the user's degree of social participation, as well as an index indicating the user's exercise, based on the user's location and activity level measured by the sensor possessed by the client 200, and calculates the risk of needing nursing care and/or the overall score based on these indices. In this way, the behavior assessment system can objectively calculate an index measuring the user's participation in social activities, without placing a large burden on the user, and calculate the risk of needing nursing care and/or the overall score based on the indices.

The behavioral evaluation system of this embodiment also generates advice text based on the user's level of social participation and tendencies (tendencies in comparison with others, such as tendencies with people of the same sex and age, and trends over time, such as self-tendencies). Furthermore, the behavioral evaluation system presents the user with improvement values and improvement ranges for indicators for improving the nursing care risk rank and/or total score rank. The behavioral evaluation system also recommends intervention services that contribute to actions to improve the user's nursing care risk and/or total score, based on the user's attributes and behavioral records. By the behavioral evaluation system performing these processes, the user can specifically understand preventive actions that are useful to him/herself, and behavioral changes in the user are expected.

In the above example, the evaluation server 100 calculates the risk of needing nursing care, as well as the standard score and the total score, using the activity ratio, interaction diversity, and exercise, but the risk of needing nursing care, as well as the standard score and the total score, may be calculated using at least one of the activity ratio, interaction diversity, and exercise.

The present invention is not limited to the above-described embodiments, but includes various modified examples. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and are not necessarily limited to those having all of the configurations described. It is also possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. It is also possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

The above configurations, functions, processing units, processing means, etc. may be realized in part or in whole in hardware, for example by designing them as integrated circuits. The above configurations, functions, etc. may be realized in software by a processor interpreting and executing a program that realizes each function. Information on the programs, tables, files, etc. that realize each function can be stored in a memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

In addition, the control lines and information lines shown are those considered necessary for the explanation, and not all control lines and information lines on the product are necessarily shown. In reality, it can be assumed that almost all components are interconnected.

100 Evaluation server, 101 CPU 101, 102 Memory, 103 Auxiliary storage device, 104 Input device, 105 Output device, 106 Communication device, 111 Timeline aggregation unit, 112 Activity ratio calculation unit, 113 Exchange diversity calculation unit, 114 Movement calculation unit, 115 Nursing care risk prediction unit, 116 Behavior evaluation unit, 117 Advice generation unit, 118 Recommendation processing unit, 119 Processing model update unit, 120 Standard score calculation unit, 132 Intervention service information, 133 Position information, 134 Activity amount information, 137 Timeline information, 138 Activity ratio information, 139 Exchange diversity information, 140 Movement information, 151 Behavior evaluation information, 142 Nursing care risk rank determination information, 143 Total score rank determination information, 145 Behavior tendency information, 146 Reference part information, 148 Inverse simulation result information, 149 Intervention service usage history information, 150 Learning data, 200 Client

Claims (11)

A behavior evaluation server,
A processor and a memory are included.
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The processor,
generating a time series of a behavior of the first user based on the time series of the location and the time series of the activity amount;
Calculating a first index indicating the number of times the first user has gone out based on the time series of the locations indicated by the generated time series of the behavior, and a second index indicating the number of places where the first user has stayed based on the time series of the locations indicated by the generated time series of the behavior ,
Calculating a third index indicating an amount of exercise including an average number of steps at a place where the first user stays based on the generated time series of the behavior;
The memory includes:
a care-needing risk prediction model that outputs a care-needing risk when a value including a user attribute, the first index, the second index, and the third index is input;
and a user attribute of the first user;
the processor calculates a care-needing risk of the first user as a fourth index that comprehensively evaluates the behavior of the first user based on the user attributes of the first user, the first index, the second index, the third index, and the care-needing risk prediction model;
The memory holds care-needing risk rank information indicating a correspondence between the care-needing risk and a care-needing risk rank,
The processor,
Identifying a care-needing risk rank from the care-needing risk of the first user by referring to the care-needing risk rank information;
a behavioral evaluation server that calculates an amount of improvement of each of the first index, the second index, and the third index, based on a care risk corresponding to a care risk rank in the care risk rank information that is a predetermined level better than the identified care risk rank, the first index, the second index, the third index, and the care risk prediction model, in order to improve the care risk rank of the first user to a care risk rank that is a predetermined level better . The behavior evaluation server according to claim 1 ,
The processor generates data for displaying the first index, the second index, and the third index, as well as an amount of improvement of each of the first index, the second index, and the third index. The behavior evaluation server according to claim 1 ,
The memory holds learning data indicating the user attributes, the first indicator, the second indicator, the third indicator, and whether or not a nursing care certification has been made within a predetermined period of time for each of a plurality of users including the first user,
The processor updates the nursing care need risk prediction model based on the learning data. A behavior evaluation server,
A processor and a memory are included.
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The processor,
generating a time series of a behavior of the first user based on the time series of the location and the time series of the activity amount;
Calculating a first index indicating the number of times the first user has gone out based on the time series of the locations indicated by the generated time series of the behavior, and a second index indicating the number of places where the first user has stayed based on the time series of the locations indicated by the generated time series of the behavior,
Calculating a third index indicating an amount of exercise including an average number of steps at a place where the first user stays based on the generated time series of the behavior;
The memory holds the first index, the second index, and the third index for each of a plurality of users including the first user;
The processor,
Calculating standard scores indicating a ranking among the plurality of users for each of the first index, the second index, and the third index of a first user included in the plurality of users;
calculating a total score of the first user by combining the standard scores of the first index, the second index, and the third index, as a fourth index for comprehensively evaluating the behavior of the first user;
the memory holds total score rank information indicating a correspondence between the total score and a total score rank;
The processor,
Identifying a total score rank from the total score of the first user by referring to the total score rank information;
A behavioral evaluation server that calculates an improvement amount of each of the first index, the second index, and the third index based on an overall score corresponding to an overall score rank in the overall score rank information that is a predetermined level better than the identified overall score rank, the first index, the second index, and the third index, in order to improve the overall score rank of the first user to an overall score rank that is a predetermined level better. The behavior evaluation server according to claim 4,
The processor generates data for displaying the first index, the second index, and the third index, as well as an amount of improvement of each of the first index, the second index, and the third index. A behavior evaluation server,
A processor and a memory are included.
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The processor,
generating a time series of a behavior of the first user based on the time series of the location and the time series of the activity amount;
Calculating a first index indicating the number of times the first user has gone out based on the time series of the locations indicated by the generated time series of the behavior, and a second index indicating the number of places where the first user has stayed based on the time series of the locations indicated by the generated time series of the behavior,
Calculating a third index indicating an amount of exercise including an average number of steps at a place where the first user stays based on the generated time series of the behavior;
The memory includes:
User attributes of a plurality of users including the first user;
Intervention service usage history information indicating an intervention service used by each of the plurality of users;
and intervention service information indicating an improvement amount of the first index, the second index, and/or the third index when each of the intervention services is used;
The processor,
determining an intervention service to be recommended to the first user from intervention services used by each of the plurality of users having the same user attribute as the first user, the intervention service usage history information indicating the intervention service usage history information;
a behavior evaluation server that refers to the intervention service information and extracts an improvement amount of the first index, the second index, and/or the third index when the determined intervention service is used. 7. The behavior evaluation server according to claim 1, 4, or 6,
The memory includes:
Holding user attributes, the first index, the second index, and the third index of each of a plurality of users including the first user;
The processor,
Calculating a self tendency indicating a tendency of a change in the time series of the first index, the second index, and the third index of the first user based on the generated time series of the behavior;
Calculating a same attribute comparison tendency indicating a tendency of each of the most recent first index, second index, and third index of the first user compared with a group of users having a common user attribute with the first user among the plurality of users;
A behavior evaluation server that generates advice to be presented to the first user based on the calculated self-tendencies and same-attribute comparison tendencies of each of the first user's first indicator, second indicator, and third indicator. 7. The behavior evaluation server according to claim 1, 4, or 6,
The time series of the first user's locations indicates whether the first user is moving or staying at each location;
The time series of the activity amount of the first user indicates a time series of the number of steps of the first user.
Behavioral assessment server. A behavior evaluation method using a behavior evaluation server, comprising:
The behavior evaluation server includes a processor and a memory,
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The behavior evaluation method includes:
The processor generates a time series of a behavior of the first user based on the time series of the positions and the time series of the activity amount;
Calculating a first index indicating the number of times the first user has gone out based on the time series of the locations indicated by the generated time series of the behavior, and a second index indicating the number of places where the first user has stayed based on the time series of the locations indicated by the generated time series of the behavior,
The processor calculates a third index indicating an amount of exercise including an average number of steps at a place where the first user stays based on the generated time series of the behavior;
The memory includes:
a care-needing risk prediction model that outputs a care-needing risk when a value including a user attribute, the first index, the second index, and the third index is input;
and a user attribute of the first user;
The behavior evaluation method includes the steps of: calculating a care-needing risk of the first user as a fourth index that comprehensively evaluates the behavior of the first user based on a user attribute of the first user, the first index, the second index, the third index, and the care-needing risk prediction model;
The memory holds care-needing risk rank information indicating a correspondence between the care-needing risk and a care-needing risk rank,
The behavior evaluation method includes:
The processor refers to the care need risk rank information to identify a care need risk rank from the care need risk of the first user;
A behavioral evaluation method in which the processor calculates an amount of improvement of each of the first index, the second index, and the third index, based on a care risk corresponding to a care risk rank in the care risk rank information that is a predetermined level better than the identified care risk rank, the first index, the second index, the third index, and the care risk prediction model, in order to improve the care risk rank of the first user to a care risk rank that is a predetermined level better. A behavior evaluation method using a behavior evaluation server, comprising:
The behavior evaluation server includes a processor and a memory,
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The behavior evaluation method includes:
The processor generates a time series of a behavior of the first user based on the time series of the positions and the time series of the activity amount;
The processor calculates a first index indicating a number of times the first user has gone out based on the time series of the location indicated by the generated time series of the behavior, and a second index indicating a number of places where the first user has stayed based on the time series of the location indicated by the generated time series of the behavior,
The processor calculates a third index indicating an amount of exercise including an average number of steps at a location where the first user stays based on the generated time series of the behavior;
The memory holds the first index, the second index, and the third index for each of a plurality of users including the first user;
The behavior evaluation method includes:
The processor calculates standard scores indicating a ranking among the plurality of users for each of the first index, the second index, and the third index of a first user included in the plurality of users;
The processor calculates a total score of the first user by totaling the standard scores of the first index, the second index, and the third index, as a fourth index for comprehensively evaluating the behavior of the first user;
the memory holds total score rank information indicating a correspondence between the total score and a total score rank;
The behavior evaluation method includes:
The processor refers to the overall score rank information to identify an overall score rank from the overall score of the first user;
A behavioral evaluation method in which the processor calculates an amount of improvement of each of the first index, the second index, and the third index, based on an overall score corresponding to an overall score rank in the overall score rank information that is a predetermined level better than the identified overall score rank, the first index, the second index, and the third index, in order to improve the overall score rank of the first user to an overall score rank that is a predetermined level better. A behavior evaluation method using a behavior evaluation server, comprising:
The behavior evaluation server includes a processor and a memory,
The memory holds a time series of a position of a first user and a time series of an activity amount of the first user;
The behavior evaluation method includes:
The processor generates a time series of a behavior of the first user based on the time series of the positions and the time series of the activity amount;
The processor calculates a first index indicating a number of times the first user has gone out based on the time series of the location indicated by the generated time series of the behavior, and a second index indicating a number of places where the first user has stayed based on the time series of the location indicated by the generated time series of the behavior,
The processor calculates a third index indicating an amount of exercise including an average number of steps at a location where the first user stays based on the generated time series of the behavior;
The memory includes:
User attributes of a plurality of users including the first user;
Intervention service usage history information indicating an intervention service used by each of the plurality of users;
and intervention service information indicating an improvement amount of the first index, the second index, and/or the third index when each of the intervention services is used;
The behavior evaluation method includes:
The processor determines an intervention service to be recommended to the first user from intervention services used by each of the plurality of users having the same user attribute as the first user, the intervention service usage history information indicating the intervention service usage history information;
A behavior evaluation method in which the processor refers to the intervention service information and extracts the amount of improvement of the first index, the second index, and/or the third index when the determined intervention service is used.

Images