Health, medical and nursing care planning methods

Abstract

To provide a method for formulating health, medical and nursing care plans that facilitates policy formulation, etc. [Solution] The plan formulation method accepts multiple problem countermeasure datasets, which are datasets of existing problem texts and existing countermeasure texts that describe existing health, medical care, and nursing care issues and existing countermeasures for those issues in natural language, and extracts from them problem countermeasure datasets that correspond to the classified type of target issue.The generation AI then reuses the problem countermeasure datasets and numerical data or prompts containing instructions given to the generation AI during the processing process as necessary to create and output a plan in which the creator's issues and countermeasures for those issues are described in natural language.

Description

[Technical Field]

[0001] The present invention relates to a method for formulating a health care plan. [Background technology]

[0002] In response to social issues such as rising medical costs due to the declining birthrate and aging population, a shortage of medical and nursing care workers, and lifestyle-related diseases, there is a strong demand for improvements in QoW (Quality of Work), such as improving the QOL (Quality of life) of hospital and nursing care facility residents and their families, improving the work efficiency of facility employees and nursing care staff, and reducing turnover rates by improving the work environment.In addition, with health awareness growing among healthy people, including young people, the well-being and health-related market has grown significantly, and there is an accelerating trend for individuals to actively manage and improve their health, happiness, and QoL.

[0003] In response to these trends in the fields of health, medical care, and nursing care in society, the inventors are conducting multifaceted analysis and consideration of optimal allocation of human resources such as medical professionals, physical resources such as medical facilities, medical equipment, and pharmaceuticals, as well as financial and information resources throughout society, in order to provide useful suggestions for the formulation of medical and nursing care policies at medical facilities such as hospitals and nursing homes, and at national and local governments.

[0004] As a method for such analysis and consideration, for example, Patent Document 1 discloses a method for estimating the severity of a patient's condition by analyzing the patient's medical information at a medical institution, and determining the number and allocation of nurses required at that medical institution. However, because this method uses the patient's medical information at a specific medical institution as an information resource, it is not possible to multifacetedly analyze and consider the optimal allocation of human and material resources in society as a whole. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-048798 Summary of the Invention [Problem to be solved by the invention]

[0006] The present inventors handle a wide variety of data, for example, as shown in Figure 1. As shown in Figure 1, not only are the types of data diverse but also the sources from which the data is obtained are diverse, and therefore the present inventors refer to this data as "multi-source, multi-dimensional data." Analyzing multi-source, multi-dimensional data makes it possible to understand the current situation, analyze issues, and formulate policies in the fields of health, medical care, and nursing care. This in turn makes it possible to proactively manage improvements to comprehensive well-being, such as individuals' health, happiness, and QoL, as well as to formulate and support mid- to long-term policies that encompass government, businesses, and individuals.

[0007] However, analyzing multi-source, multi-dimensional data and using the results to understand the current situation in the health, medical, and nursing care fields, analyze issues, and formulate policies requires advanced analytical techniques that combine multi-source, multi-dimensional data in complex ways based on specialized knowledge, which is not easy even for experts.

[0008] The problem that this invention aims to solve is to make it easy to grasp the current situation of human resources, physical resources, financial information, and information resources, analyze issues, and formulate policies in the fields of health, medical care, and nursing care. [Means for solving the problem]

[0009] One aspect of the present invention, which has been made to solve the above problems, is a method for formulating a health, medical care, and nursing care plan, comprising: Using a computer, a first receiving step of receiving current situation description data describing the current situation of the entity creating the plan regarding at least one of health, medical care, and nursing care; a task setting step of setting a target task related to the health, medical care, or nursing care of the creator based on the current situation description data received in the first receiving step; a second receiving step of receiving numerical data related to the target task set in the task setting step from the creator; a statistics calculation step of calculating statistics relating to a higher group of the creator of the numerical data received in the second receiving step; a target task classification step of classifying the target task into one of a plurality of types by analyzing the statistics calculated in the statistics calculation step; a third receiving step of receiving a plurality of problem countermeasure datasets, which are datasets of existing problem sentences and existing countermeasure sentences in which existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; an extraction step of extracting a problem solution dataset corresponding to the type of the target problem classified in the target problem classification step from the plurality of problem solution datasets received in the third reception step; a plan creation step of using a generation AI to create a plan in which the problem of the creator and the measures for the problem are described in natural language using the problem solution dataset extracted in the extraction step and the numerical data; a knowledge storage step of storing a pair of the problem solution dataset created in the plan creation step and a prompt including the instruction content given to the generation AI as knowledge data; an output step for outputting the plan created in the plan creation step; This is what is carried out.

[0010] In the above health, medical and nursing care planning method, Using the computer, A management step of managing the current situation description data received in the first receiving step and the numerical data received in the second receiving step based on a thesaurus related to health, medical care, and nursing care can be further executed. Here, "managing" includes hierarchically dividing and storing the descriptive data and numerical data into pairs with prompts containing instructions to the generating AI, based on a thesaurus related to health, medical care, and nursing care, the meaning of each piece of data, and the subordinate relationships between the data.

[0011] The health, medical care, and nursing care plan formulation method of the present invention can be applied not only to plans of the country, local government, medical area, school, company, medical facility, nursing facility, health insurance association, etc., but also to plans of individuals such as residents of the country or local government or medical area, students at schools, employees of companies, users of medical facilities or nursing facilities, members of health insurance associations, and users of health management apps. In other words, the "plan formulation entity" in the present invention refers to an organization such as the country, local government, medical area, school, company, medical facility, nursing facility, or health insurance association, or an individual.

[0012] Furthermore, if the creating entity is, for example, a country, the "higher group of the creating entity" is the region, such as Asia, Europe, or Oceania, to which multiple countries including that country belong, or if the creating entity is, for example, a company, the industry association to which that company belongs. Furthermore, if the creating entity is an individual, the "higher group of the creating entity" is the country, local government, school, company, medical facility, nursing care facility, etc. to which that individual belongs.

[0013] "Current status description data" may be any data that can grasp the current situation of the creator regarding at least one of health, medical care, and nursing care, and may include data written by a person (worker), such as text, graphs, tables, cartoons, pictures, etc.

[0014] The type of target issue includes whether the target issue relates to human resources, material resources, financial resources, or information resources in the health, medical care, and nursing care fields, the length of time required to solve the target issue (short-term, medium-term, long-term, etc.), whether the cost required to solve the target issue is high or low, the urgency and importance of the target issue, etc.

[0015] The numerical data includes numerical data representing the performance, activities, human resources, physical resources, and financial resources of the creator in the health, medical care, and nursing care fields, as well as numerical data representing demographic and socio-economic environment data. Furthermore, if the creator is an individual, it includes numerical data representing the economic and health status of the individual, and if the creator is an organization, it includes numerical data representing the economic and health status of individuals belonging to the organization, as well as vital information and well-being information that can provide an understanding of the mental health status of individuals. The statistical quantities relating to the upper group of the entity that created the numerical data are typically standard deviations or percentiles, but also include, for example, the difference or ratio between the numerical quantity of the entity that created the data and the average, minimum, or maximum value of the upper group.

[0016] Furthermore, one aspect of the present invention is a system for formulating a health care plan using a computer, comprising: a first storage unit storing current situation description data that describes the current situation of the plan creator regarding at least one of health, medical care, and nursing care; a second storage unit storing numerical data relating to the health, medical care, and nursing care field of a higher-level group of the creator including the creator; a problem setting unit that sets a target problem related to the health, medical care, or nursing care of the creator based on the current situation description data stored in the first storage unit; a statistics calculation unit that acquires from the second storage unit numerical data of a higher group of the creator that is related to the target task set by the task setting unit, and calculates statistics of the higher group of the numerical data of the creator; a target task classification unit that analyzes the statistics calculated by the statistics calculation unit and classifies the target task into one of a plurality of types; a third storage unit that stores a problem countermeasure dataset, which is a dataset of existing problem sentences and existing countermeasure sentences in which one or more existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; a plan creation unit that acquires from the third storage unit a problem countermeasure dataset corresponding to the type of target problem classified by the target problem classification unit, and uses a generation AI to create a plan in which the problem of the creator and a sentence about a countermeasure for the problem are written in natural language using the problem countermeasure dataset corresponding to the type of target problem and the numerical data; a knowledge storage unit that stores, as knowledge data, a pair of a problem solution dataset created by the plan creation unit and a prompt including instructions given to the generation AI; an output unit that outputs the plan created by the plan creation unit; It is equipped with the following. [Effects of the Invention]

[0017] In this invention, a problem (target problem) for the plan creator is set based on a current situation description, and statistics are calculated for the creator's supergroup of numerical data representing attributes related to at least one of the creator's health, medical care, and nursing care, which are related to the target problem. The statistics are then analyzed to classify the type of the target problem. Once the type of target problem is classified, a plan is created that describes the creator's problem and its countermeasures in natural language based on existing problem sentences and existing countermeasure sentences corresponding to the type, which describe existing problems and existing countermeasures for the problem in natural language, and a prompt dataset (problem countermeasure dataset) containing instructions given to the generation AI. The plan is output. Since this plan is generated using a generation AI, a trained model that learns patterns and features using a large dataset of existing problem sentences and existing countermeasure sentences is stored in a knowledge storage unit, and the problem countermeasure dataset is then used as training data for the generation AI, allowing high-quality plans to be easily created while growing knowledge even without specialized knowledge. [Brief explanation of the drawings]

[0018] [Figure 1] FIG. 1 is a diagram showing an example of multi-source and multi-dimensional data. [Figure 2] 1 is a block diagram of an example of a health care and nursing care planning system according to an embodiment of the present invention; [Figure 3] 1 is a flowchart showing the steps of operations and processes in a health care and nursing care planning method using the health care and nursing care planning system of this embodiment. [Figure 4] FIG. 10 is a diagram for explaining an example of a flow of work for creating a tentative plan and a final plan. [Figure 5] An example of a table showing the correspondence between the content of the target task sentences included in the plan and the task classification. [Figure 6] 10 is an example of a table showing the correspondence between the contents of target countermeasure statements included in the plan and the classification of the countermeasures. [Figure 7]An example of a hierarchical or systematic data management method between each data. [Figure 8] An example of the data needed from a health, medical, and nursing care perspective when generative AI makes future plans and predicts future social changes. [Figure 9] An example of data trends (2000 to 2050) predicted by the AI ​​generation system from the perspective of health, medical care, and nursing care using the data shown in Figure 8. [Figure 10] FIG. 1 is a configuration diagram of a health care and nursing care planning system according to an embodiment. [Figure 11] 1 is a diagram showing input / output data and a processing flow of a health care planning system according to an embodiment; DETAILED DESCRIPTION OF THE INVENTION

[0019] Artificial intelligence (AI) technology has evolved significantly in recent years. Generative AI, a field of machine learning, is capable of generating new documents, images, audio, and other data from patterns and features input by a user by learning in advance the correspondence between various types of data, such as documents, images, and audio, and the patterns and features represented by that data. Here, "generative AI" refers to artificial intelligence that generates various types of text based on input data, constraints, or instructions. Currently, widely known text generation AI (also known as generative AI) includes, but is not limited to, "ChatGPT" provided by Open AI, Inc. (USA), "Copilot" provided by Microsoft, Inc. (USA), "Gemini" provided by Google, Inc. (USA), and "Llama" provided by Meta, Inc. (USA).

[0020] Generative AI includes a technology called conversational AI, which is specialized for natural conversations with humans. Services using large-scale language models (LLMs), trained using large amounts of data of various types, are beginning to be offered. LLMs are designed to understand and generate text written in natural language. By applying reinforcement learning to healthcare-related content and fine-tuning parameters, plans and numerical data analysis, which traditionally required significant time and effort from healthcare experts, can be created using decision-making support methods based on unique data generated by LLMs or Retrieval Augmented Generation (RAG). This is expected to improve the efficiency of planning and numerical data analysis. The present invention relates to a method and system for supporting healthcare-related planning using generative AI.

[0021] In the present invention, a "plan" refers, for example, to a plan for formulating and implementing policies, measures, and systems related to the health, medical care, and nursing care fields of the national and local governments, but it may also include management plans for specific health, medical care, and nursing care facilities, associations, and organizations, and plans for promoting the health and improving the lives of individuals.

[0022] Hereinafter, an embodiment of a planning support method and system according to the present invention will be described in detail with reference to the drawings.

[0023] [System Configuration] FIG. 2 is a schematic block diagram of the planning system of this embodiment. As shown in FIG. 2, the planning system of this embodiment includes a terminal 1, a multi-source, multi-dimensional database (DB) 3, and a generating AI system 4, all of which are connected to the Internet or an intranet 2 (hereinafter referred to as "Internet 2" for convenience). The generating AI system 4 is typically built on a single large-scale server via the Internet, but some of its functions may be distributed across multiple medium-sized servers, or it may be built on a small-scale server on a personal computer (PC) and specialized for its own functions. Alternatively, multiple generating AI systems may be coordinated depending on the application (data volume and analytical functions).

[0024] Terminal 1 is an edge computer (EC) such as a PC or a smartphone, and software (computer program) installed on the PC or EC runs on the PC or EC, and in some cases, also utilizes functions provided by a system (not shown) existing on the Internet 2 to perform processing in each functional block described below.

[0025] The multi-source, multi-dimensional DB3 is a database that stores a wide variety of data, such as statistical data on health, medical care, and nursing care, population, and socioeconomic conditions, personal health, medical care, and nursing care data, data including geospatial information, and various needs such as policies and management plans for customer issues. This may include widely available databases (systems) that store demographic data provided by national and local governments and survey results of health, medical care, and nursing care facilities, as well as various databases provided by private businesses. The multi-source, multi-dimensional DB3 is not limited to a single database, but may be composed of multiple databases. In other words, in this embodiment, "multi-source, multi-dimensional data" refers to a wide variety of data related to health, medical care, and nursing care. Furthermore, as shown in Figure 10, the multi-source, multi-dimensional DB3 stores a large number of optimal problem sentences and solution sentences generated by the generation AI, as well as prompts given to the generation AI, as knowledge in the knowledge storage unit. This knowledge data is then used again (feedback) as learning data for the generation AI, automatically strengthening and growing the knowledge data.

[0026] Figure 1 shows an example of multi-source, multi-dimensional data. For example, in Figure 1, administrative statistical data, health, medical care, and nursing care data, healthy lifestyle-related data, various government statistical survey data, and various published data are included in statistical data related to health, medical care, and nursing care, as well as demographic and socioeconomic trends. Also in Figure 1, hospital DPC data, health checkup data, medical and nursing care receipt data, and related data are included in individual health, medical care, and nursing care data, while transportation data, natural environment Earth observation data, various government statistical survey data, and various published data are included in data containing geospatial information. Furthermore, the various needs for policies and management plans addressing customer issues include support needs for national and local government administrative planning, support needs for organizations (hospitals, nursing care facilities, and companies), and well-being support needs for individuals.

[0027] The generative AI system 4 can be the whole or part of a system built by a provider of generative AI services, such as those mentioned above. The generative AI used here is commonly known as text generation AI. Text generation AI is a model that uses training data from a large amount of web text data available on the Internet, a pre-trained proprietary language model (LLM), or a method for supporting decision-making based on proprietary data from Retrieval Augmented Generation (RAG), to predict the next word following an input word and complete the sentence. "ChatGPT," a well-known GPT (Generative Pre-trained Transformer) algorithm, is a model tuned to output appropriate responses to user-input text, voice, and image input.

[0028] The terminal 1 includes, as functional blocks, a data input reception unit 10, a problem (needs) setting unit 11, a statistical quantity calculation unit 12, a problem countermeasure dataset extraction unit 13, a target problem classification unit 14, a plan creation unit 15, and a memory unit 16, and the memory unit 16 includes a current situation description data memory unit 160, a numerical data memory unit 161, a problem countermeasure data memory unit 162, a tentative plan storage unit 163, etc. The functions or processes of each functional block are typically realized by the above-mentioned generative AI system 4, but may also be realized by a machine learning model or generative AI built into the terminal 1.

[0029] An input unit 17 and an output unit 18 are connected to the terminal 1 as a user interface. The input unit 17 includes, for example, a touch panel, hardware keys such as a keyboard, a pointing device such as a mouse, a camera (image input), and a microphone (audio input). The output unit 18 includes, for example, a display, a touch panel, a speaker, etc. In the following description, the output unit 18 will be described as a display.

[0030] [Outline of the planning process steps] Next, an example of the procedure for formulating and processing a regional medical plan for a certain secondary medical area using the above system will be explained using the flowchart in Figure 3. In this example, the "certain secondary medical area" is the planning entity.

[0031] Secondary medical areas are regional areas established by prefectures in accordance with the Medical Care Act, a Japanese law, to develop medical care provision systems.

[0032] First, the user (the person in charge of creating the plan at the creation entity) operates the input unit 17 to input current situation description data that describes the current situation regarding at least one of health, medical care, and nursing care (hereinafter referred to as "health, medical care, and nursing care") in the secondary medical area, and the current situation description data is accepted by the data input accepting unit 10 (step 1). The current situation description data is typically text data or audio data that describes the current content of health, medical care, and nursing care in a natural language, but it may also be data in the form of graphs, tables, cartoons, or pictures, or may be data consisting of a combination of two or more of text, graphs, tables, cartoons, and pictures.

[0033] The current situation description data includes the current situation, evaluation and types of issues, content of evaluations and issues, types of measures, content of measures, etc. in the health, medical care, and nursing care fields of any of the prefectures that include the secondary medical area that is the creator, as well as for each secondary medical area within that prefecture. The current situation description data is data that can reinforce the evaluation, setting of issues, and formulation of measures for the prefecture as a whole and for each secondary medical area.

[0034] The current situation description data may be input directly by the user by operating the input unit 17, or may be input by the user by having the terminal 1 load text data prepared in advance, or may be acquired by accessing a multi-source / multi-dimensional DB 3 or other database via the Internet 2. The data input receiving unit 10 stores the received current situation description data of the secondary medical area in the current situation description data storage unit 160.

[0035] Next, when the user operates input unit 17 to issue an instruction to extract issues related to health, medical care, and nursing care in the secondary medical area, issue setting unit 11 reads out the current situation description data from current situation description data storage unit 160, extracts issues related to health, medical care, and nursing care, and sets those issues as target issues (step 2). Issues are extracted, for example, by understanding the content of the current situation description and searching for characteristic words contained in the current situation description, such as words representing human resources, material resources, and information resources related to health, medical care, and nursing care (doctors, caregivers, hospitals, nursing facilities, hospital beds, drugs, etc.), and words representing qualitative and quantitative characteristics of human resources, material resources, and information resources (shortage, uneven distribution, aging, etc.). For example, examples of target issues related to medical care include "shortage of doctors," "working style and working environment of doctors," "career development of doctors," "shortage of medical institutions," and "access to medical institutions." The issue setting unit 11 causes the output unit 18 to sequentially display the set target issues.

[0036] Next, when the user performs a predetermined operation on the input unit 17, the statistical calculation unit 13 accesses the multi-source, multi-dimensional DB 3 via the Internet 2 and retrieves numerical data related to the target issue set in step 2 from the numerous numerical data stored in the multi-source, multi-dimensional DB 3 (step 3). The numerical data visualizes and evaluates the characteristics of the prefecture containing the secondary medical care area that created the report, as well as the number of physicians, medical care facilities, medical personnel, population, income, and other characteristics of each secondary medical care area within that prefecture. The numerical data is expressed in text, numbers, and charts, and can be used to evaluate the prefecture as a whole and each secondary medical care area, identify issues, and formulate countermeasures. While the example shown here illustrates retrieving numerical data from the multi-source, multi-dimensional DB 3, the user may input prepared numerical data by loading it into the terminal 1, or access a database via an intranet or other means and retrieve the data from there.

[0037] Here, the large number of numerical data stored in the multi-source, multi-dimensional DB3 is linked to at least information specifying the source of the numerical data, such as the country, prefecture, etc., and information specifying the type of the numerical data, and it is further preferable that the large number of numerical data be linked to information that allows the large number of numerical data to be managed by linking them hierarchically and systematically.

[0038] For example, if the set target issue is "physician shortage," numerical data on the number of physicians in all secondary medical areas included in the prefecture to which the secondary medical area belongs is acquired and stored in the numerical data storage unit 161. Note that instead of the numerical data for secondary medical areas included in a prefecture, numerical data for secondary medical areas nationwide may be acquired, or numerical data for secondary medical areas in an intermediate region between a prefecture and a country (for example, the Kinki region, Chugoku region, or Hokuriku region) may be acquired. In other words, compared to the secondary medical area that is the creator, the prefecture, country, Kinki region, etc. correspond to a higher-level group.

[0039] Next, the statistics calculation unit 13 performs statistical processing on the acquired numerical data and calculates statistics (e.g., standard deviation) for the top group of physicians in the secondary medical area that created the data (step 4). The calculated statistics are then analyzed to classify the type of target problem (step 5). For example, if the standard deviation for the number of physicians in the secondary medical area is within the range of 40-50, it is determined that the number of physicians in the secondary medical area is not insufficient, and the area is recognized as a "potential physician shortage." Furthermore, if the standard deviation for the number of physicians is greater than 50, it is determined that there is a sufficient number of physicians, and the area is recognized as "no physician shortage problem." However, if the standard deviation for the number of physicians is less than 40, it is determined that there is a physician shortage, and the area is recognized as "a physician shortage has been identified." However, these numerical values ​​are tentative, and the numerical values ​​(thresholds) used to determine whether or not there is a "physician shortage" can be set as appropriate.

[0040] Next, the user performs a predetermined operation on the input unit 17, accesses the multi-source / multi-dimensional DB3 via the problem countermeasure dataset extraction unit 13 via the Internet 2, and acquires one or more problem countermeasure datasets that match the classification of the target problem identified in step 5 from the numerous problem countermeasure datasets stored in the multi-source / multi-dimensional DB3 (step 6). A problem countermeasure dataset is a dataset of sentences (existing problem sentences and existing countermeasure sentences) that describe in natural language the problems and countermeasures for those problems included in plans previously created for the formulation of health, medical care, and nursing care policies. The series of operations in the "plan formulation process procedure" described above is saved in the multi-source / multi-dimensional DB3 as a pair of optimal problem sentences and countermeasure sentences in programming language format (JSON, Python, etc.) as a dataset of instructions (prompts) to the generation AI, and is used again as a training dataset for the generation AI as needed.

[0041] Figure 10 shows a detailed system configuration diagram of a healthcare and nursing care planning system according to one embodiment, and Figure 11 shows the basic input / output data and processing flow of the healthcare and nursing care planning system according to one embodiment. In Figure 10, the user "extracts or modifies the optimal set from the linguistic data with numerical values ​​+ System Code set (JSON, Python, etc.) format" and stores it as knowledge data in the knowledge storage unit, thereby accumulating the user's knowledge in code format. In the basic input / output data relationship to the generation AI and the processing instruction flow to the generation AI shown in Figure 11, in order to obtain more optimal output (challenge sentences and solution sentences), the knowledge data is automatically strengthened and grown by instructions (prompts) for organization and efficiency as feedback from the knowledge storage unit to Field 3 (generation AI).

[0042] Here, the multiple problem-solving datasets stored in the multi-source, multi-dimensional DB3 contain at least information that can identify the entity that created them (hereinafter referred to as "specific information"), and in addition to the specific information, they may also contain information that can manage the multiple problem-solving datasets by linking them hierarchically and systematically (hereinafter referred to as "management information"), and instructions (prompts) to the generation AI.

[0043] A large number of problem-solving data sets stored in the multi-source multi-dimensional DB 3 and a data management method for the data storage unit 160 will be described. Because the data stored in the multi-source, multi-dimensional DB3 in this embodiment is large and diverse, the data structure of the multi-source, multi-dimensional DB3 significantly impacts the performance (e.g., analysis speed, capacity, system efficiency) of the planning system according to this embodiment, and is also closely related to the accuracy of the analysis and predictions of the generation AI. For this reason, personal information is managed according to the attributes of the multi-source, multi-dimensional data, i.e., public nature or the individual's privacy and intentions. For example, as shown in Figure 7, in the multi-source, multi-dimensional DB3 or data storage unit 160, each data is hierarchically classified into folders by target organization, region, year, data attribute, and data details, and managed according to these hierarchies. Alternatively, flags (L1 to Ln) are assigned to each data level according to the public nature, privacy, or intention of each data, and during statistical processing, the data is processed in order of priority (e.g., assigned numbers) and confidentiality (e.g., assigned markers) according to the flags. During processing, a prompt dataset containing instructions given to the generation AI is stored in the knowledge storage unit in a programming language (e.g., JSON, Python), along with a pair of optimal problem statements and solution statements.

[0044] Furthermore, as a method for managing the numerous problem countermeasure datasets and data storage unit 160, the planning system of this embodiment hierarchically divides and manages multi-source, multi-dimensional data according to its meaning, a thesaurus of healthcare, medical care, or the subordination relationships between data (for example, chapters, sections, items, paragraphs, and content of a text). The generation AI reads the hierarchically divided and managed current situation description data, or prompts containing instructions given to the generation AI during processing, from the current situation description data storage unit 160, extracts healthcare, medical care, and nursing issues, and when setting those issues as target issues, it can calculate the meaning of the input sentence, the thesaurus or the subordination relationships that can be read from it, and the vector distance of words, improving the accuracy and speed of the analysis and prediction of the generation AI.

[0045] When the creator is a secondary medical area, the user typically searches for problem-solving datasets for each secondary medical area nationwide, using, for example, a word describing the target problem (and / or the secondary medical area number, name, etc.) as a search term, and extracts one or more problem-solving datasets that match the type of the target problem. The extracted problem-solving datasets may include problem-solving datasets previously created by the user (creator) himself. When the number of extracted problem-solving datasets is large, the user may use the specific information and the management information as search terms to narrow down the search to problem-solving datasets for secondary medical areas that have similar populations, regions, demographic and socioeconomic factors, climates, etc. to the creator's secondary medical area.

[0046] Returning to the flowchart in Figure 3, the problem countermeasure dataset acquired in step 6 is stored in pair with a prompt containing instructions to the generation AI in the problem countermeasure dataset storage unit 162. In other words, the problem countermeasure dataset storage unit 162 and the knowledge storage unit correspond to the third storage unit.

[0047] Next, when the user performs a predetermined operation on the input unit 17, the plan creation unit 15 inputs one or more problem countermeasure datasets stored in the problem countermeasure dataset storage unit 162 to the generation AI system 4, as shown in Figure 10, and identifies the target problem and the classification of countermeasures corresponding to that classification from the existing problem sentences and existing countermeasure sentences included in the problem countermeasure datasets, and generates a prompt requesting the generation AI to create sentences for the problem and countermeasures, and gives this to the generation AI system 4. The plan creation unit 15 is responsible for the exchange of information between the user and the generation AI system 4, for example, using an API (Application Programming Interface) provided by the generation AI system 4. The prompts are sent to the generation AI system 4, which creates a tentative plan in which sentences about the target problem and its corresponding classification and countermeasures (target problem sentence and target countermeasure sentence) are written in natural language based on the existing problem sentence and existing countermeasure sentence, as shown in Figure 11, and sequentially displays on the output unit 18 a data set of prompts including instructions given to the generation AI during the processing based on the user's instructions in a programming language (JSON, Python, etc.) format and a pair of optimal problem sentence and countermeasure sentence (step 7). The plan creation unit 15 saves the first draft of the created tentative plan in the tentative plan storage unit 163 and sequentially displays it on the output unit 18 (step 8).

[0048] If the input unit 17 is a microphone, the user can use voice commands to create sentences corresponding to the target problem and its classification from existing problem sentences and existing solution sentences included in the problem solution dataset. This configuration allows the user to provide instructions based on the classification of the target problem, numerical data, or specialized knowledge through dialogue with the generative AI system 4, and convert the existing problem sentences and existing solution sentences into content that reinforces the target problem or solution. Furthermore, through dialogue with the generative AI system 4, the generative AI can predict future plans and social changes and present the user with problems. The user can then formulate instructions based on specialized knowledge to solve the problems in the form of prompts or code.

[0049] Figure 8 shows an example of the data required from the perspective of health, medical care, and nursing care when the generative AI predicts future plans and future social changes. Figure 9 also shows an example of data trends (2000 to 2050) predicted by the generative AI from the perspective of health, medical care, and nursing care using the data shown in Figure 8.

[0050] As shown in FIG. 11, the user checks the contents of the tentative plan displayed sequentially on the output unit 18, checks whether there are any points that need to be corrected or revised, and if there are any points that need to be corrected or revised, performs a predetermined operation on the input unit 17 to input the content to be corrected or revised, and then issues a command to recreate (feedback) the plan sequentially (if No in step 9). This returns to the processing of step 3, and steps 3 to 8 are repeatedly executed until Yes in step 9. Here, the "contents to be corrected or revised" input by the user include information identifying the source and type of the numerical data accepted in step 3, the contents of the statistics calculated in step 4, etc. Therefore, in the processing of steps 3 to 8 from the second time onwards, the type of target problem is classified based on numerical data and statistics that are different from those used in the previous processing of steps 3 to 8 (step 5), and a problem countermeasure dataset is extracted (step 6). As a result, the generative AI system 4 creates a tentative plan with revised content that differs from the previous one (step 8).

[0051] If there are no corrections or revisions to be made to the contents of the tentative plan displayed on the output unit 18, the user performs a predetermined operation on the input unit 17 to issue an instruction to end the plan creation process (Yes in step 9). As a result, the tentative plan is finalized as the final plan, and is stored in the tentative plan storage unit 163 and the knowledge storage unit, and is also displayed on the output unit 18 (step 10).

[0052] Figure 4 shows an example of the process leading up to the creation of the provisional plan and the final plan when "physician shortage" is extracted as the target issue in Step 2 and the target issue is classified as "physician shortage identification" in Step 5. In this example, the number of physicians for each secondary medical area is accepted as numerical data in Step 3, the deviation value for the entire prefecture, which is the upper group of the secondary medical area created by the creator, is calculated as a statistic in Step 4, the classification of each issue and countermeasure is identified in Steps 5 and 6, and points to be corrected or revised in the contents of the provisional plan are entered in Step 9. Steps 3 to 8 are then repeated, resulting in the creation of a plan that includes proposed issues and countermeasures based on multiple combinations of issue classifications and countermeasure classifications.

[0053] Figure 5 shows the problem classification and the specific problem content (text) related to that problem classification, and Figure 6 shows the countermeasure classification and the specific countermeasure content (text) related to that countermeasure classification. Figures 4 to 6 show that in this example, when "physician shortage" was identified, the types of problems were classified into four, and four countermeasure classifications were identified accordingly, resulting in a regional medical plan that includes four types of problems and countermeasures. It can be seen that both regional medical plans are more specific in content than the provisional plan.

[0054] The contents of the target problem statements and target measure statements included in the regional medical plan, as well as the classification of the problems and measures, vary depending on the size of the population and area of ​​the entity that prepares it, its geographical conditions, etc., and may also change due to technological advances, etc. In short, the above embodiment is merely one example of the present invention, and it goes without saying that any appropriate changes, modifications, additions, etc. made within the spirit of the present invention will also be encompassed within the scope of the claims of this application.

[0055] [Aspect] It will be appreciated by those skilled in the art that the exemplary embodiments described above are examples of the following aspects.

[0056] (Item 1) One aspect of the present invention is a method for formulating a health care plan, comprising: Using a computer, a first receiving step of receiving current situation description data describing the current situation of the entity creating the plan regarding at least one of health, medical care, and nursing care; a task setting step of setting a target task related to the health, medical care, or nursing care of the creator based on the current situation description data received in the first receiving step; a second receiving step of receiving numerical data related to the target task set in the task setting step from the creator; a storage and management step of hierarchically dividing, managing, and storing the current state description data and the numerical data according to the meanings, thesauruses, or subordinate relationships between the data; a statistics calculation step of calculating statistics relating to a higher group of the creator of the numerical data received in the second receiving step; a target task classification step of classifying the target task into one of a plurality of types by analyzing the statistics calculated in the statistics calculation step; a third receiving step of receiving a plurality of problem countermeasure datasets, which are datasets of existing problem sentences and existing countermeasure sentences in which existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; an extraction step of extracting a problem solution dataset corresponding to the type of the target problem classified in the target problem classification step from the plurality of problem solution datasets received in the third reception step; a plan creation step in which a generation AI is used to create a plan in which the problem of the creator and a sentence about a solution to the problem are written in natural language using the problem solution dataset extracted in the extraction step and the numerical data; a knowledge storage step of storing, as knowledge data, the problem solution dataset created in the plan creation step and a set of prompts including instructions given to the generation AI; an output step for outputting the plan created in the plan creation step; This is what is carried out.

[0057] (10) The health, medical care, and nursing care planning system according to paragraph 10 is a computer-based system, a first storage unit storing current situation description data that describes the current situation of the plan creator regarding at least one of health, medical care, and nursing care; a second storage unit storing numerical data relating to the health, medical care, and nursing care field of a higher-level group of the creator including the creator; a problem setting unit that sets a target problem related to the health, medical care, or nursing care of the creator based on the current situation description data stored in the first storage unit; a statistics calculation unit that acquires from the second storage unit numerical data of a higher group of the creator that is related to the target task set by the task setting unit, and calculates statistics of the higher group of the numerical data of the creator; a target task classification unit that analyzes the statistics calculated by the statistics calculation unit and classifies the target task into one of a plurality of types; a third storage unit that stores a problem countermeasure dataset, which is a dataset of existing problem sentences and existing countermeasure sentences in which one or more existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; a plan creation unit that acquires from the third storage unit a problem countermeasure dataset corresponding to the type of target problem classified by the target problem classification unit, and uses a generation AI to create a plan in which the problem of the creator and a sentence about a countermeasure for the problem are written in natural language using the problem countermeasure dataset corresponding to the type of target problem and the numerical data; a knowledge storage unit that stores, as knowledge data, a set of prompts including the problem solution dataset created by the plan creation unit and instructions given to the generation AI; an output unit that outputs the plan created by the plan creation unit; It is equipped with the following.

[0058] According to the health, medical care, and nursing care planning method and health, medical care, and nursing care planning system described in paragraph 1 and paragraph 10, a problem (target problem) for the plan creator is set based on a current situation description. Statistics are calculated for the plan creator's supergroup of numerical data representing attributes related to at least one of the plan creator's health, medical care, and nursing care, which are related to the target problem. The statistics are analyzed to classify the type of the target problem. Once the type of target problem is classified, a plan is created that describes the plan creator's problem and its countermeasures in natural language based on a dataset of existing problem sentences and existing countermeasure sentences (problem countermeasure dataset) that describes existing problems and existing countermeasures for the problem in natural language, and the plan is output. Because this plan is generated using a generative AI, a trained model that learns the patterns and characteristics of a large dataset of existing problem sentences and existing countermeasure sentences can be created, making it easy to create high-quality plans even without specialized knowledge. The above method and system can present problems and solutions by using large amounts of existing multi-source and multi-dimensional data and predicting future plans and social changes.

[0059] (2) The method for formulating a health, medical, and nursing care plan under paragraph 2 is the method for formulating a health, medical, and nursing care plan under paragraph 1, Using the computer, A management step can be further executed to manage the current situation description data received in the first reception step, and the numerical data received in the second reception step, or prompts including instructions given to the generation AI during the processing process, based on a thesaurus related to health, medical care, and nursing care.

[0060] According to the method for formulating a health, medical, and nursing care plan described in Section 2, the current situation description data, numerical data, and instructions (prompts) given to the generation AI are managed based on a thesaurus related to health, medical, and nursing care, so that when the generation AI creates a plan in the plan creation step, it is possible to improve the accuracy of analysis of the issues and countermeasures of the creator and shorten processing time.

[0061] (3) The method for formulating a health, medical and nursing care plan under paragraph 3 is the method for formulating a health, medical and nursing care plan under paragraph 1 or paragraph 2, The creator may be an organization including the country, local government, living area, school, company, medical facility, nursing care facility, health insurance association (National Health Insurance, Long-Term Care Insurance System for the Elderly, National Health Insurance Association, health insurance association, etc.), or an individual.

[0062] (4) The method for formulating a health, medical, and nursing care plan under paragraph 4 is the method for formulating a health, medical, and nursing care plan under paragraph 1 or 2, The creating entity is a national government, a local government, or a living area, The numerical data may be the number and location information of human or material resources related to health, medical care, and nursing care that exist in the creating entity.

[0063] (5) The method for formulating a health, medical, and nursing care plan according to paragraph 5 is a method for formulating a health, medical, and nursing care plan according to any one of paragraphs 1 to 4, The numerical data may include numerical data of geospatial information of the creator.

[0064] (6) The method for formulating a health, medical, and nursing care plan according to paragraph 6 is a method for formulating a health, medical, and nursing care plan according to any one of paragraphs 1 to 5, Using a computer, a voice or conversation sentence receiving step of receiving input of voice or conversation sentence from a user; and a prompt creation step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, The plan creation step The plan is created by using a generation AI to convert the existing problem sentences included in the problem countermeasure dataset extracted in the extraction step and the sentences related to the target problem included in the existing countermeasure sentences based on the type of the target problem, the numerical data, or the prompt created in the prompt creation step. The user's knowledge is accumulated by saving a pair of the problem solution dataset created in the plan creation step and the instructions (prompts) given to the generation AI as knowledge data. If necessary, the created plan dataset can be used again as learning data for the generation AI, allowing knowledge to evolve automatically even without specialized knowledge, making it possible to easily create high-quality plans.

[0065] In the health, medical care, and nursing care planning method according to paragraph 6, the voice or conversation receiving step may have a function of converting the language of the input voice or conversation into the language of another region, as necessary, depending on the usage environment, user instructions, etc.

[0066] (7) The method for formulating a health, medical, and nursing care plan according to paragraph 7 is a method for formulating a health, medical, and nursing care plan according to any one of paragraphs 1 to 6, The plan creation step The plan can be created by using a generative AI to predict future plans for health, medical care, and nursing care, as well as social changes, and by converting existing problem sentences contained in the problem countermeasure dataset extracted in the extraction step and sentences related to the target problem contained in the existing countermeasure sentences.

[0067] (8) The method for formulating a health, medical, and nursing care plan according to paragraph 8 is a method for formulating a health, medical, and nursing care plan according to any one of paragraphs 1 to 7, The problem setting step can be configured to accept input of prompts including terms and numerical data related to health, medical care, and nursing care, or instructions given to the generation AI during the processing process, and set the target problem based on the terms, the numerical data, and the current situation description received in the first reception step.

[0068] (Item 9) The planning support method according to item 9 is a planning support method according to any one of items 1 to 8, Using a computer, a voice or conversation sentence receiving step of receiving input of voice or conversation sentence; and a prompt creation step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, In the second receiving step, the numerical data is received based on the prompt created in the prompt creating step; In the extracting step, the problem countermeasure dataset can be extracted based on the prompt created in the prompt creating step.

[0069] In the health, medical care, and nursing care planning method according to paragraph 9, the voice or conversation receiving step may have a function of converting the language of the input voice or conversation into the language of another region, as necessary, depending on the usage environment, user instructions, etc.

[0070] (11) The planning support method according to paragraph 11 is a planning support method according to any one of paragraphs 1 to 9, Using a computer, a voice or conversation sentence receiving step of receiving input of voice or conversation sentence; and a prompt creation step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, In the prompt creation step, when a prompt is created that instructs the re-creation of the plan output in the output step, a generation AI can be used to execute the problem setting step, the second reception step, the statistics calculation step, the target problem classification step, the third reception step, the extraction step, and the plan creation step, using the plan as the problem countermeasure dataset.

[0071] (Clause 12) The data management method of clause 12 manages the current situation description data received in the first reception step of the health, medical care, and nursing care planning method described in any one of claims 1 to 9, or the instructions (prompts) given to the generation AI during the processing, and the numerical data received in the second reception step, according to the attributes of the entity that created the current situation description data and the numerical data.

[0072] (Clause 13) The data management method according to clause 13 manages the current situation description data received in the first reception step of the method for formulating health, medical care, and nursing care described in any one of claims 1 to 9, or the instructions (prompts) given to the generation AI during the processing, and the numerical data received in the second reception step, according to a thesaurus related to health, medical care, and nursing care. [Explanation of symbols]

[0073] 1...Device 10...Data entry reception section 11...Problem Setting Section 12…Statistics calculation section 13...Problem solution dataset extraction section 14...Target Issue Classification Section 15...Planning Department 16...Storage section 160...Current status description data storage unit 161...Numerical data storage unit 162...Problem solution dataset storage section 163…Temporary Plan Preservation Department 17...Input section 18...Output section 2. Internet 3…Multi-source multi-dimensional DB 4. Generative AI system

Claims

1. A method for formulating a health care plan, comprising: Using a computer, a first receiving step of receiving current situation description data describing the current situation of the entity creating the plan regarding at least one of health, medical care, and nursing care; a task setting step of setting a target task related to the health, medical care, or nursing care of the creator based on the current situation description data received in the first receiving step; a second receiving step of receiving numerical data related to the target task set in the task setting step from the creator; a statistical quantity calculation step of acquiring numerical data related to the target task set in the task setting step of a higher group of the creators including the creators, and calculating statistical quantities related to the higher group of the creators of the numerical data received in the second receiving step; a target task classification step of analyzing the statistics calculated in the statistics calculation step and classifying the target task into one of a plurality of types; a third receiving step of receiving a plurality of problem countermeasure datasets, which are datasets of existing problem sentences and existing countermeasure sentences in which existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; an extraction step of extracting a problem solution dataset corresponding to the type of the target problem classified in the target problem classification step from the plurality of problem solution datasets accepted in the third acceptance step; a plan creation step of inputting a prompt to the generation AI, and using the problem solution dataset extracted in the extraction step and the numerical data, causing the generation AI to create a plan in which the target problem and the solution to the target problem are described in natural language; a knowledge storage step of storing the problem solution data set used in the plan creation step and a pair of prompt sets input to the generation AI as knowledge data; an output step for outputting the plan created in the plan creation step; A method for formulating health, medical and nursing care plans to implement the above.

2. 2. The health, medical care, and nursing care planning method according to claim 1, Using the computer, The health, medical, and nursing care planning method of claim 1, further comprising a management step of managing the current situation description data received in the first reception step, the numerical data received in the second reception step, and the prompts input to the generation AI in the plan creation step based on a thesaurus related to health, medical, and nursing care.

3. The health, medical, and nursing care planning method of claim 1, wherein the creating entity is an organization including the national government, local government, living area, school, company, medical facility, nursing care facility, health insurance association (National Health Insurance, Long-Term Care Insurance System for the Elderly, National Health Insurance Association, health insurance association, etc.), or an individual.

4. The creating entity is a national government, a local government, or a living area, The health, medical care, and nursing care plan formulation method according to claim 1 , wherein the numerical data is the number and location information of human or material resources related to health, medical care, and nursing care present in the planning entity.

5. The health, medical care, and nursing care planning method according to claim 1 , wherein the numerical data includes numerical data of geospatial information of the creator.

6. Using a computer, a voice or conversation sentence receiving step of receiving input of voice or conversation sentence from a user; and a prompt creation step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, The plan creation step A method for formulating a health, medical, and nursing care plan as described in claim 1, wherein the plan is created by using a generation AI to convert existing task sentences included in the task countermeasure dataset extracted in the extraction step and sentences related to the target task included in the existing countermeasure sentences based on the type of the target task, the numerical data, or the prompt created in the prompt creation step, and the plan and the prompt including the instructions of the generation AI are saved as a pair of knowledge data, and a new plan is created by again using the knowledge data as learning data for the generation AI.

7. The plan creation step Using generative AI, health The health, medical, and nursing care planning method of claim 6, wherein the plan is created by predicting future medical and nursing care plans and social changes, and converting existing problem sentences contained in the problem countermeasure dataset extracted in the extraction step and sentences related to the target problem contained in the existing countermeasure sentences.

8. 2. The health, medical, and nursing care planning method of claim 1, wherein the task setting step accepts input of a prompt including terms and numerical data related to health, medical care, and nursing care, or instructions given to the generation AI during the processing process, and sets the target task based on the terms, the numerical data, and the current situation description data received in the first reception step.

9. Using a computer, The method further executes a voice or conversation sentence receiving step of receiving input of a voice or conversation sentence, and a prompt creating step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, In the second receiving step, the numerical data is received based on the prompt created in the prompt creating step; The health, medical care, and nursing care planning method according to claim 1 , wherein the extraction step extracts the problem countermeasure dataset based on the prompt created in the prompt creation step.

10. A system for formulating a health care plan using a computer, comprising: a first storage unit storing current situation description data describing the current situation of the plan creator regarding at least one of health, medical care, and nursing care; a second storage unit storing numerical data relating to the health, medical care, and nursing care field of a higher-level group of the creator including the creator; a problem setting unit that sets a target problem related to the health, medical care, or nursing care of the creator based on the current situation description data stored in the first storage unit; a statistics calculation unit that acquires, from the second storage unit, numerical data of a higher group of the creator that is related to the target task set by the task setting unit, and calculates statistics of the higher group of numerical data of the creator; a target task classification unit that analyzes the statistics calculated by the statistics calculation unit and classifies the target task into one of a plurality of types; a third storage unit that stores a problem countermeasure dataset, which is a dataset of existing problem sentences and existing countermeasure sentences in which one or more existing problems related to at least one of health, medical care, and nursing care and existing countermeasures for the problems are described in natural language; a plan creation unit that acquires from the third storage unit a problem countermeasure dataset corresponding to the type of target problem classified by the target problem classification unit, and uses a generation AI to create a plan in which the target problem of the creator and a sentence related to the countermeasure for the target problem are written in natural language using the problem countermeasure dataset corresponding to the type of target problem and the numerical data; an output unit that outputs the plan created by the plan creation unit; A health, medical and nursing care planning system that includes:

11. Using a computer, a voice or conversation sentence receiving step of receiving input of voice or conversation sentence; and a prompt creation step of analyzing the voice or conversation sentence received in the voice or conversation sentence receiving step and creating a prompt including a question and / or an instruction for the generation AI, 2. The health, medical, and nursing care planning method of claim 1, wherein, in the prompt creation step, a prompt is created instructing the re-creation of the plan output in the output step, and then, using a generation AI, the plan is used as the problem countermeasure dataset to execute the problem setting step, the second reception step, the statistical calculation step, the target problem classification step, the third reception step, the extraction step, and the plan creation step.

12. The health, medical care, and nursing care planning method according to any one of claims 1 to 9, A method for formulating a health, medical, and nursing care plan, comprising a step of hierarchically dividing and managing the current situation description data received in the first reception step and the numerical data received in the second reception step based on the dependency relationship between the data.

Images