Support devices for home care service providers

Abstract

To enable comprehensive and compassionate care and assistance, we provide technical means to reduce the burden on service providers in obtaining the special service provider allowance. [Solution] The support device 1 for home care service providers according to the present invention comprises a display command unit 111 that displays visit information relating to visits related to home care services on a screen, and an instruction generation unit 112 that automatically generates text to be communicated by the service provider responsible for home care services to home care workers, etc., regarding the visit. The display command unit 111 provides an evaluation and modification user interface for the service manager responsible for evaluating and / or modifying the text automatically generated by the instruction generation unit 112. The instruction generation unit 112 automatically generates text by a procedure for randomly selecting from a group of text candidates including multiple text candidates, or by an arithmetic process that takes visit data relating to the visit as input.

Description

Technical Field

[0001] The present invention relates to a support device for home nursing care service providers.

Background Art

[0002] In order to continuously carry out the home nursing care service business and contribute to the improvement of national healthcare and the enhancement of welfare, appropriate business operations are required.

[0003] Regarding appropriate business operations in the home nursing care service business, Patent Document 1 discloses a centralized home nursing care management method executed by a centralized home nursing care management device for a business office engaged in home nursing care. When an instruction is given by the person in charge of creating the visit schedule to move the high-demand display to the column of the home nursing care staff with high experience in the allocation table of home nursing care staff of any other business office in the high-demand vacancy display, the inter-operator movement instruction information is acquired, and the visit schedule list of the home nursing care staff of the reorganized business office is displayed on the screen. The technology of Patent Document 1 can reduce the indirect work in visit schedule management in home nursing care services.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] By the way, in the proper operation of a visiting care service business, it is important to meet various requirements regarding the specific workplace addition and obtain such addition. Such various requirements include, for example, system requirements including the requirement "(3) Transmission by documents such as user information, reports from visiting care workers, etc.". When operating the business so that intimate care and assistance can be provided without omission, the burden on the service provider responsible for transmission and reporting regarding this requirement is significant.

[0006] Although the technology of Patent Document 1 can reduce indirect work in visit schedule management, it does not lead to a reduction in the burden on the service provider responsible for transmission and reporting. Therefore, there is room for further improvement in the technology in terms of reducing the burden on the service provider regarding obtaining the specific workplace addition so that intimate care and assistance can be provided without omission.

[0007] The present invention has been made to solve the problems of the prior art as described above, and an object thereof is to provide an advanced technical means for reducing the burden on the service provider regarding obtaining the specific workplace addition so that intimate care and assistance can be provided without omission.

Means for Solving the Problem

[0008] As a result of intensive studies to solve the above problems, the inventors of the present invention have found that the above problems can be solved by a configuration in which the text transmitted by the service provider to the visiting care worker, etc. is automatically generated by arithmetic processing using visit data related to the visit and other technical means. And the inventors of the present invention have completed the present invention. Specifically, the present invention provides the following.

[0009] An invention according to a first feature of the present invention includes a display command unit that causes a screen to display visit information regarding a visit related to a visiting care service, and an instruction generation unit that automatically generates a text to be transmitted by a service management person in charge of the visiting care service to a person in charge of actual work who visits the user's home to provide care services to the service user regarding the visit. The display command unit provides an evaluation and correction user interface for the service management person in charge to evaluate and / or correct the text automatically generated by the instruction generation unit. The instruction generation unit automatically generates the text by a procedure of randomly extracting from a group of text candidates including a plurality of text candidates, or by arithmetic processing taking visit data related to the visit as input. The arithmetic processing calculates a weight of the text candidates calculated based on the visit data for a group of text candidates including a plurality of text candidates, and extracts a text candidate from the group of text candidates based on the weight, or includes at least one of a procedure of inputting a prompt including the visit data to a generation AI and causing the generation AI to generate a response in natural language, and provides a support device for a visiting care service provider.

[0010] According to the invention, regarding "(3) Transmission by documents of user information, reports from visiting care workers, etc." among the institutional requirements related to the specific office addition, a transmission text that the service provider in charge transmits to the person in charge of actual work is automatically generated and displayed on the screen, thereby reducing the burden on the service provider in charge regarding the drafting of the transmission text.

[0011] According to the invention, the instruction generation unit automatically generates the text by arithmetic processing taking visit data related to the visit as input. Thereby, the invention can automatically generate a more situation-appropriate transmission text by an advanced technical means of arithmetic processing based on visit data, rather than mere random extraction. Therefore, the invention can automatically generate a transmission text that can provide thorough and personalized care and assistance, including in situations where a more situation-appropriate transmission text is required.

[0012] Therefore, the invention according to the first feature can provide an advanced technical means for reducing the burden on the service provider in charge regarding obtaining the specific office addition so as to be able to provide thorough and personalized care and assistance without omission.

[0013] The invention according to the second feature of the present invention is the invention according to the first feature, wherein the instruction generation unit automatically generates the sentence by arithmetic processing including a procedure of calculating the weight of the sentence candidates for a group of sentence candidates including a plurality of sentence candidates based on the access data and extracting a sentence candidate from the group of sentence candidates based on the weight, and provides a support device.

[0014] According to the invention, the instruction generation unit automatically generates the transmission sentence by a more advanced arithmetic process of extracting a sentence candidate from the group of sentence candidates based on the weight calculated based on the access data. Thereby, the invention can automatically generate the transmission sentence in accordance with the priority indicated by the weight. Therefore, the invention can automatically generate a transmission sentence that can provide meticulous care and assistance without omission, including situations where a transmission sentence according to the priority is required.

[0015] Therefore, the invention according to the second feature can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business office addition so as to be able to provide meticulous care and assistance without omission.

[0016] The invention according to the third feature of the present invention is the invention according to the second feature, wherein the instruction generation unit automatically generates the sentence by arithmetic processing including a procedure of extracting a sentence candidate from the group of sentence candidates using a probabilistic method based on the weight, and provides a support device.

[0017] In the automatic generation of the transmission sentence, in addition to generating a sentence suitable for the situation, it is also important that the same or similar sentences are not continuously and automatically generated. This is because if such automatic generation is performed, the service provided by the person in charge of the actual work will be biased.

[0018] According to the invention, the instruction generation unit automatically generates a transmission text by means of a more advanced arithmetic process of extracting a text candidate from a group of text candidates using a probabilistic method based on weights calculated based on access data. Thereby, the invention can achieve both automatically generating a transmission text in accordance with the priorities indicated by the weights and suppressing the continuous automatic generation of the same transmission text. Therefore, the invention can automatically generate a transmission text that can provide thorough and considerate care and assistance without omission, including cases where diversity of the transmission text is required.

[0019] Therefore, the invention according to the third feature can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business office addition so as to be able to provide thorough and considerate care and assistance without omission.

[0020] The invention according to the fourth feature of the present invention is the invention according to the second or third feature, and the instruction generation unit automatically generates the text by means of an arithmetic process including a procedure for calculating the weights of the text candidates based on the access data and a learning model for a group of text candidates including a plurality of text candidates, and provides a support device.

[0021] In order to perform extraction in accordance with the priorities at the home care visit site in extraction using weights, it is preferable that the weights be assigned in accordance with the priorities. However, due to various circumstances related to home care visits, the weights initially assigned to the access data may not match the priorities. Examples of the various circumstances include progress in home care visits, changes in the circumstances surrounding home care visits, changes at the site, changes in the persons in charge of the actual work, changes in service users, and the difficulty of predicting the actual state of the service before the start of the home care service.

[0022] According to the invention, the instruction generation unit automatically generates a transmission text by performing more advanced arithmetic processing of calculating the weights of text candidates based on the access data and the learning model, and extracting text candidates based on the weights. As a result, the support device can continue to calculate appropriate weights by updating the learning model so that the weights match the priorities. That is, the invention can realize automatic generation of a transmission text that can provide meticulous care and assistance without omission, including cases where the weights initially assigned to the access data do not match the priorities.

[0023] Therefore, the invention according to the fourth feature can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business establishment addition so as to be able to provide meticulous care and assistance without omission.

[0024] The invention according to the fifth feature of the present invention is an invention according to any one of the first to fourth features, and provides a support device in which the instruction generation unit automatically generates the text by arithmetic processing including a procedure of inputting a prompt including the access data to a generation AI and causing the generation AI to generate a response in natural language.

[0025] Visits related to home care may have various individual circumstances. If an attempt is made to prepare in advance a group of text candidates corresponding to a wide variety of individual circumstances, a great deal of labor and time are required for the consideration and examination thereof.

[0026] According to the invention, the instruction generation unit inputs a prompt including the access data to a generation AI and causes the generation AI to generate a transmission text in natural language. As a result, the invention can flexibly generate a transmission text in natural language corresponding to the individual circumstances of the visit without preparing in advance a group of text candidates corresponding to a wide variety of individual circumstances. Therefore, the invention can automatically generate a transmission text that can provide meticulous care and assistance without omission regardless of the presence or absence of foresight of individual circumstances.

[0027] Therefore, the invention according to the fifth feature can provide an advanced technical means for reducing the burden on the service provider who obtains the specific business establishment addition so that intimate care and assistance can be provided without omission.

[0028] The invention according to the sixth feature of the present invention is the invention according to the fifth feature, and further includes a machine learning unit that performs machine learning of the generation AI using the learning data including the above text, and provides a support device.

[0029] The generation AI can generate natural language in which findings indicated by general natural language sentences are reflected. On the other hand, it is difficult for the generation AI to generate natural language in which findings not included in the learning data are reflected. Therefore, it is expected that the generation AI cannot generate a transmission text in which findings accumulated through the continuous provision of home care services are reflected.

[0030] According to the invention, the machine learning unit performs machine learning of the generation AI using the learning data including the transmission text. As a result, the invention can automatically generate a transmission text in which various findings exemplified by the findings accumulated through the continuous provision of home care services are reflected. Therefore, the invention can automatically generate a transmission text that enables intimate care and assistance without omission, including cases where various findings exemplified by the accumulated findings are required.

[0031] Therefore, the invention according to the sixth feature can provide an advanced technical means for reducing the burden on the service provider who obtains the specific business establishment addition so that intimate care and assistance can be provided without omission.

Effects of the Invention

[0032] The present invention can provide an advanced technical means for reducing the burden on the service provider who obtains the specific business establishment addition so that intimate care and assistance can be provided without omission.

Brief Description of the Drawings

[0033] [Figure 1]Figure 1 is a block diagram showing an example of the hardware and software configuration of system S according to an embodiment of the present invention. [Figure 2] Figure 2 shows an example of a text candidate group table 131. [Figure 3] Figure 3 is a continuation of the previous figure. [Figure 4] Figure 4 is a continuation of the previous figure. [Figure 5] Figure 5 is a continuation of the previous figure. [Figure 6] Figure 6 shows an example of schedule table 132. [Figure 7] Figure 7 is a main flowchart showing an example of a preferred flow of support processing performed by support device 1. [Figure 8] Figure 8 is a continuation of the previous figure. [Figure 9] Figure 9 is a continuation of the previous figure. [Figure 10] Figure 10 shows an example of the display of the schedule area R1. [Figure 11] Figure 11 shows an example of the display of the visitor information area R2. [Figure 12] Figure 12 is a flowchart showing an example of a preferred flow of support processing performed by the support device 1 of the second embodiment. [Figure 13] Figure 13 is a continuation of the previous figure. [Figure 14] Figure 14 is a continuation of the previous figure. [Figure 15] Figure 15 is a continuation of the previous figure. [Figure 16] Figure 16 shows an example of a weighted, deterministic extraction procedure among the automatic text generation processes performed in step S109 of Figure 13. [Figure 17] Figure 17 shows an example of a weighted, probabilistic extraction procedure used in automatic text generation processing. [Figure 18] Figure 18 shows an example of a procedure using AI for automatic text generation. [Modes for carrying out the invention]

[0034] Firstly, although the following disclosures, figures, and / or claims are described either individually or in combination with one or more other aspects, the subject matter of the immediate disclosure is not intended to be limited in that way. That is, the immediate disclosures, figures, and claims are intended to encompass the various aspects described herein, either individually or in one or more combinations with each other. For example, even if the immediate disclosure describes and illustrates the first, second, and third embodiments in such a way that the first embodiment is described and illustrated particularly in relation to the second embodiment, or the second embodiment is described and illustrated only in relation to the third embodiment, the immediate disclosures and illustrations are not limited in that way and may include only the first embodiment, only the second embodiment, only the third embodiment, or one or more combinations of the first, second, and / or third embodiments, such as the first and second embodiments, the first and third embodiments, the second and third embodiments, or the first, second, and third embodiments.

[0035] In this text, the phrase "or" is used to mean a "non-exclusive" arrangement unless explicitly specified otherwise. For example, "item x is A or B" means either (1) item x is either A or B, or (2) item x is both A and B. In other words, the word "or" is not used to define an "exclusive" arrangement.

[0036] Furthermore, when the phrases "contain at least one" or "contain at least one of the following" are used in the text, they mean that the system or element contains one or more of the elements listed after the phrase. For example, if there are three types of elements, from element 1 to element 3, the phrases "contain at least one" or "contain at least one of the following" are interpreted as any of the following structural arrangements: a device containing element 1, a device containing element 2, a device containing element 3, a device containing element 1 and element 2, a device containing element 1 and element 3, a device containing element 2 and element 3, or a device containing element 1, element 2, and element 3.

[0037] The same interpretation is intended when the phrase "used in at least one of the following" is used in the text. Furthermore, "and / or" as used in the text is used as a linguistic conjunction to indicate that one or more of the listed elements or conditions are included or occur. For example, a device containing the first element, the second element, and / or the third element is interpreted as any of the following structural arrangements: a device containing the first element, a device containing the second element, a device containing the third element, a device containing the first and second elements, a device containing the first and third elements, a device containing the second and third elements, or a device containing the first, second, and third elements.

[0038] Furthermore, the use of the phrase "and / or" in this text signifies a "non-exclusive" arrangement, as stipulated in the Japanese Industrial Standard (JIS) "Format and Preparation Method of Standards Documents JIS Z 8301".

[0039] Hereinafter, an example of an embodiment of the present invention will be described in detail with reference to the drawings.

[0040] <System S> Figure 1 is a block diagram showing an example of the hardware and software configuration of system S according to an embodiment of the present invention.

[0041] System S comprises at least one terminal T and a support device (support device 1) of a home care service provider, which is configured to communicate with each other via a network N.

[0042] [Support device 1] The support device for home care service providers (support device 1) according to this embodiment comprises a control unit 11, a storage unit 13, a communication unit 14, a screen 15, and an input unit 16. By executing the program of this embodiment, the support device 1 performs support processing for home care service providers, including processing to automatically generate text to be communicated by the service provider manager to home care workers, etc.

[0043] [Home care workers, etc.] In this embodiment, "home care workers, etc." is used as a general term for practical personnel who visit the user's home and provide care services to the service user, as exemplified by home care workers. The service manager responsible for home care services is required to transmit user information to home care workers, etc. in writing or by other means, and to have home care workers, etc. report it.

[0044] Hereafter, home care workers and other personnel responsible for home care services will be simply referred to as "person in charge." Similarly, users of home care services may be simply referred to as "users." Service managers may be simply referred to as "service managers."

[0045] [Control Unit 11] The control unit 11 includes a Central Processing Unit (CPU), Random Access Memory (RAM), and Read Only Memory (ROM), among other things.

[0046] The control unit 11 cooperates with at least one of the storage unit 13 and the communication unit 14 as needed. The control unit 11 then implements the display command unit 111 and the instruction generation unit 112, which are software components of the program of this embodiment executed by the support device 1.

[0047] The support processing performed by the program of this embodiment will be described later with reference to Figures 7 to 9.

[0048] [Storage section 13] The storage unit 13 is a device on which data and / or files are stored, and has a storage unit that stores data non-temporarily using a hard disk, semiconductor memory, recording medium, and memory card, etc. The storage unit 13 stores programs executed by a microcomputer, a document candidate group table 131, a schedule table 132, the correspondence between working hours and wages for each person in charge, the maximum working hours for each person in charge, and notes related to each person in charge.

[0049] Preferably, the memory unit 13 stores a set of data according to this embodiment, such as macros as programs executed by a microcomputer, a text candidate group table 131, and a schedule table 132, in the form of a spreadsheet software file.

[0050] In the form of a spreadsheet software file, the storage unit 13 stores a spreadsheet software file that includes data such as a text candidate group data (text candidate group table 131) containing multiple text candidates used for automatically generating texts that service providers responsible for home care services communicate to home care workers, etc., and a macro that causes a computer (support device 1) to execute the support processing of this embodiment. The spreadsheet software file is not particularly limited as long as it is capable of executing macros, and may be, for example, a Microsoft Excel® sheet.

[0051] (Table of candidate sentences 131) Table 131 of the text candidate group stores text candidates that service providers responsible for home care services communicate to home care workers, etc.

[0052] In order to appropriately prepare sentence candidates so that sentences that meet the requirements for specific workplace addition and can provide thorough and meticulous care and assistance are automatically generated, it is preferable that the sentence candidate group table 131 stores a group of sentence candidates including a group of sentence candidates for physical condition confirmation, a group of sentence candidates for mental condition confirmation, a group of sentence candidates for daily life confirmation, and a group of sentence candidates for safety confirmation.

[0053] The number of sentence candidates stored in the sentence candidate group table 131 is preferably in the range of 48 to 80. This enables the coverage of the information necessary to meet the requirement of “(3) Transmission by documents of user information, reports from visiting caregivers, etc.” regarding specific workplace addition, while preventing bias in the automatically generated sentences that occurs when the number of sentence candidates is too large. As a result of trial and error by the inventors, it was determined that 64 is the optimal value. Therefore, the number of sentence candidates is more preferably in the range of 56 to 72. Preferred examples of the number and composition of sentence candidates belonging to each of the group of sentence candidates for physical condition confirmation, the group of sentence candidates for mental condition confirmation, the group of sentence candidates for daily life confirmation, and the group of sentence candidates for safety confirmation when the number of sentence candidates is 64 are as follows.

[0054] Figure 2 is an example of the sentence candidate group table 131. Figures 3, 4, and 5 are consecutive figures from the previous figure. Hereinafter, a preferred example of the sentence candidates included in the sentence candidate group table 131 will be described using Figures 2 to 5. In Figures 2 to 5, for the sake of explanation, each sentence candidate is assigned a serial number and a type, but the assignment of the serial number and type to each sentence candidate is not an essential requirement for the sentence candidate group table 131.

[0055] ((Group of Sentence Candidates for Physical Condition Confirmation)) Figure 2 shows 24 sentence candidates belonging to the group of sentence candidates for physical condition confirmation. Changes in physical condition due to aging, etc. (decline in physical ability, poor physical condition, decline in the five senses, etc.) are important confirmation items in nursing care services. These sentence candidates contribute to enabling visiting caregivers, etc. to accurately grasp the physical condition of the user and provide thorough and meticulous care and assistance according to the physical condition without omission.

[0056] Eating habits, including proper nutrition, medication administration and elimination during meals, and hydration, are important aspects of caregiving. Of the 24 sample sentences, eight are designated as meal confirmation sentences: "Please check if oral care is being performed after meals," "Please check if there is any food spillage during eating and drinking," "Please check if there are any changes in appetite or if the amount of food left uneaten is increasing," "Please check if the user is having difficulty using spoons or chopsticks," "Please check the frequency of elimination," "Please check the amount of water the user is drinking," "Please check if sufficient hydration is being provided," and "Please report any changes in medication status." These meal confirmation sentences encourage home care workers and others to perform important checks regarding meals, which are crucial for assessing the physical condition of service users, and contribute to ensuring that attentive care and assistance related to meals are provided without fail.

[0057] The user's physical condition, as determined by interviewing them about their subjective symptoms or checking their objective symptoms, is an important aspect of caregiving. Of the 24 candidate sentences, eight sentences—"Please check if there are any limitations or pain in the movement of their limbs when they move," "Please check if they experience shortness of breath or irregular breathing when they move," "Please let me know if there are any changes in their physical condition," "Please check if there is any swelling in their hands or feet," "Please check if they feel any discomfort in their feet or hands," "Please check if they experience any pain or numbness," "Please check if the volume of the television or radio is appropriate," and "Please report any changes in the user's complaints or special requests"—are candidates for checking the user's physical condition. These candidates for checking the user's physical condition encourage home care workers and others to check the user's physical condition, which is an important aspect of assessing the user's physical state, and contribute to ensuring that attentive care and assistance are provided without fail, taking the user's physical condition into account.

[0058] The user's activity level, determined through subjective symptom interviews or objective symptom checks, is an important aspect of caregiving. Of the 24 candidate sentences, eight are designated as activity level confirmation sentences: "Please check if the user's daytime activity level has decreased," "Please check if the user is spending more time sitting or lying down," "Please check if the user is getting tired easily," "Please check if the user appears tired," "Please check if their movements have become slower," "Please check if they are staying in the same position in bed or chair for extended periods," "Please inform us of any changes in the amount of assistance required," and "Please check for any movements that require assistance." These activity level confirmation sentences encourage home care workers and others to check the activity level, which is crucial for assessing the physical condition of service users, and contribute to ensuring that attentive care and assistance are provided comprehensively, taking activity levels into account.

[0059] ((Suggested sentences for checking mental state)) Figure 3 shows 16 candidate sentences belonging to the group of candidate sentences for assessing mental state. Changes in mental state due to aging, etc. (decline in motivation and cognitive function, sleep irregularities, etc.) are important points to check in care services. These candidate sentences will contribute to enabling home care workers and others to understand the mental state of users and to provide compassionate care and assistance that is appropriate to that state without fail.

[0060] The motivation of service users is an important point to confirm in caregiving. Of the 16 candidate sentences, eight candidate sentences—"Please tell me about any changes in their motivation regarding ADL and IADL," "Please check on their recent appetite," "Please ask about their willingness to go out," "Please check if their mood or motivation has declined," "Please check if they are able to continue their hobbies and enjoyable activities," "Please check if they are feeling easily fatigued or experiencing a decline in muscle strength," "Please check if they are becoming reluctant to move their body," and "Please check if they are getting tired more easily"—are candidate sentences for confirming motivation. These candidate sentences for confirming motivation encourage home care workers and others to check on the maintenance of motivation, which is important in confirming the mental state of service users, and contribute to ensuring that compassionate care and assistance based on motivation are provided without fail.

[0061] Psychological health is an important aspect of caregiving. Of the 16 sample sentences, eight are designated as psychological health assessment sentences: "Please report any changes in mental state," "Please check on their recent mood and feelings," "Please check if they are maintaining their daily routines," "Please check if they are able to converse and respond smoothly," "Please tell us about any changes in hygiene and cleanliness," "Please check if they are maintaining physical cleanliness," "Please ask about their sleep quality," and "Please carefully observe the user's facial expressions." These psychological health assessment sentences encourage home care workers and others to observe the psychological health of service users, which is crucial for assessing their mental state, and contribute to ensuring that compassionate care and assistance that takes psychological health into account is provided comprehensively.

[0062] ((Sentence candidates for daily life confirmation)) Figure 4 shows 16 candidate sentences belonging to the group of candidate sentences for confirming daily life. Understanding the daily life of service users, which reflects their physical and mental condition and contributes to maintaining a healthy and safe life, is an important aspect of care services. These candidate sentences will help home care workers and others understand the daily lives of users and provide attentive care and assistance that is appropriate to their daily lives without fail.

[0063] Maintaining a healthy daily life is an important aspect of caregiving. Of the 16 sample sentences, eight—"Please check if the house is being kept tidy," "Please check if mail and documents are being sorted," "Please check if food in the refrigerator is being stored properly," "Please check if laundry is piling up," "Please check if the trash is being taken out," "Please check if there are any changes in the daily routine," "Please check if the face and hands are being washed," and "Please check if the inventory of daily necessities is being managed"—are considered sample sentences for checking the maintenance of daily life. These sample sentences encourage home care workers and others to check on the maintenance of daily life, which is important for confirming the daily life of service users, and contribute to ensuring that attentive care and assistance are provided without fail, taking into account the status of maintaining daily life.

[0064] Challenges in daily life are important points to confirm in caregiving. Of the 16 candidate sentences, eight candidate sentences—"Please report any changes since the last visit report," "Please report any changes regarding family or living environment," "Please inform us if there are any changes in the user's environment, especially family relationships," "Please report any changes in the situation since the last service provision," "Please check the brightness and lighting conditions inside the house," "Please check if there are any imbalances in their diet," "Please check if they are having any difficulties in their daily life," and "Please check if they have any wishes regarding going out"—are candidate sentences for confirming daily life challenges. These candidate sentences for confirming daily life challenges encourage home care workers to confirm important challenges in the daily life of service users, and contribute to ensuring that compassionate care and assistance are provided without fail, taking these challenges into account.

[0065] ((Sentence candidates for safety confirmation)) Figure 5 shows eight candidate sentences belonging to the group of candidate sentences for safety confirmation. The safety of service users is an important point to confirm in care services. These candidate sentences will contribute to ensuring that home care workers and others can understand the safety of users and provide attentive care and assistance so that they can live more safely without fail.

[0066] The following eight sentence candidates belong to the group of safety check sentence candidates: "Please check if they often forget things or have trouble finding their belongings," "Please check if they have difficulty chewing or swallowing during meals," "Please check if they experience any anxiety or difficulty when moving around," "Please check if they can regulate their body temperature on their own and if the room temperature is appropriate," "Please check if their medication is stored properly," "Please check if emergency contact information is in an easily accessible place," "Please check if electrical cords are getting in the way," and "Please tell us about any changes in the risk of falls (such as unsteadiness, tripping, lying down, or falling while sitting)."

[0067] (Schedule Table 132) Schedule table 132 stores the schedules of the assigned personnel. The schedule includes appointments related to each assigned personnel and their visits. Each appointment includes the time of the visit, the service users, and the service content. The schedule may also include the length of time the service is provided or any notes related to the visit.

[0068] When the data set according to this embodiment is stored in the form of a spreadsheet software file, it is preferable that the schedule table 132 is a table with time on the vertical axis and the person in charge on the horizontal axis. This allows the schedule table 132 itself, as displayed by the spreadsheet software, to be visually displayed as a schedule table, enabling users such as service managers to intuitively grasp the schedule. This form is suitable for horizontal writing, but a similar table with the person in charge on the vertical axis and time on the horizontal axis is suitable for vertical writing.

[0069] At this time, it is preferable that the schedule is in a mode in which cells indicating service users, cells indicating service provision time zones, cells indicating service contents, and cells indicating the length of time are arranged in order in the same column. Thereby, the schedule table 132 allows the program of the present embodiment to search for a row indicating a service user and obtain other data from a series of cells following the cell, and also allows the display of the schedule table 132 itself to clearly show a schedule including these groups of information.

[0070] For the convenience of the service management responsible person, it is preferable that the file of the spreadsheet software is configured such that the schedules for one week are stored in separate sheets for each day. Thereby, the service management responsible person can switch sheets for each day and copy files for each week, and can reduce the labor required for transmission, schedule management, etc. using the text related to the requirement of "(3) Transmission by documents of user information, reports from visiting caregivers, etc." among the institutional requirements related to the specific office addition.

[0071] FIG. 6 is an example of the schedule table 132. This example includes a column of time indicating the time zone from 7:00 am to 12:00 pm in 10-minute increments, a column indicating the schedule of person in charge A, a column indicating the schedule of person in charge B, and a column indicating the schedule of person in charge C.

[0072] The column of person in charge A does not contain data corresponding to the schedule. Thereby, the support device 1 can calculate the operation rate and wages assuming that person in charge A has no plan to visit service users from 7:00 am to 12:00 pm.

[0073] The column for Staff Member B, from the row for 9:30 AM to the row for 10:20 AM, contains cells in order: one indicating that the service user is "Hei○ Hatsū○", another indicating that the service type is the code "△Sei3", another indicating that the time slot is "9:30~10:30", another indicating that the service content is "cleaning and shopping", and another indicating that the duration is "60" minutes. As a result, the support device 1 can automatically generate text related to the visit, assuming that Staff Member C is scheduled to visit the home of "Hei○ Hatsū○" (type "△Sei3") for 60 minutes from 9:30 AM to 10:30 AM to provide the "cleaning and shopping" service, and can also calculate Staff Member B's utilization rate and wages.

[0074] In the column for Person in Charge C, from the row for 11:00 AM to the row for 12:00 PM, there are cells in order that indicate the service user is "Hisashi ○," the type is indicated by the code "△(Disabled) Student 4," the time period is "11:00~12:00," a blank cell indicating that the service content is undecided, and the duration is "60" minutes. As a result, the support device 1 can automatically generate text related to the visit, assuming that Person in Charge B will visit the home of "Hisashi ○" (type "△(Disabled) Student 4") for 60 minutes from 11:00 AM to 12:00 PM to provide separately determined services, and can also calculate Person in Charge C's utilization rate and wages.

[0075] [Communications Section 14] The specific configuration of the communication unit 14 is not particularly limited, as long as it is for connecting the support device 1 to the network N and performing communication. Examples of the communication unit 14 include a network card compatible with the Ethernet standard, a communication device compatible with wireless LAN, etc.

[0076] [Screen 15] Screen 15 is not particularly limited as long as it is configured to display the schedule area R1, visit information area R2, wage area, utilization rate area, confirmation area, etc. Screen 15 may be, for example, a liquid crystal display, an organic electroluminescent display, etc. Screen 15 may be configured integrally with the support device 1 (for example, a built-in display of a notebook computer), or it may be a separate display device.

[0077] [Input section 16] The input unit 16 is not particularly limited, as long as it is configured for the service manager to perform various inputs. The input unit 16 may include, for example, a keyboard, various pointing devices, etc. The input unit 16 may be configured integrally with the support device 1 (for example, the keyboard and pointing device provided in a notebook computer) or it may be configured separately.

[0078] [Server configuration] The support device 1 may be a server that does not have a screen 15 and an input unit 16. In this case, the support device 1 performs processes such as sending data to a service manager terminal configured separately from the support device 1 to display a schedule area R1, a visit information area R2, a wage area, an utilization rate area, a confirmation area, etc., and receiving data from the service manager terminal indicating input from the service manager, etc.

[0079] [Network N] The type of network N is not particularly limited, as long as it is used for communication by the support device 1, etc. Examples of network N include the internet, a mobile phone network, a wireless LAN, etc.

[0080] [Terminal T] Terminal T is used by home care workers, etc. Terminal T performs processes such as receiving and displaying text sent from support device 1, and sending reports regarding said text. The type of terminal T is a mobile device, such as a smartphone or tablet.

[0081] [Main flowchart for support processing] Figure 7 is a main flowchart showing an example of a preferred flow of support processing performed by the support device 1. Figures 8 and 9 are continuations of the previous figures, respectively. The following is a description of an example of a preferred flow of support processing performed by the support device 1, using Figures 7 to 9. In the example shown below, the support device 1 is described as having a screen 15 and an input unit 16, but those skilled in the art can similarly configure a service management terminal configured separately from the support device 1 for display and input based on the example below.

[0082] To achieve the entire process from schedule input to automatic text generation, the support process preferably includes a series of processes related to displaying a screen including the schedule area R1, and inputting the schedule via the schedule area R1. Steps S1 to S3 are an example of such a process.

[0083] This process may be implemented using the functions of spreadsheet software, or it may be implemented through the cooperation of macros provided in the spreadsheet software file with the spreadsheet software itself. By utilizing the functions of spreadsheet software, the cost of introducing the support device 1 of this embodiment can be further reduced. The spreadsheet software provides functions such as input, display, and data updating. The macros provide functions such as discrimination and data linking between entered cells and other cells.

[0084] [Step S1: Display the screen] The control unit 11 works in cooperation with the storage unit 13 and the screen 15, etc., to realize the display command unit 111. Then, the control unit 11 executes a process to display the screen 15, which includes a schedule area R1 for inputting and displaying the schedules of home care workers, etc. (schedule screen display step). The control unit 11 then moves the process to step S2.

[0085] [Step S2: Determine if the schedule has been entered] The control unit 11 works in cooperation with the storage unit 13 and the input unit 16 to perform a process to determine whether or not a schedule has been entered into the schedule area R1 (schedule input determination step). If the control unit 11 determines that a schedule has been entered, it moves the process to step S3; otherwise, it moves the process to step S6.

[0086] [Step S3: Update the schedule] The control unit 11 works in cooperation with the storage unit 13 to perform a process to update the schedule stored in the schedule table 132 based on the input related to the schedule input determination step (schedule update step). The control unit 11 then moves the process to step S4.

[0087] The support process preferably includes a series of processes related to the calculation and display of wages and utilization rates via the schedule area R1. This reduces the burden on the service provider responsible for these calculations and supports the service provider responsible for providing thorough and attentive care and assistance through efficient schedule management that takes wages and utilization rates into consideration. Steps S4 to S5 are an example of such a process.

[0088] This process may be implemented using the functions of spreadsheet software, or it may be implemented through the cooperation of macros provided in the spreadsheet software file with the spreadsheet software itself. By utilizing the functions of spreadsheet software, the cost of introducing the support device 1 of this embodiment can be further reduced. The spreadsheet software provides functions such as input, display, and data update. The macros provide functions such as calculation.

[0089] [Step S4: Calculate wages and utilization rate] The control unit 11, based on the display command unit 111, executes a process to calculate wages and utilization rates based on the above-described schedule (wage calculation step). The control unit 11 then moves the process to step S5.

[0090] [Step S5: Display wages and utilization rate] The control unit 11, based on the schedule described above, executes a process to display the wages and utilization rate calculated in the wage calculation step on the screen 15 using the display command unit 111 (wage display step). The control unit 11 then moves the process to step S6. This process is achieved, for example, by displaying the visit information area R2, which has a wage area and an utilization rate area, on the screen 15.

[0091] The support process includes a series of processes for automatically generating documents that the service provider responsible for home care services will communicate to home care workers, etc. Steps S6 to S11 are an example of this series of processes.

[0092] [Step S6: Repeat for all responsible parties] The control unit 11 works in cooperation with the memory unit 13 and the screen 15, etc., to realize the instruction generation unit 112. Then, the control unit 11, using the instruction generation unit 112, executes a process that repeats the process from step S7 to step S11 for all personnel (loop start step). The control unit 11 then moves the process to step S7.

[0093] The loop in the loop start step may be implemented by iterative processing, or by a pre-expanded loop process that is expanded in advance according to the number of people in charge. When support processing is implemented by a macro included in a spreadsheet file, it is preferable that the loop be implemented by a pre-expanded loop process in order to prevent an increase in processing load due to a macro that displays information according to the number of loops.

[0094] [Step S7: Determine if it is a holiday] The control unit 11, using the instruction generation unit 112, performs a process to determine whether the person being processed in the current loop iteration is on leave or not (leave determination step). If the control unit 11 determines that the person is on leave, it moves the process to step S11; otherwise, it moves the process to step S8.

[0095] [Step S8: Determine if there are any undisclosed visits] The control unit 11 executes, through the instruction generation unit 112, a process of determining whether there is an un-displayed visit among the visits assigned to the person in charge who is the processing target in the current iteration of the loop process (un-displayed visit determination step). If the control unit 11 determines that there is, it moves the process to step S9, and if not, it moves the process to step S11.

[0096] [Step S9: Automatically generate text] The control unit 11 executes, through the instruction generation unit 112, a process of automatically generating text that the service provider responsible for the visit care service conveys to the visit care worker or the like (text automatic generation step). The control unit 11 moves the process to step S10. In this step, the instruction generation unit 112 automatically generates the above-mentioned text by randomly extracting from a text candidate group (text candidate group table 131) including a plurality of text candidates.

[0097] In order to prevent bias in the automatically generated text and to suppress the calculation processing cost, in this step, the instruction generation unit 112 is preferably configured to randomly extract text candidates without weighting the text candidates.

[0098] In order to prevent the service provider from spending effort on text selection due to excessive information provision and to suppress the calculation processing cost, in this step, the instruction generation unit 112 is preferably configured to automatically generate one text for each visit.

[0099] In this step, the instruction generation unit 112 is preferably configured to randomly extract from a text candidate group having a text candidate group for physical state confirmation, a text candidate group for mental state confirmation, a text candidate group for daily life confirmation, and a text candidate group for safety confirmation. Thereby, the support device 1 can generate text that meets the requirements related to the specific business office addition and can provide intimate care and assistance without omission by using a text candidate group with an appropriate configuration prepared in advance, and can support more appropriately performing the transmission for meeting the above requirements.

[0100] When the support process is implemented by a macro in spreadsheet software, it is preferable that the support device 1 automatically generates text each time the display of the spreadsheet software sheet is updated. This allows the support device 1 to provide an operating system that enables automatic text generation through a simple procedure via the standard operation of the spreadsheet software, which is updating the display of the sheet.

[0101] [Step S10: Display text] The control unit 11, acting on the display command unit 111, executes a process to display the text generated in the automatic text generation step on the screen 15 (display command step). The control unit 11 then moves the process to step S8.

[0102] In the display command step, the display command unit 111 displays visit information on the screen 15, which includes at least the identification information of each home care worker and service user receiving home care services, and text automatically generated by the instruction generation unit 112.

[0103] Furthermore, it is preferable that the display command unit 111 displays visit information on screen 15, which further includes transmission check information indicating the transmission status of the above-mentioned text and report check information indicating the reporting status corresponding to the above-mentioned text. This display is achieved, for example, by displaying a visit information area R2 having a confirmation area on screen 15.

[0104] [Step S11: Determine whether to continue the loop] The control unit 11, using the instruction generation unit 112, executes a process to determine whether to continue the loop related to the loop start step (loop continuation determination step). If the control unit 11 determines to continue, it changes the person to be processed in the loop repetition to a person who has not yet been processed and moves the process to step S7; otherwise, it moves the process to step S12. In the loop continuation determination step, the instruction generation unit 112 determines to continue if there is a person who has not yet been processed.

[0105] In the support process, if the support process includes displaying visit information on screen 15, which further includes transmission check information indicating the transmission status of the above-mentioned document and report check information indicating the reporting status corresponding to the above-mentioned document, it is preferable that the support process includes a series of processes to update the transmission status and report status in accordance with the input. This allows the support device 1 to provide screen 15 that allows users to easily confirm whether the requirements of "(3) transmission of user information by document, etc., and reports from home care workers, etc." have been met, without increasing the burden on the service provider.

[0106] The process of updating the transmission status, etc., may be implemented using the functions of spreadsheet software, or it may be implemented through the cooperation of macros provided in the spreadsheet software file and the spreadsheet software itself. By utilizing the functions of spreadsheet software, the cost of introducing the support device 1 of this embodiment can be further reduced. The spreadsheet software provides functions such as input, display, and data update. The spreadsheet software provides functions such as input, display, and data update. Macros provide functions such as data linking.

[0107] [Step S12: Determine if the transmission status has been entered] The control unit 11 performs a process to determine whether the transmission status of the above-mentioned text has been input by the display command unit 111 (transmission status input determination step). If the control unit 11 determines that the text has been input, it moves the process to step S13; otherwise, it moves the process to step S14.

[0108] [Step S13: Update communication status] The control unit 11, using the display command unit 111, executes a process to update the transmission status based on the above-mentioned input (transmission status update step). The control unit 11 then moves the process to step S14.

[0109] [Step S14: Determine if the report status has been entered] The control unit 11 performs a process to determine whether the report status corresponding to the above text has been entered, as indicated by the display command unit 111 (report status input determination step). If the control unit 11 determines that the report status has been entered, it moves the process to step S15; otherwise, it returns the process to step S1 and repeats the process from step S1 to step S15.

[0110] [Step S15: Update reporting status] The control unit 11, using the display command unit 111, executes a process to update the report status based on the above-mentioned input (report status update step). The control unit 11 returns the process to step S1 and repeats the process from step S1 to step S15.

[0111] [Effects of support processing] By executing the above-described support processing using the program of this embodiment, the support device 1 automatically generates (step S9) a document to be transmitted by the service provision manager to the home care workers, etc., regarding "(3) Transmission of user information by document, etc., and reports from home care workers, etc." among the organizational requirements for the specific business office addition, and displays it on the screen (step S10), thereby reducing the burden on the service provision manager in drafting the document. Herein, "home care workers, etc." is a general term for the practical staff who visit the user's home and provide care services to the service user, as exemplified by the home care worker.

[0112] Random sampling does not require sophisticated algorithms for selecting text candidates or the complex data management associated with them, and can be achieved with simple random selection computation. The instruction generation unit 112 employs a method in the automatic generation of the above text that involves random selection from a text candidate group (text candidate group table 131) containing multiple text candidates, which requires a large amount of work to create an appropriate text candidate group, but minimizes the computational processing cost related to automatic generation (step S9).

[0113] As a result, the support device 1 not only reduces the burden on the service provider responsible for drafting the above-mentioned documents, enabling comprehensive and attentive care and assistance, but also reduces the hardware performance required for the support device by simplifying the calculation process, thereby reducing the financial burden on the business operator regarding the introduction and operation of the support device 1.

[0114] Regarding this automated generation, random sampling with weighting of sentence candidates is computationally more expensive than methods that randomly select sentence candidates without weighting. Furthermore, random sampling with weighting may preferentially select certain sentence candidates, raising concerns that it could bias the automatically generated sentences and prevent the provision of comprehensive and empathetic care and assistance instructions.

[0115] The above-described automatic generation can be configured to employ a method of randomly selecting text candidates without weighting (step S9). This configuration solves the above-described problems, reduces the burden on service providers, enables comprehensive and attentive care and assistance, and further reduces the financial burden on businesses introducing the support device 1 that performs automatic generation by reducing computational processing costs.

[0116] Furthermore, the above-mentioned automatic generation can be configured to perform a minimum amount of automatic generation for each visit (step S9). This configuration prevents service providers from having to expend effort selecting and filtering text due to excessive information provision, thereby hindering their ability to provide attentive care and assistance, while further reducing the burden on businesses introducing the support device 1 that performs automatic generation by reducing computational processing costs.

[0117] The configuration in which a loop process is performed for all staff members (steps S6 to S11), and in this loop process, if it is determined that the staff member is not on leave (step S7), and further determined that there are undisclosed visits (step S8), then text is automatically generated (step S9), is an example of a configuration that performs minimal automatic generation for each visit. As a result, the support device 1 can automatically generate text so that each staff member can provide thorough and attentive care and assistance for all visits.

[0118] The above automatic generation can be configured to use a group of candidate sentences related to random extraction (

[0118] ), which has a group of candidate sentences for physical condition confirmation, a group of candidate sentences for mental condition confirmation, a group of candidate sentences for daily life confirmation, and a group of candidate sentences for safety confirmation (candidate sentence group table 131). Since the candidate sentence group table 131 in this configuration has appropriate content prepared in advance, the instruction generation unit 112 generates sentences that meet the requirements related to the specific workplace addition and can provide intimate care and assistance without omission. As a result, the support device 1 can support the more appropriate transmission of the above requirements.

[0119] Therefore, this embodiment can provide a technical means for reducing the burdens of both the service provider responsible for providing services related to obtaining the specific workplace addition and the business operator so that intimate care and assistance can be provided without omission.

[0120] By the way, in order to confirm whether intimate care and assistance are provided without omission, the service provider responsible for providing services is required to accurately grasp the transmission status and reporting status of the sentences. The support process can be configured to add a concise display including necessary and sufficient information related to the transmission status and reporting status, namely transmission check information and reporting check information (step S10, steps S12 to S15). As a result, a screen can be provided that can easily confirm whether the requirements of "(3) Transmission of user information by documents, etc., reports from visiting caregivers, etc." are met without increasing the burden on the business operator introducing the support device 1.

[0121] As a result, the support device 1 seamlessly supports a series of procedures from the generation to the transmission and reporting of sentences, prevents omission of information transmission and reporting, and enables the service management responsible person to grasp the situation in real time. Therefore, the support device 1 can support significantly reducing the burdens of the service provider responsible for providing services and the business operator in a series of procedures for efficiently meeting the requirements of the specific workplace addition.

[0122] In addition, the support process can be configured to update the schedule according to the input (from step S1 to step S3), calculate the wage and the operation rate based on the schedule (step S4), and display them (step S5). Thereby, the support device 1 reduces the burden on the service provider responsible for these calculations, and supports the service provider to provide meticulous care and assistance without omission through efficient schedule management considering the wage and the operation rate.

[0123] Thereby, the support device 1 links the creation of the visit instruction and the schedule management, and supports both the site where the service management responsible person conveys the text and the management, so that the optimization of the instruction content and the optimization of the visit plan can be realized simultaneously. Therefore, the support device 1 can support the efficient achievement of the specific business office addition requirements and reduce the management burden.

[0124] The support process of the present embodiment can be executed by a file of spreadsheet software including a macro and data of a sentence candidate group to be performed by an electronic computer (storage unit 13).

[0125] Thereby, the service provider and the contractor can make the existing electronic computer function as the support device 1 of the present embodiment by opening the file with spreadsheet software. That is, the service provider and the contractor can realize the support device 1 of the present embodiment described above without high equipment investment or complicated procedures associated with the introduction of a dedicated device, installation of new software, etc.

[0126] Thereby, the file can utilize the flexibility and real-time processing of spreadsheet software to provide integrated support for visit information management and instruction generation. The execution of the support process by a file of spreadsheet software reduces the introduction cost and operation burden of the support device 1 and contributes to efficiently achieving the requirements of the specific business office addition. Thereby, the adaptability to small-scale contractors who have difficulty bearing the financial burden associated with the introduction of a dedicated device is improved.

[0127] <Usage Example> The following describes an example of using the support device 1 and system S of this embodiment. The following describes an example in which the support device 1 and system S of this embodiment are implemented using a spreadsheet software file stored in the memory unit 13.

[0128] [Preparing the files] The service manager prepares a spreadsheet file containing macros and various data to execute the support processing of this embodiment. If the file contains a week's worth of schedules, the service manager prepares the file by duplicating it weekly and renaming the file. The service manager also adds, deletes, or changes personnel as needed.

[0129] [Entering the schedule] By opening the file using spreadsheet software, the support device 1 displays the schedule area R1.

[0130] The service manager enters the schedule for each employee into the schedule area R1. Figure 10 shows an example of the display in the schedule area R1. This example shows schedules for employees A, B, and C that are the same as those shown in schedule table 132. At this time, the service manager can adjust the schedule by referring to the utilization rate and wages, which are updated each time a schedule is entered.

[0131] [Communication of user information through documents, etc.] Support device 1 displays the visit information area R2. Figure 11 shows an example of the display of the visit information area R2. In this example, for staff member C, it is shown that they will be visiting "Hisashi〇" (Disabled Student 4) and the automatically generated text for this visit is "Please check if they are getting tired more easily." and for "Misao Naga〇" (Disabled Student 3) and the automatically generated text for this visit is "Please check if the volume of the TV or radio is appropriate." In addition, in this example, for staff member D, it is shown that they will be visiting "Yoshio Ki〇" (Disabled Student 2, Life 1) and the automatically generated text for this visit is "Please check if there are any food spills during meals." and for "Misao Naga〇" (Disabled Student 1, Life 2) and the automatically generated text for this visit is "Please check if there are any changes in their daily routine." Furthermore, in this example, for staff member E, it is shown that they are on leave by changing the column displaying the staff member's name.

[0132] The service manager takes a screenshot of the updated visit information area R2 display when entering the schedule and sends the screenshot to the person in charge via an appropriate means of communication (for example, chat using LINE®). The service manager then updates the field for recording the status of message transmission. The service manager also refers to the field for recording the status of message transmission and continues to send screenshots to the person in charge who has not yet received the message.

[0133] [Service provision and reporting by the person in charge] The staff member will review the visit information via the screenshot and provide the service according to the automatically generated text. The staff member will then send a report corresponding to the text to the service manager via the aforementioned contact method. The service manager will update the field for recording the report's transmission status. The service manager will also refer to the field for recording the report's transmission status and prompt staff members who have not yet submitted a report to do so.

[0134] <Second Embodiment> A second embodiment of the present invention is a method for automatically generating documents that service managers communicate to service providers, which is not limited to random sampling. Herein, "service providers" refers to service providers who visit the homes of service users and provide care services to service users, as exemplified by home care workers. Furthermore, "documents that service managers communicate to service providers" will also be referred to as "communication documents" below.

[0135] The hardware and software components in the second embodiment may be the same as those in the first embodiment, except for the differences listed below.

[0136] The automatic generation in the second embodiment may include methods for automatically generating text optimized by various techniques. Examples of such methods include the following methods or combinations thereof: (1) Rule-based weighted instruction generation based on situational data or reported information; (2) Dynamic text generation utilizing generative AI exemplified by large-scale language models (LLMs); (3) Learning-type response that individually optimizes instruction sentences; (4) Configurations that use multiple methods in combination; (5) Adjustments after automatic generation. Note that the methods listed above are merely examples, and other natural language processing technologies, predictive models, statistical analysis, context-dependent models, and combinations thereof for automatic generation methods are also included in this embodiment.

[0137] First, I will briefly explain each of the example methods.

[0138] [(1) Rule-based weighted instruction generation] In rule-based weighted instruction generation, the support device 1 weights each document to be extracted in a rule-based manner based on parameters derived from situation data specified by the service manager and / or report information provided by the practitioner. Then, as part of the automatic generation of communication documents, the support device 1 selects a communication document from each document to be extracted, either deterministically or probabilistically, based on the weighting.

[0139] [(1a) Definitive selection of communication texts based on weighting] In weighted definitive message selection, the support device 1 calculates a final weight by superimposing weights based on multiple parameters. Then, the support device 1 selects a message based on the final weight. As a result, this method reliably generates the optimal message for the situation indicated by the parameters.

[0140] [(1b) Probabilistic message selection based on weighting] In weighted probabilistic message selection, the support device 1 probabilistically generates message texts based on weights derived from one or more parameters. This prevents the same message from being generated repeatedly in situations where there is no significant change in the conditions indicated by the parameters.

[0141] [Regarding weighting] An example of weighting based on parameters related to the user's situation is shown below. If a parameter indicating that the user has difficulty getting up is given, the support device 1 will use rule-based processing to assign a greater weight to appropriate sentences, such as sentences instructing to check whether the user can get up on their own, sentences instructing to assist the user in getting up, and sentences instructing to check whether the user has any injuries, than to other sentences.

[0142] Furthermore, in this case, the support device 1 uses rule-based processing to assign a smaller weight to inappropriate sentences, such as those exemplified by sentences instructing the user to get up on their own, compared to other sentences. This allows the support device 1 to encourage the automatic generation of appropriate sentences and suppress the automatic generation of inappropriate sentences.

[0143] An example of weighting based on parameters related to the reported information is shown below. If a parameter is given indicating that the cleaning status has recently been checked, the support device 1 will use rule-based processing to assign a greater weight to appropriate sentences, such as a sentence instructing to check medication status or a sentence instructing to check the refrigerator status, than to other sentences.

[0144] Furthermore, in this case, the support device 1 uses rule-based processing to assign a smaller weight to, for example, a sentence instructing the user to check the cleaning status, compared to other sentences. This allows the support device 1 to encourage the automatic generation of appropriate sentences and suppress the automatic generation of inappropriate sentences.

[0145] Procedures for superimposing weights based on multiple parameters include, for example, linear weighted averages and hierarchical weighting. Furthermore, when using weights, it is possible to select only sentences whose final weights exceed a certain threshold as candidates.

[0146] In definitive selection of communication texts, there may be a configuration that presents multiple communication texts in order of weight to allow selection. Furthermore, there may be a configuration that assigns negative weights to prevent the automatic generation of texts that are unsuitable for the situation indicated by the parameters.

[0147] (2) Dynamic text generation by generation AI In dynamic text generation using a generation AI, the support device 1 inputs instruction text and other data specified by the service manager to the generation AI, causing the generation AI to generate a communication text. Examples of data input to the generation AI include instruction text, fixed prompts, status data, report information, and various data stored in the schedule table 132.

[0148] This method generates linguistically natural communication text using a generative AI. Furthermore, if the generative AI is a large-scale language model, it is expected to generate more practical communication text that reflects insights derived from its training data.

[0149] In this method, it is preferable to limit the operation of the generating AI so that it generates text that is appropriate as a communication document. Means of achieving this limitation include including a fixed prompt in the input that indicates the requirements for text that is appropriate as a communication document, and having the generating AI undergo additional learning.

[0150] The instructions given to the generating AI may be selected from a template. This allows the service manager to provide instructions without having to manually enter them one by one. Furthermore, using a template is expected to ensure stable generation results. To facilitate the selection process for the service manager, it is preferable that the templates be categorized.

[0151] Furthermore, the instruction document may include both fixed instructions selected from a template and additional instructions entered by the service manager. This allows for both the reduction of effort and stability of the generated results provided by templates, and the flexibility of manual input.

[0152] Fixed prompts may include various specifications, such as style, tone, and length. This will result in the generation of more appropriate communication text.

[0153] The generation AI may be executed on support device 1, or on an external device different from support device 1. Executing the generation AI on support device 1 enables the generation of communication texts regardless of communication conditions. Executing the generation AI on an external device allows for a reduction in the computing power and procurement costs of support device 1 itself, while simultaneously enabling the generation of communication texts by the generation AI.

[0154] (3) Use of machine learning In utilizing machine learning, support device 1 performs machine learning on a learning model using data related to practitioners, users, or combinations of practitioners and users as training data. Then, support device 1 automatically generates communication text using this learning model. An example of a learning model is the following model.

[0155] [(3a) A learning model that performs weighting related to extraction] The learning model used for weighting during extraction is a learning model that contributes to determining the weight for each sentence to be extracted. In this method, the learning model is trained to perform such weighting using the training data described above.

[0156] This allows the method to enable the learning model to determine weights that are more appropriate for individual cases. The method then determines weights optimized for individual cases and generates communication text based on these weights.

[0157] (Reinforcement learning) Machine learning may include reinforcement learning using training data in which automatically generated communication texts are accompanied by report information or evaluations from service managers. This allows the support device 1 to perform appropriate machine learning while reducing the effort spent on preparing training data.

[0158] Furthermore, by using reinforcement learning, the support device 1 can continuously incorporate the actual usage results and evaluation information of the communication text into its learning process. As a result, the support device 1 can gradually improve its accuracy in the actual operating environment and continuously generate communication texts that are highly appropriate to the situation.

[0159] [(3b) Learning model for generating communication texts] The learning model for generating communication texts is a learning model that contributes to the dynamic generation of communication texts using generative AI. In this method, the training data described above is used to train the learning model to produce examples of communication texts suitable for individual cases.

[0160] This method enables the generative AI associated with the learning model to generate communication text that is more suitable for individual cases.

[0161] (Case-based reasoning) In machine learning related to a learning model for generating communication texts, it is preferable to use training data that associates communication texts with parameters indicating the situations in which the communication texts are appropriate, so that the generating AI can perform case-based reasoning.

[0162] By using a learning model that enables case-based reasoning, it becomes possible to generate practical and reliable communication texts that refer to specific past communication examples. Furthermore, generation based on known good examples contributes to suppressing misleading expressions and reducing errors in communication.

[0163] [(4) Configurations that use multiple methods in combination] The following describes an example of a configuration that combines several of the above-mentioned methods.

[0164] [(4a) Selection of Method] Support device 1 may provide a user interface that allows the service manager to select an automated generation method. This allows the service manager to, for example, select random sampling when there are no special circumstances, and to select automated generation using rule-based, generative AI, or learning models when there are special circumstances. This enables support device 1 to achieve automated generation that leverages the advantages of each method.

[0165] The support device 1 may include a mechanism for automatically selecting an automatic generation method. This allows the support device 1 to achieve automatic generation that leverages the advantages of each method without requiring the service manager to expend effort in making a selection.

[0166] The support device 1 may include a mechanism for indicating the priority of automatic generation methods. This allows the support device 1 to reduce the effort required for the service manager to consider and select an automatic generation method, and to achieve automatic generation that takes advantage of the strengths of each method.

[0167] [(4b) Hierarchical use of different methods] The support device 1 may have a configuration that applies multiple automatic generation methods hierarchically. One example of such a configuration is one in which candidate sentences are first narrowed down by rule-based weighting, and then a generation AI generates a communication text with improved naturalness of meaning based on the candidate sentences.

[0168] Another example of such a configuration is a stepwise collaborative process in which the extraction priority is adjusted by a learning model, and the remaining candidate sentences are reconstructed by a generation AI. This enables the support device 1 to efficiently generate highly accurate and contextually relevant communication texts.

[0169] [(5) Adjustments after automatic generation] The support device 1 preferably provides means for adjusting the automatically generated communication text. The following is an example of such means.

[0170] [(5a) Corrections after automatic generation] The support device 1 may provide a user interface for modifying automatically generated communication documents. This allows the service manager to modify the automatically generated communication documents. Furthermore, this enables the support device 1 to use the modified communication documents for machine learning to improve their accuracy.

[0171] [(5b) Review after automatic generation] Support device 1 may provide a user interface for reviewing automatically generated communication documents. This allows support device 1 to use the review results for machine learning to improve its accuracy. Furthermore, the service manager can control the process through the review to ensure that only appropriate communication documents are sent to the service provider.

[0172] [(5c) Accumulation and utilization of revision and review history] The support device 1 may be configured to store the revised or reviewed communication document in a memory unit, linked to the relevant situation. This allows the support device 1 to prioritize presenting the revised or reviewed document as a candidate when a similar situation occurs in the future.

[0173] Furthermore, if a large number of similar histories are accumulated, the support device 1 may update its learning model and weighting rules based on those histories. This allows the support device 1 to achieve continuous accuracy improvement.

[0174] [Details of the second embodiment] The following describes the details of the second embodiment. In the description of the second embodiment, the first embodiment described in this specification will also be referred to as the "first embodiment" in order to distinguish it from the first embodiment described herein. Furthermore, the "invention according to the second embodiment" will also be referred to simply as the "present invention." In addition, the "second embodiment" will also be referred to simply as the "this embodiment" below.

[0175] [System S] System S is the same as in the first embodiment, except that it may optionally include a server (not shown) that provides processing by the generating AI.

[0176] [Support device 1] Support device 1 is the same as in the first embodiment, except for the following points.

[0177] [Control Unit 11] The control unit 11 is the same as in the first embodiment, except that it can implement a machine learning unit 113 (not shown) as a software component.

[0178] [Storage section 13] The memory unit 13 is the same as in the first embodiment, except that it stores visit data, stores a learning model and / or a language model related to the generating AI as needed, and stores learning data and various data related to the learning data as needed.

[0179] (Visit data) Visit data refers to data relating to visits for home care services. For example, visit data may include situation data indicating the circumstances of the visit, and one or more pieces of report information related to the visit. By including situation data in the visit data, the support device 1 can achieve automatic generation that reflects the circumstances. By including report information in the visit data, the support device 1 can achieve automatic generation that reflects the report content.

[0180] The situation data may include, for example, one or more data points indicating the situation of the service user, the situation of the person providing the service, the temporal situation of the home care service, and the geographical situation of the home care service. This allows the support device 1 to automatically generate data that reflects each situation.

[0181] The data indicating the service user's situation may include, for example, one or more of the following: level of care required, presence or absence of dementia, physical condition, medical history, and lifestyle habits. This enables the support device 1 to automatically generate communication documents tailored to the care situation of the service user.

[0182] The data indicating the status of the person in charge of practical work may include, for example, one or more of the following: caregiving qualifications, work experience, past work history, work shifts, and the types of services that can be provided. This enables the support device 1 to automatically generate communication documents that are appropriate to the abilities and / or experience of the person in charge of practical work.

[0183] The data indicating the temporal circumstances of home care visits may include the scheduled date of the visit, the scheduled time of the visit, the duration of the visit, whether or not it is a consecutive visit, and one or more intervals between the scheduled visits. This enables the support device 1 to automatically generate communication documents that take into account the visit time slot and / or time constraints.

[0184] The data indicating the geographical circumstances of home care visits may include, for example, the address of the visit destination, the structure of the building at the visit destination, the surrounding traffic conditions, and one or more geographical distances to other visit destinations. This enables the support device 1 to automatically generate communication messages that take into account travel time and / or geographical constraints.

[0185] The report information is information that shows the content of the report from the person in charge of the work. The report information may also include data that further includes, for example, one or more pieces of information that identifies the reporter, information that indicates the date and / or time of the visit related to the report, and information that identifies the service user related to the report. By including information that identifies the reporter in the report information, the support device 1 can achieve automatic generation tailored to the reporter. By including information that indicates the date and / or time of the visit related to the report information, the support device 1 can achieve automatic generation tailored to the temporal circumstances of the visit. By including information that identifies the service user related to the report information, the support device 1 can achieve automatic generation tailored to the service user.

[0186] [Communications Section 14] The communication unit 14 is the same as in the first embodiment, except that it is configured to communicate with a server (not shown) that provides processing by the generation AI as needed.

[0187] [Network N] Network N is the same as in the first embodiment, except that it is configured to provide communication with a server (not shown) that provides processing by the Generating AI, if necessary.

[0188] [Server configuration] The support device 1 of the second embodiment is the same as the first embodiment in that it can be configured as a server and in the various additional processes performed in that case.

[0189] [Main flowchart for support processing] Figure 12 is a main flowchart showing an example of a preferred flow of support processing performed by the support device 1 of the second embodiment. Figures 13, 14, and 15 are continuations of the previous figures. The following is a description of an example of a preferred flow of support processing performed by the support device 1 of the second embodiment, using Figures 12 to 15.

[0190] [Step S101: Display the screen] [Step S102: Determine if a schedule has been entered] [Step S103: Update the schedule] [Step S104: Calculate wages and utilization rate] [Step S105: Display wages and utilization rate] Steps S101 to S105 are, in order, the same as steps S1 to S5 of the first embodiment. In the description of these steps, the step numbers will be read as the step numbers of the first embodiment plus 100.

[0191] The support process includes a series of processes for automatically generating documents that the service provider responsible for home care services will communicate to home care workers, etc. Steps S106 to S111 are an example of this series of processes.

[0192] [Step S6: Repeat for all responsible parties] [Step S7: Determine if it is a holiday] [Step S8: Determine if there are any undisclosed visits] Steps S106 to S108 are, in order, the same as steps S6 to S8 of the first embodiment. In the description of these steps, the step numbers will be read as the step numbers of the first embodiment plus 100.

[0193] [Step S109: Automatic text generation process] The control unit 11 executes an automatic text generation process (automatic text generation process step) in which the instruction generation unit 112 automatically generates text to be communicated by the service provider manager for home care services to the practical staff member, using visit data as input. The control unit 11 then moves the process to step S110.

[0194] Details of the automatic text generation process performed in the automatic text generation step will be described later in the "Flowchart of Automatic Text Generation Process" section.

[0195] [Step S110: Display text] [Step S111: Determine whether to continue the loop] [Step S112: Determine if the transmission status has been entered] [Step S113: Update communication status] [Step S114: Determine if the report status has been entered] [Step S115: Update reporting status] Steps S110 to S115 are, in order, the same as steps S10 to S15 of the first embodiment. Note that the step numbers in the description of these steps should be read as the step numbers of the first embodiment plus 100. Also, the statement equivalent to "The control unit 11 returns the process to step S1 and repeats the process from step S1 to step S15" should be read as "The control unit 11 moves the process to step S116."

[0196] The support process preferably includes procedures for evaluating and / or revising the document. Step S116 is an example of such a procedure. Evaluation is also referred to as review.

[0197] [Step S116: Provide evaluation correction UI] The control unit 11 executes a process to provide an evaluation and correction user interface (evaluation and correction UI) for the service manager to evaluate and / or correct the text automatically generated in step S109, as instructed by the display command unit 111 (evaluation and correction UI provision step). The control unit 11 then moves the process to step S117.

[0198] The evaluation correction UI preferably includes user interface elements for evaluating automatically generated text. These elements include, for example, one or more user interface elements exemplified by a button for giving a positive evaluation to text, a button for giving a negative evaluation to text, a button for assigning a score to text, a button for instructing text regeneration, and a window for entering evaluations. This allows the support device 1 to reflect the evaluations entered by the service manager via these user interface elements into the training data. If the button for instructing text regeneration is operated, the support device 1 may perform text regeneration and determine that a negative evaluation has been given to the text.

[0199] The evaluation correction UI preferably includes user interface elements for correcting automatically generated text. These elements include, for example, one or more user interface elements exemplified by an input window in which the automatically generated text is pre-entered, and a button for displaying the input window. This allows the support device 1 to reflect the text corrected by the service manager in the training data via these user interface elements. When text is corrected, the support device 1 may correct the text and also determine that a negative evaluation has been assigned to the text.

[0200] To facilitate transmission to the person in charge of the work, it is preferable that the evaluation and correction UI has a user interface element that sends the evaluated and / or corrected text, or the automatically generated text as is, to the person in charge of the work. In this case, if the user interface element is operated to send the automatically generated text as is to the person in charge of the work, the support device 1 may send the text and determine that the text has been given a positive evaluation.

[0201] The support device 1 may be configured to store the text after an operation using the evaluation / correction UI, i.e., the corrected or evaluated communication text, in the memory unit 13, linked to the relevant situation. This allows the support device 1 to prioritize presenting the corrected or evaluated text as a candidate when a similar situation occurs in the future.

[0202] Furthermore, if a large number of similar communication messages are accumulated in the history, the support device 1 may update its learning model and weighting rules based on that history. This allows the support device 1 to achieve continuous accuracy improvement.

[0203] The support process preferably includes a procedure for updating the learning data based on operations related to the evaluation and / or modification of a text. Steps S117 to S118 are an example of such a procedure.

[0204] [Step S117: Determine if the document has been evaluated or revised] The control unit 11 works in cooperation with at least the memory unit 13 to implement the machine learning unit 113. The control unit 11 then performs a process to determine whether the text automatically generated in step S109 has been evaluated or modified by the machine learning unit 113 (evaluation / modification determination step). If the control unit 11 determines that at least one of evaluation or modification has been performed, it moves the process to step S118; otherwise, it moves the process to step S101, and repeats the process from step S101 to step S119.

[0205] In this step, the support device 1 can perform a process to reflect the results of the evaluation or modification of the text into the training data.

[0206] [Step S118: Update training data] The control unit 11 performs a process to update the training data based on the operation determined by the machine learning unit 113 to have been performed in step S117 (training data update step). The control unit 11 then moves the process to step S119.

[0207] Examples of procedures for updating training data in the training data update step include adding data that associates automatically generated sentences with evaluations to the training data, and adding data that associates automatically generated sentences with revised sentences to the training data.

[0208] The data added to the training data may be data that further associates some or all of the visit data used for automatic generation. This allows the support device 1 to update the training data, including the context of automatic generation, to contribute to machine learning.

[0209] [Step S119: Run machine learning] The control unit 11 executes a machine learning process based on the updated training data in step S118, using the machine learning unit 113 (machine learning step). The control unit 11 then moves the process to step S101 and repeats the processes from step S101 to step S119.

[0210] To improve learning efficiency and resource efficiency, it is preferable that the support device 1 executes a machine learning step when the amount of training data updated since the previous machine learning session meets a given criterion. Alternatively, as another means of improving learning efficiency and resource efficiency, it is preferable that the support device 1 executes a machine learning step when a given time has elapsed since the previous machine learning session.

[0211] If the computational processing related to automatic text generation includes a procedure for calculating weights for a group of text candidates, which includes multiple text candidates, based on visit data and a learning model, it is preferable that the machine learning step performs machine learning on the learning model that calculates the weights based on the learning data. This allows the support device 1 to update the learning model so that the weights match the priorities in the practical work of home care.

[0212] In the machine learning step, the support device 1 may perform reinforcement learning using training data in which automatically generated communication texts are accompanied by reporting information and / or evaluations from the service manager. This allows the support device 1 to perform appropriate machine learning while reducing the effort spent on preparing training data. Furthermore, this enables the support device 1 to continuously reflect the actual usage results and evaluation information of the communication texts in its learning. By continuously reflecting this in its learning, the support device 1 can gradually improve its accuracy in the actual operating environment and continuously generate communication texts that are highly appropriate to the situation.

[0213] If the computational processing related to the automatic generation of text includes a procedure for inputting a prompt containing visit data into a generating AI and causing the generating AI to generate a natural language response, it is preferable that the machine learning step performs machine learning on the generating AI based on the training data. This enables the support device 1 to automatically generate communication text that reflects various insights exemplified by the knowledge accumulated through the continuous provision of home care services.

[0214] Machine learning for generative AI preferably uses training data that associates communication text with parameters indicating the situations in which the communication text is appropriate, so that the generative AI can perform case-based reasoning. This makes it possible to generate practical and reliable communication text by referring to specific past communication examples. Furthermore, generation based on known good examples also contributes to suppressing misleading expressions and reducing errors in responses.

[0215] [Flowchart for automatic text generation process] Figure 16 shows an example of a deterministic extraction procedure based on weighting, which is part of the automatic text generation process performed in step S109 of Figure 13. Figure 17 shows an example of a probabilistic extraction procedure based on weighting, which is part of the automatic text generation process. Figure 18 shows an example of a procedure using a generation AI, which is part of the automatic text generation process. The following is a description of a preferred flow example for various methods of the automatic text generation process performed in step S109 of Figure 13, using Figures 16 to 18. Note that these methods may be executed individually or in combination. Combinations of these methods will be described later.

[0216] Steps S201 to S204 are an example of a deterministic extraction procedure based on weighting. In this example, the computational process related to automatic text generation includes a procedure for calculating the weights of a group of text candidates, which includes multiple text candidates, based on visit data, and for extracting text candidates from the group of text candidates using a deterministic method based on said weighting.

[0217] [Step S201: Calculate weights] The control unit 11, using the instruction generation unit 112, performs a process to calculate the weight of candidate sentences based on the visit data (weighting calculation step). The control unit 11 then moves the process to step S202.

[0218] The process for calculating weights includes, for example, a procedure for weighting each sentence to be extracted based on visit data using a rule-based approach. This allows the support device 1 to perform weighting based on rules constructed in line with the objective of providing comprehensive and attentive care and assistance. To suppress the selection of inappropriate sentence candidates, the weighting may include negative weights.

[0219] In this case, it is preferable that the instruction generation unit 112 calculates the final weight by superimposing weights based on multiple parameters related to the visit data. This allows the support device 1 to reflect multiple parameters related to the visit data in the weight by a less computationally intensive process, namely weight superposition, without having to process complex rules that cover all possible combinations of multiple parameters. Examples of procedures for superimposing weights based on multiple parameters include linear weighted average and hierarchical analytical weighting.

[0220] (Use of a learning model) The process for calculating weights preferably includes a step of calculating the weights of a group of text candidates, which includes multiple text candidates, based on visit data and a learning model. This allows the support device 1 to continue calculating appropriate weights by updating the learning model so that the weights match the priority.

[0221] [Step S202: Extract sentence candidates using a definitive method] The control unit 11 performs a process to extract document candidates from the document candidate group using a deterministic method based on the weights calculated by the instruction generation unit 112 (deterministic extraction step). The control unit 11 then moves the process to step S203. Here, "document candidate group" refers to, for example, the document candidate group stored in the document candidate group table 131.

[0222] The extraction of text candidates using a deterministic method may be, for example, a procedure to extract the text candidate with the highest weight, or a procedure to extract multiple text candidates in descending order of weight. By extracting text candidates using a deterministic method, the support device 1 can reliably and automatically generate the most appropriate communication text for the situation indicated by the visit data. To prevent inappropriate text candidates from remaining, the procedure to extract multiple text candidates in descending order of weight preferably extracts only text candidates whose weights satisfy the requirement indicated by a given threshold.

[0223] When there are multiple candidate sentences with the highest weight, and in the procedure for extracting multiple candidate sentences in descending order of weight, the support device 1 may provide a sentence selection user interface (sentence selection UI). Steps S203 and S204 are examples of processing procedures for using the sentence selection UI.

[0224] [Step S203: Determine whether to display the text selection UI] The control unit 11 performs a process to determine whether to display a text selection UI configured by the instruction generation unit 112 to select one text candidate from multiple text candidates (text selection UI display determination step). If the control unit 11 determines to display the UI, it moves the process to step S204; otherwise, it returns the process to step S110.

[0225] This determination process includes, for example, a procedure to display a text selection UI when there are multiple text candidates with the highest weighting, or when multiple text candidates are extracted.

[0226] [Step S204: Display text selection UI] The control unit 11 executes a process to display a text selection UI configured by the instruction generation unit 112 to select one text candidate from a plurality of extracted text candidates (text selection UI display execution step). After one text candidate to be treated as an automatically generated text is selected via the text selection UI, the control unit 11 returns to step S110.

[0227] The text selection UI preferably presents multiple text candidates in order of weighting. This allows the support device 1 to both leave the selection from the text candidates to the discretion of the service manager and make a selection based on the calculated weights.

[0228] Steps S301 to S304 are an example of a weighted probabilistic extraction procedure. In this example, the computational process for automatic text generation includes a procedure for calculating the weights of a group of text candidates, which includes multiple text candidates, based on visit data, and for extracting text candidates from the group of text candidates using a probabilistic method based on these weights.

[0229] [Step S301: Calculate weights] Step S301 is the same as step S201, except that the destination is step S302.

[0230] [Step S302: Extract text candidates using a probabilistic method] The control unit 11 performs a process to extract document candidates from the document candidate group using a probabilistic method based on the weights calculated by the instruction generation unit 112 (probabilistic extraction step). The control unit 11 then moves the process to step S303. Here, "document candidate group" refers to, for example, the document candidate group stored in the document candidate group table 131.

[0231] The procedure for extracting text candidates using a weighted probabilistic method is not particularly limited, as long as it is a procedure in which text candidates with higher weights have a higher probability of being extracted. For example, such a procedure may be a conventionally known procedure used in weighted lotteries.

[0232] By performing a probabilistic sampling step, the support device 1 can automatically generate communication messages according to the priority indicated by the weighting, while simultaneously suppressing the automatic generation of identical communication messages consecutively. Therefore, the support device 1 can automatically generate communication messages that ensure comprehensive and attentive care and assistance, even in situations where diversity in communication messages is required.

[0233] [Step S303: Determine whether to display the text selection UI] Step S303 is the same as step S203, except that if it is determined that the display should be shown, the destination is step S304.

[0234] [Step S304: Display the text selection UI] Step S304 is the same as step S204, except that the texts can be presented in an order determined by a probabilistic method.

[0235] Steps S401 to S403 are an example of a procedure using a generative AI. In this example, the computational processing related to the automatic generation of text includes a procedure in which a prompt containing visit data is input to the generative AI, and the generative AI generates a natural language response.

[0236] [Step S401: Generate prompt] The control unit 11 executes a process to generate a prompt including visit data using the instruction generation unit 112 (prompt generation step). The control unit 11 then moves the process to step S402.

[0237] In the prompt generation step, the instruction generation unit 112 generates a prompt that includes an instruction sentence for the generation AI. This instruction sentence may be, for example, an instruction sentence obtained from the input of the service manager via the string input user interface. The instruction generation unit 112 may also generate a prompt that further includes one or more fixed prompts, status data, report information, and various data stored in the schedule table 132.

[0238] The fixed prompts preferably include instructions that limit the operation of the generation AI so that a text appropriate as a communication document is generated. The fixed prompts may include, for example, specifications for writing style, tone of text, length of text, and other various specifications. This results in the generation of a more appropriate communication document that satisfies the various specifications.

[0239] The instruction generation unit 112 may command the display of an instruction selection user interface (instruction selection UI) for selecting an instruction from a given template, and generate an instruction based on the service manager's selection using the instruction selection UI. This allows the service manager to provide instructions without having to manually input them one by one. Furthermore, the templates are expected to ensure stable generation results. To facilitate the selection process for the service manager, it is preferable that the templates are categorized.

[0240] Furthermore, the instruction document may include both fixed instructions selected from a template and additional instructions entered by the service manager. This allows for both the reduction of effort and stability of the generated results provided by templates, and the flexibility of manual input.

[0241] [Step S402: Enter the prompt] The control unit 11, using the instruction generation unit 112, executes the process of inputting the prompt related to step S401 to the generation AI (prompt input step). The control unit 11 then moves the process to step S403. If the generation AI is implemented on an external server, it is preferable that the support device 1 inputs the prompt via the API corresponding to the generation AI.

[0242] The generative AI is not particularly limited as long as it generates natural language. Preferably, the generative AI includes a large-scale language model (LLM) as exemplified by ChatGPT (developed by OpenAI, registered trademark) and LLaMA (developed by Meta). This allows the support device 1 to generate natural language communication texts that reflect insights shown by general natural language sentences contained in the vast amount of training data used to construct the LLM.

[0243] It is preferable that LLM undergoes additional training to limit the operation of the generating AI so that appropriate texts as communication texts are generated. This training is achieved, for example, by using a set of candidate texts stored in the candidate text table 131 as training data with positive evaluations, and / or by using a set of texts that are inappropriate as communication texts as training data with negative evaluations.

[0244] [Step S403: Obtain the generation result] The control unit 11 executes a process to acquire the generation result of the generation AI as a text automatically generated by the instruction generation unit 112 (generation result acquisition step). The control unit 11 returns the process to step S110.

[0245] In a procedure where a prompt containing visit data is input to a generative AI, and the generative AI generates a natural language response, the generative AI produces linguistically natural communication text. In addition, if the generative AI is an LLM (Learning Language Machine), it is expected to generate more practical communication text that reflects insights derived from its training data.

[0246] [Regarding automated text generation processes that use multiple methods] In order to realize automatic generation that makes the most of the advantages of each method based on the discretion of the service management responsible person, the support device 1 may provide a generation method selection user interface (generation method selection UI) that allows the service management responsible person to select an automatic generation method. As a result, the service management responsible person can select, for example, random extraction when there are no special circumstances, and select automatic generation using a rule-based system, generative AI, or a learning model when there are special circumstances.

[0247] The support device 1 may be equipped with a mechanism for automatically selecting an automatic generation method. As a result, the support device 1 can realize automatic generation that makes the most of the advantages of each method without causing the service management responsible person to expend the effort of selection.

[0248] The support device 1 may be configured to hierarchically apply a plurality of automatic generation methods. As an example of such a configuration, there is a configuration in which candidate sentences are first narrowed down by rule-based weighting, and then a generative AI generates a communication text with enhanced naturalness of meaning based on the candidate sentences.

[0249] Also, as an example of such a configuration, there is a step-by-step cooperation process in which a learning model adjusts the extraction priority and the remaining candidate sentences are reconstructed by a generative AI. As a result, the support device 1 can efficiently generate a communication text with high accuracy and high situational suitability.

[0250] <Other inventions included in system S> System S further includes the following inventions.

[0251] (1) The support device (system S) for home visit care service providers includes a display command unit that causes a screen to display visit information regarding visits related to home visit care services, and an instruction generation unit that automatically generates a text to be transmitted by a service provider responsible for providing home visit care services to a practitioner who visits the user's home to provide care services to the service user regarding the visit. The display command unit causes the screen to display visit information including at least the identification information of each of the practitioner and the service user to whom the home visit care service is provided, and the text automatically generated by the instruction generation unit. The instruction generation unit automatically generates the text by arithmetic processing that takes the visit data related to the visit as input.

[0252] According to the invention of (1), regarding "(3) Transmission by documents of user information, reports from home visit care workers, etc." among the institutional requirements related to specific business office addition, the service provider automatically generates a transmission text to be transmitted to the practitioner and displays it on the screen, thereby reducing the burden on the service provider regarding the drafting of the transmission text.

[0253] According to the invention of (1), the instruction generation unit automatically generates the text by arithmetic processing that takes the visit data related to the visit as input. Thereby, the invention can automatically generate a more situation-appropriate transmission text by an advanced technical means of arithmetic processing based on visit data, rather than simply random extraction. Therefore, the invention can automatically generate a transmission text that enables thorough and personal care and assistance, including scenes where a more situation-appropriate transmission text is required.

[0254] Therefore, the invention of (1) can provide an advanced technical means for reducing the burden on the service provider regarding obtaining specific business office addition so that thorough and personal care and assistance can be provided without omission.

[0255] (2) The support device (system S) is the invention of (1), and the instruction generation unit automatically generates the text by arithmetic processing including a procedure of calculating the weighting of the text candidates based on the visit data for a group of text candidates including a plurality of text candidates and extracting a text candidate from the group of text candidates based on the weighting.

[0256] (2) According to the invention, the instruction generation unit automatically generates a transmission text by means of a more advanced arithmetic process of extracting a text candidate from a group of text candidates based on the weighting calculated based on the access data. Thereby, the invention can automatically generate a transmission text in accordance with the priority indicated by the weighting. Therefore, the invention can automatically generate a transmission text that is more appropriate to the situation than simple random extraction. Therefore, the invention can automatically generate a transmission text that can provide thorough and personalized care and assistance without omission, including situations where a transmission text in accordance with the priority is required.

[0257] Therefore, the invention of (2) can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business office addition so that thorough and personalized care and assistance can be provided without omission.

[0258] (3) The support device (system S) is the invention of (2), and the instruction generation unit automatically generates the text by means of an arithmetic process including a procedure of extracting a text candidate from the group of text candidates using a probabilistic method based on the weighting.

[0259] In the automatic generation of a transmission text, in addition to generating a text appropriate to the situation, it is also important that the same or similar texts are not continuously and automatically generated. This is because if such automatic generation is performed, the service provided by the person in charge of the actual work will be biased.

[0260] (3) According to the invention, the instruction generation unit automatically generates a transmission text by means of a more advanced arithmetic process of extracting a text candidate from a group of text candidates using a probabilistic method based on the weighting calculated based on the access data. Thereby, the invention can achieve both the automatic generation of a transmission text in accordance with the priority indicated by the weighting and the suppression of the continuous automatic generation of the same transmission text. Therefore, the invention can automatically generate a transmission text that can provide thorough and personalized care and assistance without omission, including cases where the diversity of the transmission text is required.

[0261] Therefore, the invention of (3) can provide an advanced technical means for reducing the burden on service providers who obtain specific workplace additions so that intimate care and assistance can be provided without omission.

[0262] (4) The support device (system S) is the invention of (2) or (3), and the instruction generation unit automatically generates the text by arithmetic processing including a procedure for calculating the weight of the text candidates for a group of text candidates including a plurality of text candidates based on the access data and the learning model.

[0263] In order to perform extraction in accordance with the priority at the site of home care in extraction using weighting, it is preferable that weighting in accordance with the priority is performed. However, due to various circumstances related to home care, the weighting initially assigned to the access data may not match the priority. Examples of various circumstances include the progress of home care, changes in the circumstances surrounding home care, changes at the site, changes in the persons in charge of practice, changes in service users, and the difficulty of predicting the actual state of the service before the start of home care services.

[0264] According to the invention of (4), the instruction generation unit automatically generates a transmission text by more advanced arithmetic processing of calculating the weight of text candidates based on access data and a learning model and extracting text candidates based on the weight. As a result, the support device can continue to calculate an appropriate weight by updating the learning model so that the weight matches the priority. That is, the invention can realize automatic generation of a transmission text that enables intimate care and assistance without omission, including cases where the weighting initially assigned to the access data does not match the priority.

[0265] Therefore, the invention of (4) can provide an advanced technical means for reducing the burden on service providers who obtain specific workplace additions so that intimate care and assistance can be provided without omission.

[0266] (5) The support device (system S) is an invention according to any one of (1) to (4), and the instruction generation unit inputs a prompt including the visit data into the generation AI, and automatically generates the text by arithmetic processing including a procedure of causing the generation AI to generate a response in natural language.

[0267] Visits related to home care may have various individual circumstances. If we try to prepare a group of text candidates in advance according to a wide variety of individual circumstances, a great deal of labor and time are required in its consideration and review.

[0268] According to the invention of (5), the instruction generation unit inputs a prompt including the visit data into the generation AI, and causes the generation AI to generate a communication text in natural language. Thereby, the invention can flexibly generate a communication text in natural language according to the individual circumstances of the visit without preparing a group of text candidates in advance according to a wide variety of individual circumstances. Therefore, the invention can automatically generate a communication text that can provide affectionate care and assistance without omission regardless of the presence or absence of anticipation of individual circumstances.

[0269] Therefore, the invention of (5) can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business office addition so that affectionate care and assistance can be provided without omission.

[0270] (6) The support device (system S) is the invention of (5), and further includes a machine learning unit that performs machine learning of the generation AI using the learning data including the text.

[0271] The generation AI can generate natural language in which the knowledge indicated by general natural language texts is reflected. On the other hand, it is difficult for the generation AI to generate natural language in which knowledge not included in the learning data is reflected. Therefore, it is expected that the generation AI cannot generate a communication text in which the knowledge accumulated through the continuous provision of home care services is reflected.

[0272] According to the invention of (6), the machine learning unit performs machine learning of the generative AI using learning data including the transmission text. As a result, the invention can automatically generate a transmission text in which various findings exemplified by the findings accumulated through the continuous provision of the visiting care service are reflected. Therefore, the invention can automatically generate a transmission text that enables thorough and personalized care and assistance, including cases where various findings exemplified by the accumulated findings are required.

[0273] Therefore, the invention of (6) can provide an advanced technical means for reducing the burden on the service provider responsible for obtaining the specific business office addition so that thorough and personalized care and assistance can be provided without omission.

[0274] <Appendix> In addition, within the scope of the idea of the present invention, those skilled in the art can conceive of various modification examples and correction examples, and it is understood that those modification examples and correction examples also belong to the scope of the present invention. For example, with respect to the above-described embodiments, those in which those skilled in the art have appropriately added, deleted, or changed the design of components, or added, omitted, or changed the conditions of the steps, are also included in the scope of the present invention as long as they have the gist of the present invention.

Explanation of Signs

[0275] S System 1 Support Device 11 Control Unit 111 Display Instruction Unit 112 Instruction Generation Unit 113 Machine Learning Unit 13 Storage Unit 131 Text Candidate Group Table 132 Schedule Table 14 Communication Unit 15 Screen 16 Input Unit R1 Schedule Area R2 Visit Information Area N Network T Terminal

Claims

1. A display control unit that displays visit information related to home care services on the screen, Regarding the aforementioned visit, the service manager responsible for the aforementioned home care service will automatically generate instructions to be conveyed to the practical staff member who visits the user's home and provides care services to the service user. Equipped with, The display command unit provides an evaluation and modification user interface for the service manager to evaluate and / or modify the text automatically generated by the instruction generation unit. The instruction generation unit automatically generates the text by randomly selecting from a group of text candidates that includes multiple text candidates, or by performing calculations that take visit data relating to the visit as input, and the calculations include at least one of the following steps: calculating weights for the text candidates that include multiple text candidates based on the visit data, and extracting text candidates from the group of text candidates based on said weights, or inputting a prompt including the visit data to a generation AI and causing the generation AI to generate a natural language response. A support device for home care service providers.

2. The support device according to claim 1, wherein the instruction generation unit automatically generates the text by performing calculation processing that includes a procedure for calculating weights of the text candidates, which are a group of text candidates including a plurality of text candidates, based on the visit data, and for extracting text candidates from the group of text candidates based on the weights.

3. The support device according to claim 2, wherein the instruction generation unit automatically generates the text by an arithmetic process that includes a procedure for extracting text candidates from the text candidate group using a probabilistic method based on the weighting.

4. The support device according to claim 2 or 3, wherein the instruction generation unit automatically generates the text by performing calculation processing that includes a procedure for calculating the weights of the text candidates based on the visit data and the learning model for a group of text candidates that includes a plurality of text candidates.

5. The support device according to claim 1, wherein the instruction generation unit automatically generates the text by computational processing that includes a procedure for inputting a prompt including the visit data to a generating AI and causing the generating AI to generate a natural language response.

6. The support device according to claim 5, further comprising a machine learning unit that performs machine learning of the generating AI using the aforementioned text-based learning data.

Images