JP7726289B2 - Care plan creation support device, care plan creation support method, and care plan creation support program - Google Patents

Description

The present invention relates to a care plan creation support device, a care plan creation support method, and a program recording medium.

Under the long-term care insurance system, care is comprehensively managed by care professionals called care managers. Care managers create home care plans, which are the core of care management. Care support for care recipients is provided in accordance with the care plan. The care plan includes items such as needs and long-term goals.

Non-Patent Document 1 discloses a technology that inputs assessment information and life tasks into an AI (Artificial Intelligence) prediction model and outputs long-term goals, short-term goals, and service content that are highly effective in supporting independence. Specifically, the technology described in Non-Patent Document 1 performs a step-by-step analysis of possible long-term goals based on the input of a user's assessment information and life tasks, possible short-term goals based on the input of those long-term goals, and possible service content and types for those short-term goals.

Patent document 1 discloses a technology that uses a correspondence table to determine corresponding care goals, service items corresponding to the care goals, and service resources, service volume, parallelism, and priority corresponding to the service items from assessment results.

Japanese Patent Application Laid-Open No. 2000-259749

Institute for International Socio-Economic Studies, Inc., “Research Report on Structuring Text Description Data, Which Difficult to Learn from AI in Supporting the Creation of Care Plans,” [online], March 2019, Institute for International Socio-Economic Studies, Inc., [Retrieved January 29, 2020], Internet <URL: https://www.i-ise.com/jp/report/2018/rep_it_201903h.pdf>

Issues (needs), long-term goals, short-term goals, and service content are all related to higher and lower concepts. It is conceivable that a certain lower-level concept item may correspond to multiple higher-level concept items. For example, Non-Patent Document 1 states that the item "Long-term goal: Able to eat meals properly" can be selected from "Needs: Want to eat" as well as "Needs: Want to bathe." Therefore, in order to select the most appropriate long-term goal for a given need, it is necessary to take into account the complex correspondence between care plan items, as described above.

However, the method disclosed in Non-Patent Document 1 only systems up the superordinate concept items and subordinate concept items one-to-one, and does not take into account the complex correspondence relationships described above. As a result, it may not be possible to appropriately select the subordinate care plan items, which are subordinate concepts, for the superordinate care plan items, which are superior concepts. Furthermore, Patent Document 1 does not mention the complex correspondence relationships described above.

The present invention has been made to solve the above problem, and its objective is to provide a device or the like that can appropriately select lower-level care plan items for higher-level care plan items.

In one aspect of the present invention, a care plan creation support device comprises: a prediction means for predicting the probability that a specified care plan sub-item set to satisfy a care plan sub-item will be selected based on care recipient information indicating information about the care recipient and the care plan sub-item set for the care recipient; a calculation means for calculating the contribution of the care plan sub-item to the probability that the specified care plan sub-item will be selected; an item combination extraction means for extracting combinations of a care plan sub-item associated with the specified care plan sub-item from among the care plan sub-items based on the contribution; and an output means for outputting information regarding the combinations.

In addition, a care plan creation support method in one embodiment of the present invention predicts the probability that a specified care plan sub-item set to satisfy a care plan sub-item will be selected based on care recipient information indicating information about the care recipient and the care plan sub-item set for the care recipient, calculates the contribution of the care plan sub-item to the probability that the specified care plan sub-item will be selected, extracts combinations of the specified care plan sub-item and a care plan sub-item related to the specified care plan sub-item from the care plan sub-items based on the contribution, and outputs information about the combinations.

In one embodiment of the present invention, a recording medium non-temporarily records a program that causes a computer to execute the following processes: predicting the probability that a specified care plan sub-item set to satisfy a care plan upper-level item will be selected based on care recipient information indicating information about the care recipient and the care plan upper-level items set for the care recipient; calculating the contribution of the care plan upper-level item to the probability that the specified care plan lower-level item will be selected; extracting combinations of the specified care plan lower-level item and a care plan upper-level item associated with the specified care plan lower-level item from the care plan upper-level items based on the contribution; and outputting information regarding the combinations.

According to the present invention, lower-level care plan items can be appropriately selected for higher-level care plan items.

FIG. 1 is an explanatory diagram showing an example of a care plan. FIG. 2 is a block diagram showing the configuration of the care plan creation support device 1 according to the first embodiment. FIG. 3 is a flowchart showing the flow of the care plan creation support operation performed by the care plan creation support device 1 in the first embodiment. FIG. 4 is a block diagram showing the configuration of a care plan creation support system 2 according to the second embodiment. FIG. 5 is an explanatory diagram showing an example of the data structure of care recipient information such as assessment information stored in the past care plan information storage device 23. FIG. 6 is an explanatory diagram showing an example of the data structure of care plan item information of a care recipient stored in the past care plan information storage device 23. As shown in FIG. FIG. 7 is a flowchart showing the flow of operations of the model generating device 22 in the second embodiment. FIG. 8 is a flowchart showing the flow of the operation of the care plan creation support device 21 in the second embodiment. FIG. 9 is a block diagram showing an example of a hardware configuration.

Hereinafter, care plan creation support devices and care plan creation support systems as embodiments of the present invention will be described with reference to the drawings. First, care plans and the creation of care plans, which are the premise of each embodiment of the present invention, will be described.

Before starting care support for a care recipient, the care manager creates a care plan based on care recipient information, such as assessment information for the care recipient. The care plan creation support device supports the creation of the care plan based on the care recipient information. The care manager also reassesses the care recipient's condition after care support begins, and modifies the care plan as appropriate based on the care recipient's condition. In this case, the care plan creation support device may support changes to the care plan based on the reassessment information for the care recipient.

A care plan is created to clarify the direction of care support for the care recipient and their specific goals. The care plan is shared between the care recipient and the care provider. It includes care plan items such as "lifestyle challenges," "long-term goals," "short-term goals," "service content," and "service type." A care plan may also include plans for care support in all aspects of the care recipient's physical function, language ability, mental function, etc. The care manager determines the higher-level care plan items and then the lower-level care plan items. Here, the higher-level care plan items and the lower-level care plan items are related to each other, and the lower-level care plan items are matters necessary to fulfill the higher-level care plan items and are subordinate concepts to the higher-level care plan items. Care plan items, from the higher-level item to the lower-level item, include lifestyle challenges, long-term goals, short-term goals, service content, and service type. For example, lifestyle challenges are superior to long-term goals, while long-term goals are subordinate concepts to lifestyle challenges. Furthermore, the higher-level care plan items and the lower-level care plan items do not have to be consecutive concepts. For example, short-term goals are sub-items of the life tasks, and service contents are sub-items of the life tasks. When a care manager creates a care plan, the care plan creation support device supports the care manager in selecting care plan sub-items for the care plan super-items set for the care recipient.

"Lifestyle issues" are issues that need to be resolved in the care recipient's overall life. Lifestyle issues are also referred to as "needs" in care plans. Hereinafter, in each embodiment of the present invention, they will be referred to as needs. Needs are set as goals that the care recipient hopes to achieve through nursing care support, such as "I want to continue treatment and manage my physical condition while alleviating pain," "I want to manage my physical condition and prevent a decline in physical function," or "I want to be able to walk with a cane." Needs may also describe the care recipient's current difficulties, such as "I have difficulty bathing alone." Needs may also vary in granularity depending on the care recipient. In other words, needs of various granularities may be set based on care recipient information, such as assessment information. For example, a short-term goal for a care recipient who is currently able to live independently may be set as a need for a care recipient who requires care in all aspects of their life. The granularity of the needs set may also vary depending on the care manager creating the care plan.

"Long-term goals" and "short-term goals" are each set to be gradually more specific in order to resolve needs. In other words, long-term goals may be specific and relatively long-term goals for resolving needs. Furthermore, short-term goals are even more specific goals for achieving long-term goals, and may be set for a shorter period than the target period for achieving the long-term goal. For example, long-term goals may be goals that can be achieved in six months to a year, while short-term goals may be goals that can be achieved in one month to several months. However, the target period for achievement when setting long-term and short-term goals is not limited to this. The long-term and short-term goals set may vary in granularity depending on the difference in granularity of the needs.

"Service content" refers to, for example, the content of the care support provided by a caregiver to a care recipient in order to gradually achieve short-term and long-term goals and ultimately resolve their needs. Furthermore, "service type" refers to, for example, the method and form in which the service content is provided.

Figure 1 is an explanatory diagram showing an example of a care plan. As a specific example of a care plan item, for example, as shown in Figure 1, if the need is set as "I want to avoid placing a burden on my family in my daily life," then the long-term goal is set as "Be able to go to the toilet on my own." Then, to achieve this long-term goal, the short-term goal of "Be able to move around indoors safely" is set. To gradually achieve the short-term and long-term goals and ultimately resolve the need, the caregiver needs to provide assistance to the care recipient, such as "assisting with mobility" and "supervising mobility." The care manager sets these items as service contents. Furthermore, if the care manager determines that these service contents should be provided through "home rehabilitation" or "home care," they are set as "service types."

Assessment information is information about a care recipient obtained by a care manager through an assessment of the care recipient. It is an example of care recipient information. Care recipient information includes information about the care recipient's physical function, language ability, mental function, etc. at the time of the assessment. Care recipient information may also include, for example, the care recipient's age, weight, medical history, and medical history, as well as information about the care recipient's living arrangements (living alone, living with family, etc.), the type of bedding used, the presence or absence of handrails in the home, the presence or absence of steps in the home, the family's health condition, and whether the family is able to care for the care recipient. Care recipient information may also include information about the care recipient's condition, such as the level of care required, determined based on indicators established by local governments. Care recipient information may also be obtained from the care recipient themselves or from a relative or guardian acting on their behalf. If the care recipient has been transferred or discharged from another medical institution or nursing facility at the time of the assessment, the care recipient information may include information shared with the care recipient from their original facility. The care recipient information is not limited to information about the care recipient, as long as it is information about the care recipient that may be used in creating a care plan, excluding information about care plan items.

First Embodiment
The configuration of the care plan creation support device 1 will be described below with reference to Fig. 2. Fig. 2 is a block diagram showing the configuration of the care plan creation support device 1 of this embodiment. The care plan creation support device 1 shown in Fig. 2 includes a prediction unit 11, a calculation unit 12, an item combination extraction unit 13, and an output unit 14.

The prediction unit 11 predicts the probability that a specific care plan subitem will be selected based on care recipient information, such as assessment information, and the higher-level care plan items set for the care recipient. For example, if the higher-level care plan items are needs and the lower-level care plan items are long-term goals, the prediction unit 11 predicts the probability that a specific long-term goal will be selected based on the care recipient information and the needs items set for the care recipient. Specifically, the prediction unit 11 predicts the probability that a specific care plan subitem will be selected using a model generated based on care recipient information in a previously created care plan and information on the higher-level care plan items and lower-level care plan items set for the care recipient. The specific care plan subitem may be set as appropriate by the care plan creation support device 1 or a care manager, etc. A model may be created for each care plan subitem and calculate the probability that a specific care plan subitem will be selected based on the input care recipient information and higher-level care plan item information.

The calculation unit 12 calculates the degree of contribution of the higher-level care plan items set for the care recipient to the probability that a specific lower-level care plan item predicted by the prediction unit 11 will be selected. Specifically, the calculation unit 12 calculates the degree of contribution that indicates the degree to which each higher-level care plan item input into the model used by the prediction unit 11 contributes to increasing the probability that a specific lower-level care plan item predicted by the prediction unit 11 will be selected.

Based on the contribution degree calculated by the calculation unit 12, the item combination extraction unit 13 extracts combinations of care plan upper-level items associated with a specified care plan lower-level item from among the care plan upper-level items set for the care recipient, and the specified care plan lower-level item. For example, among the care plan upper-level items set for the care recipient, items with a high contribution degree are determined to have contributed to increasing the probability that the specified care plan lower-level item is selected. The item combination extraction unit 13 extracts care plan upper-level items that can be determined to have contributed to increasing the probability that the specified care plan lower-level item is selected as care plan upper-level items associated with the specified care plan lower-level item. Then, the item combination extraction unit 13 extracts combinations consisting of care plan upper-level items associated with the specified care plan lower-level item and the specified care plan lower-level item.

The output unit 14 outputs information regarding the combinations extracted by the item combination extraction unit 13. That is, the output unit 14 outputs information indicating that a care plan upper level item in the combination extracted by the item combination extraction unit 13 is associated with a predetermined care plan lower level item. The output unit 14 may also output information indicating that a predetermined care plan lower level item in the combination extracted by the item combination extraction unit 13 is an item that should be selected for the care plan upper level item in the combination. The output unit 14 may further output information indicating that a care plan upper level item in a combination that was not extracted and a predetermined care plan lower level item are not associated with each other.

Next, the care plan creation support operation performed by the care plan creation support device 1 will be explained with reference to Figure 3. Figure 3 is a flowchart showing the flow of the care plan creation support operation performed by the care plan creation support device 1.

In step S11, the prediction unit 11 predicts the probability that a specified care plan sub-item will be selected based on the information of the care recipient and the care plan sub-items set for the care recipient.

In step S12, the calculation unit 12 calculates the contribution of the higher-level care plan items to the probability that a specified lower-level care plan item predicted in step S11 will be selected.

In step S13, the item combination extraction unit 13 extracts combinations of items related to a specified care plan sub-item from among the higher-level care plan items set for the care recipient, based on the contribution rate calculated in step S12, and the specified care plan sub-item.

In step S14, the output unit 14 outputs information regarding the combination extracted in step S13.

As described above, the care plan creation support device 1 of this embodiment outputs information regarding combinations of care plan upper-level items associated with specific care plan lower-level items among the care plan upper-level items set for the care recipient, and the specific care plan lower-level items. The care plan creation support device 1 calculates the contribution of the care plan upper-level items set for the care recipient to the probability that the specific care plan lower-level item will be selected, and can output information indicating whether the care plan upper-level items and the care plan lower-level items are related to each other based on the contribution. In other words, the care plan creation support device 1 is able to select appropriate care plan lower-level items for a care plan upper-level item, even if the correspondence between the care plan upper-level items and the care plan lower-level items is complex. Furthermore, the care plan creation support device 1 is able to select appropriate care plan lower-level items for a care plan upper-level item without necessarily storing a correspondence table or the like indicating the correspondence between the care plan upper-level items and the care plan lower-level items in advance. Furthermore, because the prediction unit 11 only needs to use a model for the number of care plan lower-level items, the care plan creation support device 1 can reduce the computational load required for prediction.

Furthermore, according to this embodiment, the care manager can recognize the care plan subitems related to the care plan superitems set for the care recipient based on the information output by the care plan creation support device 1. That is, the care manager can select appropriate care plan subitems for the care recipient information and the care plan superitem information based on the information output by the care plan creation support device 1. The information output by the care plan creation support device 1 reflects information on care plan subitems that were frequently selected for similar care recipient information and care plan superitems in previously created care plans. In other words, the information output by the care plan creation support device 1 to support the selection of appropriate care plan subitems for the care plan superitems can be said to be information that was supported by many care managers in previously created care plans. Therefore, the care manager can select the care plan subitem with a sense of satisfaction.

Second Embodiment
Next, a second embodiment of the present invention will be described. As an example, a case will be described in which the upper level care plan items are needs and the lower level care plan items are long-term goals. However, as mentioned above, the upper level care plan items and the lower level care plan items are not limited to this.

Figure 4 is a block diagram showing the configuration of a care plan creation support system 2 in a second embodiment of the present invention. As shown in Figure 4, the care plan creation support system 2 in the second embodiment includes a care plan creation support device 21, a model generation device 22, a past care plan information storage device 23, a model storage device 24, and a predicted target person information storage device 25. The model generation device 22 includes a past care plan information acquisition unit 221 and a learning unit 222. The care plan creation support device 21 includes a prediction unit 211, a calculation unit 212, an item combination extraction unit 213, a selection unit 215, and an output unit 214. The prediction unit 211 includes a model acquisition unit 2111, a predicted target person information acquisition unit 2112, and a sub-item selection probability prediction unit 2113.

The care plan creation support device 21 in the second embodiment of the present invention differs from the care plan creation support device 1 in the first embodiment in that it includes a selection unit 215. The prediction unit 211 included in the care plan creation support device 21 in the second embodiment of the present invention shows a detailed configuration example of the prediction unit 11 included in the care plan creation support device 1 in the first embodiment. The calculation unit 212, item combination extraction unit 213, and output unit 214 correspond to the calculation unit 12, item combination extraction unit 13, and output unit 14 included in the care plan creation support device 1 in the first embodiment of the present invention, respectively.

The past care plan information storage device 23 stores information about care plans created in the past as past care plan information. Specifically, the past care plan information storage device 23 stores care recipient information such as assessment information and care plan item information set for the care recipient as information about care plans created in the past. The care plan item information stored in the past care plan information storage device 23 includes needs, long-term goals, short-term goals, service content, service type, etc.

5 and 6 are explanatory diagrams showing an example of the data structure of past care plan information stored in the past care plan information storage device 23. FIG. 5 is an explanatory diagram showing an example of the data structure of care recipient information, such as assessment information, stored in the past care plan information storage device 23. For example, for each item of the care recipient information shown in FIG. 5, the item contents, such as "IIb" and "apartment," may be converted to integer values or real numbers corresponding to each item and its content and then stored. In this case, the past care plan information storage device 23 may store the correspondence between the item contents of the care recipient information and their numerical values. FIG. 6 is an explanatory diagram showing an example of the data structure of care plan item information for a care recipient stored in the past care plan information storage device 23. The care plan item information for a care recipient indicates whether or not each care plan item has been set for the care recipient. Therefore, as shown in FIG. 6, each item of the care plan item information set for a care recipient has two values, for example, "1" or "0," with items set for the care recipient being stored as "1" and items not set being stored as "0."

As shown in Figures 5 and 6, the past care plan information storage device 23 may separately store care recipient information, such as assessment information, related to previously created care plans, and care plan item information set for the care recipient. The past care plan information storage device 23 may also store past care plan information in association with a care recipient ID, which is the care recipient's identification information. Furthermore, the past care plan information storage device 23 may store other items that may be included in a care plan, such as those shown in Figure 1, as past care plan information. For example, the date the assessment was conducted, the date the care plan was created, and the period set for each care plan item. The past care plan information storage device 23 may also store information about the care plan creator linked to each care plan as past care plan information. Information about the care plan creator may include, for example, the name of the care plan creator and the creator's level of expertise.

Next, we will explain the care plan past information acquisition unit 221 and the learning unit 222 provided in the model generation device 22.

The care plan past information acquisition unit 221 acquires care plan past information from the care plan past information storage device 23.

The learning unit 222 learns the past care plan information acquired by the past care plan information acquisition unit 221. Then, the learning unit 222 generates a long-term goal selection probability prediction model as the learning result. The long-term goal selection probability prediction model is a model that predicts the probability that a specific long-term goal will be selected for a care recipient based on input of care recipient information such as assessment information and information on the needs set for the care recipient. The selection probability of a long-term goal indicates the probability that the long-term goal will be selected for the care recipient information and needs. When the learning unit 222 learns the past care plan information, the selection probability of a long-term goal predicted by the long-term goal selection probability prediction model indicates a value based on the probability that the long-term goal was selected for the care recipient information and needs in the care plans created in the past that were learned.

Specifically, the learning unit 222 uses past care plan information to learn combinations of long-term goals to be selected based on care recipient information and the needs set for the care recipient. The learning unit 222 then generates a long-term goal selection probability prediction model as a result of the learning. The learning unit 222 can use known machine learning techniques when learning combinations of long-term goals to be selected based on care recipient information and the needs set for the care recipient. The long-term goal selection probability prediction model is generated, for example, for each long-term goal item. The long-term goal selection probability prediction model is generated, for example, for each long-term goal item, and predicts the probability of selection for each long-term goal item. The learning unit 222 uses, for example, a linear regression model as the long-term goal selection probability prediction model. The learning unit 222 can use other learning methods as long as they use a linear regression model. The learning unit 222 may also use a model generated by deep learning as the long-term goal selection probability prediction model. The long-term goal selection probability prediction model is not limited to this, as long as it is a model that can predict the selection probability of a long-term goal and calculate the contribution rate, as described below.

The model storage device 24 stores the long-term goal selection probability prediction model generated by the model generation device 22. The model storage device 24 may also store the date and time when the long-term goal selection probability prediction model was generated by the model generation device 22 and the date and time when it was updated.

The predicted subject information storage device 25 stores predicted subject information. Specifically, the predicted subject information storage device 25 includes, as predicted subject information, care subject information such as assessment information regarding the care subject who is the predicted subject of the long-term goal, and information on needs set for the care subject. The predicted subject information storage device 25 may store predicted subject information output from an input device such as a tablet into which a care manager inputs predicted subject information. Furthermore, the information on needs set for the care subject may be set by the care manager based on the care subject information, or may be determined using a known method based on the care subject information.

Next, we will explain the care plan creation support device 21, which is equipped with a model acquisition unit 2111, a predicted subject information acquisition unit 2112, a calculation unit 212, an item combination extraction unit 213, a selection unit 215, and an output unit 214.

The model acquisition unit 2111 acquires a long-term goal selection probability prediction model from the model storage device 24. If a long-term goal selection probability prediction model is generated for each long-term goal, the model acquisition unit 2111 may acquire a long-term goal selection probability prediction model for each long-term goal.

The predicted subject information acquisition unit 2112 acquires predicted subject information from the predicted subject information storage device 25. The predicted subject information acquisition unit 2112 may acquire predicted subject information directly from an input device such as a tablet, which the care manager has input using that input device. The predicted subject information acquisition unit 2112 may also acquire predicted subject information for the care subject to be predicted each time a prediction is made.

The sub-item selection probability prediction unit 2113 inputs the predicted target person information acquired by the predicted target person information acquisition unit 2112 into the long-term goal selection probability prediction model acquired by the model acquisition unit 2111, and predicts the selection probability of a specified long-term goal. Specifically, the sub-item selection probability prediction unit 2113 uses the long-term goal selection probability prediction model to predict the selection probability of a specified long-term goal for the predicted target person information. For example, when the learning unit 222 learns past care plan information, items with a high selection probability for the long-term goal are items that were frequently selected by care managers for past care targets who have similar care target information and needs information as the predicted target.

The calculation unit 212 calculates the contribution of needs to the selection probability of a specified long-term goal predicted by the sub-item selection probability prediction unit 2113. Here, the contribution is a value indicating the degree to which each item of needs set for the care recipient contributes to increasing the selection probability of the specified long-term goal. The contribution may be a value expressed as an integer value or a real number. The contribution may also be expressed as a negative value. For example, a larger contribution indicates a greater contribution to increasing the selection probability of the specified long-term goal. For example, in some cases, a larger positive contribution indicates a greater contribution to increasing the selection probability of the specified long-term goal.

Based on the contribution rate calculated by the calculation unit 212, the item combination extraction unit 213 extracts combinations of items related to a specific long-term goal among the needs set for the care recipient and the specific long-term goal. It can be determined that items with a high contribution rate among the needs set for the care recipient contributed to increasing the selection probability of the specific long-term goal. In other words, the item combination extraction unit 213 can determine that the specific long-term goal and a need item with a high contribution rate are highly related, or that a need item with a high contribution rate is related to the specific long-term goal. In this way, the item combination extraction unit 213 extracts combinations of a specific long-term goal and a need item related to the specific long-term goal. The item combination extraction unit 213 may also determine whether the contribution rate is greater than a specific threshold to determine whether the contribution rate contributed to increasing the selection probability of the specific long-term goal. In this case, the threshold may be zero. In other words, the item combination extraction unit 213 may determine whether a need item contributed to increasing the selection probability of the specific long-term goal by determining whether the contribution rate is positive or negative. In this case, the item combination extraction unit 213 can determine that a positive contribution rate contributes to an increase in the selection probability, and that a negative contribution rate does not contribute to an increase in the selection probability. Here, the item combination extraction unit 213 may also determine that a negative contribution rate contributes to a decrease in the selection probability.

The selection unit 215 selects a combination to output from among multiple combinations including a specific need using the selection probability of the long-term goal. The combination to output from among multiple combinations including a specific need is extracted by the item combination extraction unit 213. The selection probability of the long-term goal is predicted by the sub-item selection probability prediction unit 2113. Specifically, when there are multiple combinations including a specific need item, the selection unit 215 selects the combination extracted by the item combination extraction unit 213 that has a high selection probability for the specific long-term goal predicted by the sub-item selection probability prediction unit 2113. Here, the case where there are multiple combinations including a specific need item refers to the case where there are multiple long-term goal items related to a specific need item. For example, the selection unit 215 may select a combination including a long-term goal item whose selection probability predicted by the sub-item selection probability prediction unit 2113 is equal to or greater than a specific value. The selection unit 215 may also select corresponding combinations in descending order of the selection probability of the long-term goal item. The selection unit 215 may select combinations including long-term goals whose selection probability is equal to or greater than a predetermined value in descending order of the selection probability.

The output unit 214 outputs information about the combination selected by the selection unit 215. If the selection unit 215 selects a combination that includes a long-term goal item with a selection probability equal to or greater than a predetermined value, the output unit 214 may output all of the selected combinations as information about the combination. The output unit 214 may also output the long-term goals included in the combination as information about the combination, as appropriate long-term goals to be selected for the needs items in the predetermined combination. For example, if the selection unit 215 selects corresponding combinations in descending order of the selection probability of the long-term goal items, the output unit 214 may output the long-term goal items in that order. In this case, the output unit 214 may rank and output the long-term goal items in an order of appropriate long-term goals for the care recipient. The output unit 214 may also output the long-term goal item with the highest selection probability from among the combinations selected by the selection unit 215 as the long-term goal item optimal for the needs included in the combination.

The care plan creation support system 2 may be equipped with a display device not shown in FIG. 4. The display device displays the information output by the care plan creation support device 21. In this case, the display device may display the information together with the care recipient information stored in the predicted recipient information storage device 25. For example, in addition to the information on long-term goals appropriate for the needs set for the care recipient output by the care plan creation support device 21, the display device also shows assessment information for the care recipient. In this way, for example, a care manager can confirm whether the long-term goal items output by the care plan creation support system 2 are appropriate for the assessment information of the care recipient.

At least one of the devices included in the care plan creation support system 2 may be realized within a single device such as a server. Furthermore, at least one of the devices included in the care plan creation support system 2 may be realized within an information terminal such as a computer owned by a care manager. Furthermore, at least one of the devices included in the care plan creation support system 2 may be a server installed in a facility where a care manager or caregiver works, or may be realized on a cloud server.

Next, we will explain the operations performed by the care plan creation support system 2.

First, the operations performed by the model generation device 22 will be explained with reference to Figure 7. Figure 7 is a flowchart showing the flow of operations performed by the model generation device 22.

As shown in FIG. 7, in step S211, the care plan past information acquisition unit 221 acquires information regarding stored care plans created in the past from the care plan past information storage device 23.

In step S212, the learning unit 222 uses the past care plan information acquired in step S211 to learn a combination of long-term goals to be selected for the care recipient information and the needs set for the care recipient.

In step S213, the learning unit 222 generates a long-term goal selection probability prediction model.

In step S214, the generated long-term goal selection probability prediction model is stored in the model memory device 24.

The model generation device 22 may perform the above-described model generation operations at any timing and update the model. For example, when past care plan information is added, the care plan past information storage device 23 sends a model generation request to the model generation device 22. The model generation device 22 may receive this request and generate/update the model. The model generation device 22 may generate/update the model periodically at a predetermined timing. The timing at which the model generation device 22 generates/updates the model is not limited to this.

Next, the operations performed by the care plan creation support device 21 will be explained with reference to Figure 8. Figure 8 is a flowchart showing the flow of operations performed by the care plan creation support device 21.

As shown in FIG. 8, in step S221, the model acquisition unit 2111 acquires a long-term goal selection probability prediction model from the model storage device 24.

In step S222, the predicted subject information acquisition unit 2112 acquires predicted subject information including care subject information such as assessment information and information on the needs set for the care subject from the predicted subject information storage device 25.

In step S223, the sub-item selection probability prediction unit 2113 inputs the predicted subject information acquired by the predicted subject information acquisition unit 2112 into the specified long-term goal selection probability prediction model acquired by the model acquisition unit 2111, and predicts the selection probability of the specified long-term goal.

In step S224, the calculation unit 212 calculates the contribution of each of the need items set for the care recipient to the probability of selecting the long-term goal predicted in step S223.

In step S225, the item combination extraction unit 213 extracts combinations of items related to the specified long-term goals among the needs set for the care recipient, based on the contribution rates calculated in step S224. Note that, as described in the first embodiment, the output unit 214 may output the combinations selected at this point.

If a long-term goal selection probability prediction model is generated for each long-term goal, the care plan creation support device 21 performs steps S223 to S225 for each long-term goal item. In other words, the care plan creation support device 21 may repeat steps S223 to S225 for the number of models, i.e., the number of types of long-term goals.

In step S226, the selection unit 215 obtains the combinations extracted in step S225 and the selection probabilities of the specified long-term goals predicted in step S223 for each long-term goal item. Furthermore, the selection unit 215 selects the combinations extracted in step S225 based on this information.

In step S227, the output unit 214 outputs information regarding the combination selected in step S226.

Below, the operation of the care plan creation support device 21 is explained using the specific example where the long-term goal selection probability prediction model is a linear regression model.

The number of care plan items is assumed to be m needs and n long-term goals. N long-term goal selection probability prediction models are generated for each long-term goal item. The long-term goal selection probability prediction model takes as input r pieces of care recipient information, such as assessment information, and information on the m needs of the care recipient, and outputs the probability that a certain long-term goal item i (i=1, 2, ..., n) will be selected.

Needs are expressed as values that indicate whether or not they can be selected for the care recipient. Needs can have two values, for example, "0" or "1", where needs that are set for the care recipient are "1" and needs that are not set are "0". Care recipient information is expressed as integer or real values.

When the long-term goal selection probability prediction model is a linear regression model, the probability p _i that a long-term goal item i (i=1, 2, ..., n) is selected, i.e., the long-term goal selection probability p _i , is expressed by the following prediction formula given by a linear combination:
p _i = a _ij x _j + n _ik y _k + b _i
Here, x _j is the input value of care recipient information item j (j=1,2,...,r). y _k is the input value of needs item k (k=1,2,...,m). _{a ij} and n _ik are coefficients for the input values of care recipient information and needs. _{b i} is a bias term.

In step S221, the model acquisition unit 2111 acquires the above prediction formula as a long-term goal selection probability prediction model from the model storage device 24. In step S222, the prediction target information acquisition unit 2112 acquires care target information and needs information from the prediction target information storage device 25. The prediction target information acquisition unit 2112 converts the acquired care target information and needs information into input values xj _{, yk} for the prediction formula, respectively, as necessary. Then, in step S223, the sub-item selection probability prediction unit 2113 inputs xj _{, yk} into the prediction formula and predicts the selection probability _pj of long-term goal item i. In step S224, the calculation unit 212 calculates the product _nikyk of the input value yk of needs item _k and the corresponding _{coefficient nik} as the contribution _Cik of needs item k to long-term goal item i. In step S225, the item combination extraction unit 213 determines whether the needs item k contributed to the selection probability p _i of the long-term goal item i, for example, by determining whether the contribution C _ik is greater than a threshold. The item combination extraction unit 213 determines that the needs item k determined to have contributed is an item related to the long-term goal item i, and extracts combinations of the needs item k determined to have contributed and the long-term goal item i. The care plan creation support device 21 performs the above-described steps S223 to S225 for each of the long-term goal items 1 to n to extract combinations of the needs item k and the long-term goal item i. The output unit 214 may output the combinations extracted in step S225. Furthermore, if there are multiple combinations including a certain needs item k among the combinations extracted in steps S223 to S225, the process of step S226 is performed. In step S226, the selection unit 215 makes a selection based on the selection probability p _i of each long-term goal. Here, when there are multiple long-term goal items related to a certain need item k, this indicates that there are multiple long-term goal items related to a certain need item k. Then, in step S227, the output unit 214 outputs information about the selected combination.

For example, if the long-term goal selection probability prediction model is a linear regression model and need items 1, 3, and 4 are set for the care recipient, the input values for the need items are y = ( _y1 , _y2 , _y3 , _y4 , _y5 , ..., _ym ) = (1, 0, 1, 1, 0, ..., 0). In this case, the selection probability _p1 for long-term goal item 1 (i = 1) is
p ₁ = a _1j x _j + n ₁₁ + n ₁₃ + n ₁₄ + b ₁ (j=1,2,…,r)
The lower-level item selection probability prediction unit 2113 predicts _p1 . The calculation unit 212 calculates the contribution _C1 = ( _n11 , 0, _n13 , _n14 , 0, ..., 0) of long-term goal item 1. When the contribution threshold is n and _n11 > _n13 > n > n14 holds, the item combination extraction unit 213 determines that the needs items that contributed to the increase in the selection probability _p1 of long-term goal item 1 are items 1 and 3 (k = 1, 3). That is, the item combination extraction unit 213 extracts ( _i , k) = (1, 1), (1, 3) as combinations of a specified long-term goal and needs items related to the specified long-term goal. The care plan creation support device 21 performs similar processing for long-term goal items 1 to n to extract the above combinations. As a result, it is assumed that (i, k) = (1, 1), (2, 1), (4, 1) is extracted for the combination related to needs item 1. In other words, this indicates that it is appropriate to select items 1, 2, and 4 of the long-term goal for needs item 1. In this case, the output unit 214 may output the extraction result of this combination. Furthermore, the selection unit 215 makes selection for the combination (i, k) = (1, 1), (2, 1), and (4, 1) using the values of selection probabilities _p1 , _p2 , and _p4 of items 1, 2, and 4 of the long-term goal. The output unit 214 outputs the long-term goal items that are appropriate to select for needs item 1 as information about the combination selected by the selection unit 215. Specifically, the output unit 214 may output long-term goals whose selection probabilities _p1 , _p2 , and _p4 are higher than predetermined values. Furthermore, the output unit 214 may output the long-term goal items in descending order of selection probability, or may output the item with the highest selection probability as the optimal long-term goal. For example, when selection probabilities _p1 = 0.9, _p2 = 0.1, and _p4 = 0.8 and a predetermined value of 0.5 are set, the output unit 214 outputs a message indicating that it is appropriate to select long-term goal items 1 and 4 for needs item 1. Furthermore, for example, the output unit 214 may output a message indicating that it is appropriate to select long-term goal items 1, 4, and 2 in this order for needs item 1, or may output a message indicating that long-term goal item 1 is optimal.

Furthermore, when the long-term goal selection probability prediction model is a model generated by deep learning, important need items are identified based on their sensitivity to changes in the input value of the need items, and the product of the importance of the identified item and the input value is used as the contribution to the probability of selecting a specific long-term goal item. The importance is derived using a known method, for example, as disclosed in the following document.
・RIBEIRO, Marco Tulio; SINGH, Sameer; GUESTRIN, Carlos. “Why should I trust you?: Explaining the predictions of any classifier,” Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, Association for Computing Machinery, 2016, p. 1135-1144
・LUNDBERG, Scott M.; LEE, Su-In. “A unified approach to interpreting model predictions,” Proceedings of the Advances in Neural Information Processing Systems 30, 2017, p. 4765-4774

As described above, the care plan creation support system 2 of this embodiment can narrow down and output combinations of needs set for a care recipient and long-term goals related to those needs by selecting them using the selection probability of the long-term goals. In other words, the care plan creation support system 2 can output long-term goals that are more likely to be selected for a given need. In particular, when there are multiple long-term goal items related to the needs set for a care recipient, the care plan creation support system 2 of this embodiment selects combinations of needs and long-term goals using the selection probability of the long-term goals. In this way, it is possible to more appropriately indicate the long-term goals to be selected for a given need.

Based on the information output by the care plan creation support device 21, the care manager can recognize the long-term goals that should be selected for the needs set for the care recipient. In other words, the care manager can select more appropriate long-term goal items as step-by-step goals for resolving the needs of the care recipient. The information output by the care plan creation support device 21 reflects information on long-term goals that were frequently selected for similar care recipient information and needs in past care plan information. In other words, the appropriate long-term goals indicated by the care plan creation support device 21 can be said to be those that were supported by many care managers in care plans created in the past. Therefore, the care manager can select more appropriate long-term goal items with a sense of conviction.

Furthermore, by having the care plan creation support device 21 rank and output the long-term goals to be selected based on needs, the care manager can recognize the order in which long-term goals should be selected based on the needs of the care recipient. This allows the care manager to select the optimal long-term goals. Furthermore, when creating a care plan, the care manager may need to consider circumstances other than the care recipient information, such as when the care content that can be provided to the care recipient is limited due to equipment. In such cases, if the long-term goal indicated as optimal by the care plan creation support device 21 cannot be selected, the care manager can select the most optimal long-term goal within the scope of what is achievable.

(Variation 1)
In this embodiment, the learning unit 222 may use past care plan information categorized by the content of a specific item in the past care plan information to generate a long-term goal selection probability prediction model for each item. For example, the learning unit 222 performs learning for each stage of nursing care certification using past care plan information categorized by the level of nursing care certification of the care recipient. It is considered that the condition of the care recipient and the degree to which the care recipient improves with the caregiver's support vary significantly depending on the level of nursing care certification. Therefore, by categorizing the past care plan information by the level of nursing care certification, the learning unit 222 can learn past care plan information with a strong correlation between care plan items set for the care recipient. This improves the accuracy of the long-term goal selection probability prediction model. By using the long-term goal selection probability prediction model, the care plan creation support device 21 can output more appropriate information.

In this embodiment, the learning unit 222 may generate a long-term goal selection probability prediction model using, for example, past care plan information categorized by information about the care plan creator. Care plan creation proficiency varies depending on the care plan creator, i.e., the care manager, and it is conceivable that the appropriateness of care plan item selection may vary. Therefore, the learning unit 222 can generate a highly reliable long-term goal selection probability prediction model by, for example, selecting and using information about care plans created by highly skilled care managers from past care plan information categorized by the care manager's proficiency. The model generation device 22 generates a long-term goal selection probability prediction model using care plans created by highly skilled care managers, and the care plan creation support device 21 uses this model, allowing the care plan creation support device 21 to output even more appropriate information.

(Variation 2)
In this embodiment, an example is shown in which the upper level care plan items are needs and the lower level care plan items are long-term goals. However, the upper level care plan items and the lower level care plan items are not limited to this. For example, the upper level care plan items may be long-term goals and the lower level care plan items may be short-term goals. In this case, the care plan creation support device 21 may output short-term goals to be selected for the long-term goals set for the care recipient. Furthermore, the care plan creation support device 21 may output long-term goal items appropriate for the needs and then output short-term goals appropriate for the long-term goals, thereby gradually outputting care plan items to be set for the care recipient. In this way, the care plan creation support device 21 can gradually propose appropriate long-term goals, short-term goals, service contents, and service types based on the care recipient information and set needs of the care recipient.

The upper and lower care plan items do not have to be consecutive upper and lower care plan items. For example, the upper care plan items may represent needs and the lower care plan items may represent service contents, and the care plan creation support device 21 may output service contents to be selected for the needs set for the care recipient. In this case, the model generation device 22 learns the probability that a specific service content will be selected for the care recipient information and the set needs information from past care plan information, and generates a model that outputs this probability. The care plan creation support device 21 performs steps S221 to S227 using this model, making it possible to directly output service contents to be selected for the needs set for the care recipient.

The care plan creation support device 21 of this embodiment may output all care plan items, such as appropriate long-term goals, short-term goals, service content, and service type, based on the care recipient information and needs. In this case, as described above, other care plan items may be predicted in stages based on the needs, or each care plan item may be predicted directly based on the needs. The care plan creation support device 21 may also output the information predicted in this manner. Specifically, the care plan creation support device 21 may output, for example, as shown in FIG. 1, an example of a care plan to be created to a display device not shown in FIG. 4.

<Hardware configuration for realizing each part of the embodiment>
In each embodiment of the present invention, each component of each device (system) represents a functional block. Some or all of the components of each device (system) are realized by any combination of an information processing device 3 and a program, for example, as shown in Fig. 9. The information processing device 3 includes, as an example, the following configuration.

・CPU (Central Processing Unit) 301
・ROM (Read Only Memory) 302
・RAM (Random Access Memory) 303
Program 304 loaded into RAM 303
A storage device 305 for storing the program 304
A drive device 307 that reads and writes data from and to the recording medium 306
A communication interface 308 that connects to a communication network 309
Input/output interface 310 for inputting and outputting data
A bus 311 that connects the components

Each component of each device in each embodiment is realized by CPU 301 acquiring and executing program 304 that realizes these functions. Program 304 that realizes the functions of each component of each device is stored in advance in storage device 305 or RAM 303, for example, and is read by CPU 301 as needed. Program 304 may be supplied to CPU 301 via communication network 309, or may be stored in advance on recording medium 306, with drive device 307 reading the program and supplying it to CPU 301.

There are various variations in how each device can be realized. For example, each device may be realized by any combination of a separate information processing device 3 and a program for each component. Furthermore, multiple components provided by each device may be realized by any combination of a single information processing device 3 and a program.

In addition, some or all of the components of each device may be realized by general-purpose or dedicated circuits including processors, etc., or a combination of these. These may be configured by a single chip, or by multiple chips connected via a bus. Some or all of the components of each device may be realized by a combination of the above-mentioned circuits, etc., and programs.

When some or all of the components of each device are realized by multiple information processing devices, circuits, etc., the multiple information processing devices, circuits, etc. may be centrally located or distributed. For example, the information processing devices, circuits, etc. may be realized as a client-server system, cloud computing system, etc., in which each device is connected via a communications network.

Note that the above explanation shows an example of providing support for creating a care plan, but the present invention is not limited to the creation of care plans and can be applied to any situation in which support is provided for proposing sub-items that are determined in correspondence with a higher-level item.

The present invention has been described above with reference to embodiments, but the present invention is not limited to the above embodiments. Various modifications to the configuration and details of the present invention that are understandable to those skilled in the art can be made within the scope of the present invention. Furthermore, the configurations of the above-described embodiments may be combined, some components may be interchanged, or some components may be carried by other devices.

A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes.
(Appendix 1)
a prediction means for predicting the probability that a predetermined care plan lower item set to satisfy a care plan upper item will be selected based on care recipient information indicating information about the care recipient and the care plan upper items set for the care recipient;
a calculation means for calculating a contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item is selected;
an item combination extraction means for extracting a combination of a care plan upper level item related to the predetermined care plan lower level item among the care plan upper level items and the predetermined care plan lower level item based on the contribution degree;
an output means for outputting information about the combination;
A care plan creation support device comprising:
(Appendix 2)
The method further comprises a selection means for selecting, using the probability, a combination to be output by the output means from a plurality of combinations including the predetermined upper level items of the care plan extracted by the item combination extraction means.
2. A care plan creation support device according to claim 1.
(Appendix 3)
The selection means selects the combination including the care plan sub-item whose probability is equal to or greater than a predetermined value.
3. A care plan creation support device according to claim 2.
(Appendix 4)
The care plan creation support device according to claim 2 or 3, wherein the selection means selects the combinations in descending order of the probability of the care plan subitems.
(Appendix 5)
the output means outputs the care plan sub-item with the highest probability from among the combinations selected by the selection means.
5. A care plan creation support device according to any one of Supplementary Note 2 to Supplementary Note 4.
(Appendix 6)
The item combination extraction means extracts top care plan items whose contribution rate is equal to or greater than a predetermined value.
6. A care plan creation support device according to any one of claims 1 to 5.
(Appendix 7)
the prediction means predicts the probability using a model that outputs the probability for the care recipient information and the higher-level items of the care plan.
7. A care plan creation support device according to any one of claims 1 to 6.
(Appendix 8)
the prediction means predicts the probability using a model that outputs the probability generated for each of the care plan sub-items;
8. A care plan creation support device according to any one of claims 1 to 7.
(Appendix 9)
the model is a linear regression model;
The care plan creation support device according to claim 7 or 8, wherein the calculation means calculates the product of a coefficient of a prediction formula expressed by a linear combination and an input value indicating whether the higher-level item of the care plan can be selected as the contribution of the higher-level item of the care plan.
(Appendix 10)
The care plan creation support device according to any one of Supplementary Note 1 to Supplementary Note 9, wherein the information on the care recipient includes at least one of information on the care recipient's physical function, language ability, and mental function.
(Appendix 11)
The top items in the care plan are needs,
11. A care plan creation support device according to any one of claims 1 to 10.
(Appendix 12)
The care plan sub-items are long-term goals;
12. A care plan creation support device according to any one of claims 1 to 11.
(Appendix 13)
predicting the probability that a predetermined care plan sub-item set to satisfy the care plan sub-item will be selected based on care recipient information and the care plan sub-item set for the care recipient;
calculating the contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item will be selected;
extracting a combination of a care plan upper level item related to the predetermined care plan lower level item from among the care plan upper level items based on the contribution degree, and the predetermined care plan lower level item;
outputting information about the combination;
How to support the creation of care plans.
(Appendix 14)
On the computer,
A process of predicting the probability that a predetermined care plan sub-item set to satisfy the care plan super-item will be selected based on care recipient information and the care plan super-item set for the care recipient;
A process of calculating the contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item will be selected;
A process of extracting a combination of a care plan upper level item related to the predetermined care plan lower level item from among the care plan upper level items based on the contribution degree, and the predetermined care plan lower level item;
outputting information about the combination;
A program recording medium that records a care plan creation support program for executing the above.

1 Care plan creation support device 11 Prediction unit 12 Calculation unit 13 Item combination extraction unit 15 Output unit 2 Care plan creation support system 21 Care plan creation support device 211 Prediction unit 2111 Model acquisition unit 2112 Prediction target person information acquisition unit 2113 Sub-item selection probability prediction unit 212 Calculation unit 213 Item combination extraction unit 214 Output unit 215 Selection unit 22 Model generation device 221 Care plan past information acquisition unit 222 Learning unit 23 Care plan past information storage device 24 Model storage device 25 Prediction target person information storage device 3 Information processing device 301 CPU
302 ROM
303 RAM
304 Program 305 Storage device 306 Recording medium 307 Drive device 308 Communication interface 309 Communication network 310 Input/output interface 311 Bus

Claims (14)

a prediction means for predicting the probability that a predetermined care plan lower item set to satisfy a care plan upper item will be selected based on care recipient information indicating information about the care recipient and the care plan upper items set for the care recipient;
a calculation means for calculating a contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item is selected;
an item combination extraction means for extracting a combination of a care plan upper level item related to the predetermined care plan lower level item among the care plan upper level items and the predetermined care plan lower level item based on the contribution degree;
an output means for outputting information about the combination;
A care plan creation support device comprising: The method further comprises a selection means for selecting, using the probability, a combination to be output by the output means from a plurality of combinations including the predetermined upper level items of the care plan extracted by the item combination extraction means.
The care plan creation support device according to claim 1 . The selection means selects the combination including the care plan sub-item whose probability is equal to or greater than a predetermined value.
The care plan creation support device according to claim 2. The selection means selects the combinations in descending order of the probabilities of the care plan sub-items.
The care plan creation support device according to claim 2 or 3. the output means outputs the care plan sub-item with the highest probability from among the combinations selected by the selection means.
5. The care plan creation support device according to claim 2. The item combination extraction means extracts top care plan items whose contribution rate is equal to or greater than a predetermined value.
6. The care plan creation support device according to claim 1. the prediction means predicts the probability using a model that outputs the probability for the care recipient information and the higher-level items of the care plan.
7. The care plan creation support device according to claim 1. the prediction means predicts the probability using a model that outputs the probability generated for each of the care plan sub-items;
8. The care plan creation support device according to claim 1. the model is a linear regression model;
The care plan creation support device according to claim 7 or claim 8, wherein the calculation means calculates the product of a coefficient of a prediction formula expressed by a linear combination and an input value indicating whether the higher-level item of the care plan can be selected as the contribution of the higher-level item of the care plan. The care plan creation support device according to any one of claims 1 to 9, wherein the information about the care recipient includes at least one of information about the care recipient's physical function, language ability, and mental function. The top items in the care plan are needs,
11. The care plan creation support device according to claim 1. The care plan sub-items are long-term goals;
12. The care plan creation support device according to claim 1. The computer predicts the probability that a predetermined care plan sub-item set to satisfy the care plan sub-item will be selected based on care recipient information indicating information about the care recipient and the care plan sub-item set for the care recipient;
calculating the contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item will be selected;
extracting a combination of a care plan upper level item related to the predetermined care plan lower level item from among the care plan upper level items based on the contribution degree, and the predetermined care plan lower level item;
outputting information about the combination;
How to support the creation of care plans. On the computer,
A process of predicting the probability that a predetermined care plan sub-item set to satisfy a care plan super-item will be selected based on care recipient information indicating information about the care recipient and the care plan super-item set for the care recipient;
A process of calculating the contribution of the care plan superordinate item to the probability that the predetermined care plan subordinate item will be selected;
A process of extracting a combination of a care plan upper level item related to the predetermined care plan lower level item from among the care plan upper level items based on the contribution degree, and the predetermined care plan lower level item;
outputting information about the combination;
A program to help create care plans to help implement these.