Information providing apparatus, information providing method, and program

Abstract

To enable even a person having no medical expert knowledge to grasp a medicine suitable for himself / herself.SOLUTION: The information providing device 1 includes an information acquisition unit 131 that acquires health information indicating the health condition of the user, a query creation unit 133 that creates a search query of a drug database for searching for a drug suitable for the user by inputting the first instruction text to the first large-scale language model together with the health information, a drug identification unit 134 that identifies one or more drugs corresponding to the health information by searching the drug database using the search query, and an output unit 135 that outputs drug information related to the drug identified by the drug identification unit 134.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information providing device, an information providing method, and a program for providing information about medicines. [Background technology]

[0002] Conventionally, there has been known a method of searching a database for a drug or treatment method suitable for a patient's symptoms (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2022-531006 Summary of the Invention [Problem to be solved by the invention]

[0004] To obtain appropriate search results using a database, it is necessary to input appropriate keywords into the database. Therefore, with conventional methods, it has been difficult for people without specialized medical knowledge to perform appropriate searches.

[0005] The present invention has been made in consideration of these points, and aims to enable even people without specialized medical knowledge to know which medication is suitable for them. [Means for solving the problem]

[0006] An information providing device according to a first aspect of the present invention includes an information acquisition unit that acquires health information indicating a user's health condition, a query creation unit that creates a search query for a drug database to search for drugs suitable for the user by inputting a first instruction text together with the health information into a first large-scale language model, a drug identification unit that identifies one or more drugs corresponding to the health information by searching the drug database using the search query, and an output unit that outputs drug information related to the drugs identified by the drug identification unit.

[0007] The query creation unit may create the search query including the user's age and the user's symptoms as keywords by inputting the first instruction text, which instructs the first large-scale language model to create the search query including keywords related to the user's attributes and the user's symptoms.

[0008] The information acquisition unit may further acquire paper information, and the information providing device may further have a database creation unit that inputs second instruction text together with the paper information into a second large-scale language model and creates structured paper data in which the paper information is structured, thereby creating the drug database.

[0009] The database creation unit may create the drug database by inputting the second instruction text, which instructs the extraction of drug names and symptom-related keywords from the paper information, into the second large-scale language model.

[0010] The drug identification unit may identify the paper information associated with words included in the search query in the drug database, and input the identified paper information, the health information, and third instruction text instructing to output the drug information into a third large-scale language model to identify the drug name and the reason for recommending the drug to be output from the third large-scale language model, and the output unit may output the drug information including the drug name and reason identified by the drug identification unit.

[0011] In the drug database, each of a plurality of drugs is associated with other drugs that cannot be used in combination, and the output unit may output the names of the other drugs that cannot be used in combination along with the name of the drug identified by the drug identification unit.

[0012] In the drug database, each of a plurality of drugs is associated with a country in which the drug is approved, and the output unit may output information indicating the country in which the drug is approved along with the name of the drug identified by the drug identification unit.

[0013] A second aspect of the information provision method of the present invention includes the steps of: acquiring health information indicating a user's health condition, executed by a computer; inputting a first instruction text together with the health information into a first large-scale language model to create a search query for a drug database for searching for drugs suitable for the user; searching the drug database using the search query to identify one or more drugs corresponding to the health information; and outputting drug information regarding the identified drugs.

[0014] A third aspect of the program of the present invention is a program for causing a computer to execute the steps of acquiring health information indicating a user's health condition, creating a search query for a drug database to search for drugs suitable for the user by inputting a first instruction text together with the health information into a first large-scale language model, searching the drug database using the search query to identify one or more drugs corresponding to the health information, and outputting drug information regarding the identified drugs. [Effects of the Invention]

[0015] The present invention has the effect of enabling even a person without specialized medical knowledge to know which medicine is suitable for him or her. [Brief explanation of the drawings]

[0016] [Figure 1]FIG. 1 is a diagram for explaining an overview of an information providing system S. [Figure 2] 1 is a diagram showing an outline of a processing flow in the information providing device 1. FIG. [Figure 3] 1 is a diagram showing a configuration of an information providing device 1. FIG. [Figure 4] FIG. 10 is a diagram showing an example of keyword extraction instruction text. [Figure 5] FIG. 10 is a diagram showing items included in structured paper data. [Figure 6] FIG. 10 is a diagram showing an example of structured paper data. [Figure 7] FIG. 10 is a diagram showing an example of a query generation instruction text. [Figure 8] 10 is a diagram showing an example of drug information output by an output unit 135. FIG. [Figure 9] 10 is a flowchart showing the flow of processing by the information providing device 1 to create a thesis database. [Figure 10] 10 is a flowchart showing a process flow in which the information providing device 1 outputs drug information. DETAILED DESCRIPTION OF THE INVENTION

[0017] [Outline of Information Provision System S] FIG. 1 is a diagram for explaining an overview of an information provision system S. The information provision system S is a system for providing information (hereinafter referred to as "drug information") about drugs recommended to a user U based on health information indicating the health condition of the user U. The user U is a person who wants to purchase drugs suitable for him or her. Drugs include medicines, supplements, or health foods. The drug information includes the name of the drug. The drug information may also include information indicating the reason for recommending the drug.

[0018] The information providing system S includes an information providing device 1 and a user terminal 2. The information providing device 1 is a computer, such as a server, that provides drug information based on health information transmitted from the user terminal 2. The health information is text indicating the results of a health check for user U, or a sentence indicating the health condition of user U written by user U or a person related to user U. The information providing device 1 inputs the health information into a large-scale language model (LLM) to determine keywords for searching for drugs, and identifies drugs to recommend to user U by searching a database using the determined keywords.

[0019] The user terminal 2 is a computer used by the user U. As an example, the user terminal 2 accesses a drug search site provided by the information providing device 1 and displays a screen of the drug search site, thereby accepting input of health information on the screen. The user terminal 2 transmits the input health information to the information providing device 1. The user terminal 2 may transmit the results of a health check input by the user U to the information providing device 1 as health information, or may transmit a text input by the user U explaining the health condition of the user U to the information providing device 1 as health information.

[0020] 2 is a diagram showing an overview of the processing flow in the information providing device 1. The information providing device 1 acquires paper information including descriptions of drugs, and identifies drugs to recommend to the user U based on the paper information. The paper information may include descriptions of the results of research and development related to drugs or medicine, or may include product descriptions of drugs.

[0021] The information providing device 1 creates structured paper data by inputting the acquired paper information and keyword extraction instruction text (corresponding to the second instruction text) that instructs to extract keywords related to drug names and symptoms from the paper information into the paper structuring LLM (corresponding to the second large-scale language model). Details of the structured paper data will be described later, but it is data in which multiple words related to items such as patient attributes, drug names, comparison results, and efficacy contained in the paper information are associated with each other.

[0022] The information providing device 1 creates a drug database based on the structured paper data. Specifically, the information providing device 1 creates the drug database by arranging words contained in the structured paper data in any of the columns, records, or fields of a table in the drug database. In the drug database, words contained in the structured paper data are associated with paper information containing those words. The information providing device 1 updates the drug database in response to the acquisition of new paper information.

[0023] In the drug database, each of the multiple drugs may be associated with other drugs that cannot be used in combination with the multiple drugs. Also, in the drug database, each of the multiple drugs may be associated with the countries in which the drug is approved.

[0024] When the information providing device 1 receives health information from the user terminal 2 together with a request for provision of drug information, the information providing device 1 creates a search query by inputting the health information and a query creation instruction text (corresponding to the first instruction text) instructing the user to create a search query including keywords related to the user's attributes and symptoms into the query creation LLM (corresponding to the first large-scale language model). The query creation instruction text is, for example, text that reads, "Please create a search query including keywords related to the patient's attributes and symptoms corresponding to the health information."

[0025] If the query creation LLM is a large-scale language model for creating a search query whose keywords include patient attributes and symptoms corresponding to the health information, the query creation instruction text may simply be text instructing the user to create a search query, such as "Please create a search query." If the query creation LLM is configured to create a search query when health information is input, the health information may also serve as the query creation instruction text. The information providing device 1 searches the drug database using the created search query to identify related paper information, which is paper information containing explanatory text related to the words included in the search query.

[0026] The information providing device 1 inputs drug information output instruction text (corresponding to the third instruction text) for instructing the output of drug information together with related paper information and health information into the drug search LLM (corresponding to the third large-scale language model), thereby acquiring text indicating the name of a drug recommended to the user U and the reason for recommending the drug from the drug search LLM. The information providing device 1 provides the drug information including the acquired text to the user terminal 2.

[0027] By configuring the information providing system S as described above, even if the user U does not have specialized knowledge about medicines or medical care, the user U can know which medicines are suitable for his or her health condition. The configuration and operation of the information providing device 1 will be described in detail below.

[0028] [Configuration of information providing device 1] 3 is a diagram showing the configuration of the information providing device 1. The information providing device 1 includes a communication unit 11, a storage unit 12, and a control unit 13.

[0029] The information providing device 1 transmits and receives data between a user terminal 2, an article information server 3, an article structuring LLM server 4, a query creation LLM server 5, and a drug search LLM server 6. The article information server 3 stores article information. The article structuring LLM server 4 stores article structuring LLMs. The query creation LLM server 5 stores query creation LLMs. The drug search LLM server 6 stores drug search LLMs. The information providing device 1 may have the functions of at least one of the article information server 3, the article structuring LLM server 4, and the query creation LLM server 5.

[0030] The communication unit 11 has a communication interface for transmitting and receiving data to and from other devices. The communication unit 11 receives health information and instruction text from the user terminal 2. The communication unit 11 also receives paper information from the paper information server 3. The communication unit 11 transmits instruction text to the paper structuring LLM server 4, the query creation LLM server 5, and the drug search LLM server 6, and receives response text transmitted from the paper structuring LLM server 4, the query creation LLM server 5, and the drug search LLM server 6. The communication unit 11 stores the received data in the memory unit 12 and notifies the control unit 13.

[0031] The memory unit 12 has storage media such as a read-only memory (ROM), a random access memory (RAM), and a solid-state drive (SSD). The memory unit 12 stores programs executed by the control unit 13. The memory unit 12 also temporarily stores health information and instruction texts transmitted from the user terminal 2. The memory unit 12 temporarily stores response texts transmitted from the thesis structuring LLM server 4, the query creation LLM server 5, and the drug search LLM server 6.

[0032] The storage unit 12 also stores a drug database (drug DB) 121. In the drug database, patient attribute information, intervention information, comparison information, and result information (outcome information) are associated with each other. The patient attribute information includes, for example, the patient's age, sex, medical history, etc. The intervention information includes information on drugs that may be effective in treating the patient, such as the drug name and dosage. The comparison information includes information on the drug name and dosage of drugs that can replace the drug indicated in the intervention information. The result information includes the name of an index of the patient's condition that changes when the drug indicated in the intervention information and the comparison information is administered, the value of the condition change and its unit, and information indicating statistical significance. The drug database 121 may be stored in an external server.

[0033] The control unit 13 has, for example, a CPU (Central Processing Unit). The control unit 13 executes the programs stored in the storage unit 12, thereby functioning as an information acquisition unit 131, a database creation unit 132, a query creation unit 133, a drug identification unit 134, and an output unit 135.

[0034] The information acquisition unit 131 acquires paper information from the paper information server 3. The information acquisition unit 131 may acquire unacquired paper information by periodically accessing the paper information server 3, or may acquire paper information in response to receiving a notification from the paper information server 3 that new paper information is available. The information acquisition unit 131 stores the acquired paper information in the memory unit 12, and notifies the database creation unit 132 that new paper information has been acquired.

[0035] The information acquisition unit 131 also acquires health information indicating the health condition of the user U. For example, the information acquisition unit 131 acquires text data indicating the health condition input by the user U at the user terminal 2 as the health information. The text indicating the health condition is, for example, text such as "I am a 77-year-old man. I am worried about myocardial infarction due to high blood pressure. What medicine should I take?" The information acquisition unit 131 may also acquire text data or image data indicating the results of the health checkup of the user U as the health information. The results of the health checkup are represented by, for example, the names of the examination items in the health checkup and the examination results (numerical values ​​or evaluation ranks). The information acquisition unit 131 associates the acquired health information with the user U and stores it in the memory unit 12, and notifies the query creation unit 133 that the health information has been acquired.

[0036] The database creation unit 132 inputs the keyword extraction instruction text along with the article information into the article structuring LLM, and creates structured article data in which the article information is structured based on the text data output from the article structuring LLM, thereby creating a drug database. The keyword extraction instruction text is text that contains content specifying the text to be extracted from multiple texts included in the article information.

[0037] The database creation unit 132 creates a drug database by, for example, inputting keyword extraction instruction text that instructs the article structuring LLM to extract keywords related to drug names and symptoms from article information. The database creation unit 132 updates the drug database by arranging the acquired structured article data in any of columns, records, or fields in the drug database table stored in the storage unit 12.

[0038] Figure 4 is a diagram showing an example of keyword extraction instruction text. In the keyword extraction instruction text shown in Figure 4, the criteria for extracting text are shown categorized into major and minor categories. By inputting such keyword extraction instruction text into the thesis structuring LLM, the database creation unit 132 can acquire structured data of paper information from the thesis structuring LLM.

[0039] Fig. 5 is a diagram showing items included in the structured paper data. The items shown in Fig. 5 correspond to the items in the drug database, and the structured paper data includes names and numerical values ​​corresponding to the patient attribute information, intervention information, comparison information, and result information described above.

[0040] Fig. 6 is a diagram showing an example of structured paper data. As shown in Fig. 6, in the structured paper data, items such as the age and attributes of the subjects, the type of drug, the dosage, the effects of the drug, etc., contained in the experimental results described in the paper are associated with the content of each item.

[0041] The query creation unit 133 inputs the query creation instruction text together with the health information into the query creation LLM, thereby creating a search query for a drug database to search for drugs suitable for the user U. For example, the query creation unit 133 inputs the query creation instruction text that instructs the query creation LLM to extract keywords related to the attributes of the user U and the symptoms of the user U from the health information, thereby creating a search query that includes the age of the user U and the symptoms of the user U as keywords.

[0042] 7 is a diagram showing an example of a query creation instruction text. The query creation instruction text shown in FIG. 7 indicates that "elderly," "middle-aged," or "young" is selected as the age attribute of user U. It also indicates that male or female is selected as the gender, that "healthy" or "with complications" is selected as the health condition, and that "low blood pressure," "cardiovascular event," "death," or "other" is selected as the symptom.

[0043] The query creation unit 133 inputs the query creation instruction text shown in Figure 7 into the search query LLM along with health information such as "I am a 77-year-old man. I am worried about myocardial infarction due to high blood pressure. What medicine should I take?", and creates a search query such as {'Patient attributes': ['elderly', 'male'], 'Results': ['low blood pressure', 'cardiovascular event']}.

[0044] The query creation unit 133 may create a search query for specifying the names of other drugs that cannot be used in combination with the recommended drug, along with the name of the drug recommended to the user U. The search query includes information for searching for drugs that cannot be used in combination. The query creation unit 133 may also create a search query for specifying the countries in which the recommended drug has been approved, along with the name of the drug recommended to the user U. The search query includes information for searching for countries in which the drug has been approved.

[0045] The drug identification unit 134 identifies one or more drugs corresponding to the health information by searching the drug database using the search query created by the query creation unit 133. For example, the drug identification unit 134 identifies paper information associated with words included in the search query in the drug database, and inputs the identified paper information and health information into the drug search LLM to identify the drug name and the reason for recommending the drug output from the drug search LLM. The drug identification unit 134 inputs the identified drug name and reason to the output unit 135.

[0046] The drug identification unit 134 may search a drug database to identify the names of other drugs that cannot be used in combination with the identified drug. The drug identification unit 134 may also search a drug database to identify countries in which the identified drug is approved.

[0047] The output unit 135 outputs drug information related to the drug identified by the drug identification unit 134. The output unit 135 outputs, for example, drug information indicating the name of the drug identified by the drug identification unit 134 and the reason for recommending the drug. The output unit 135 transmits the drug information to the user terminal 2 via the communication unit 11, for example. The output unit 135 may output the drug information to a printer for printing the drug information.

[0048] Fig. 8 is a diagram showing an example of drug information output by the output unit 135. The drug information shown in Fig. 8 is composed of health information input by the user U, a search query created by the query creation unit 133, and text indicating the name of the drug identified by the drug identification unit 134 and the reason why the drug is recommended. The drug information may include an image.

[0049] The output unit 135 may output the names of other drugs that cannot be used in combination with the drug, together with the name of the drug identified by the drug identification unit 134. Furthermore, the output unit 135 may output, together with the name of the drug identified by the drug identification unit 134, information indicating the country in which the drug is approved.

[0050] [Processing flow in information providing device 1] 9 is a flowchart showing the flow of processing by which the information providing device 1 creates a paper database. When the information acquiring unit 131 acquires paper information (S11), the database creating unit 132 inputs the paper information and keyword extraction instruction text into the paper structuring LLM (S12). The database creating unit 132 acquires structured paper data output from the paper structuring LLM (S13) and updates the paper database based on the structured paper data (S14). The information providing device 1 periodically repeats the processes from S11 to S14 to update the drug database with the latest paper information.

[0051] 10 is a flowchart showing the flow of a process for outputting drug information by the information providing device 1. When the information acquiring unit 131 acquires health information (S21), the query creating unit 133 creates a search query by inputting the health information and a query creation instruction text into the query creation LLM (S22).

[0052] The drug identification unit 134 searches the drug database using the search query created by the query creation unit 133 to identify related paper information corresponding to the words indicated by the search query (S23). The drug identification unit 134 inputs the identified related paper information, the health information acquired by the information acquisition unit 131, and drug information output instruction text for instructing the output of drug information to the drug search LLM (S24). The drug identification unit 134 identifies the name of a drug recommended to user U and the reason for recommending the drug based on the text output from the drug search LLM (S25). The output unit 135 outputs drug information including text indicating the drug name and the reason for recommending the drug (S26).

[0053] [Effects of information provision device 1] As described above, the information providing device 1 has the query creating unit 133 that creates a search query for a drug database to search for drugs suitable for the user U by inputting the query creation instruction text along with the user U's health information into the query creation LLM, and the drug identifying unit 134 that identifies one or more drugs corresponding to the health information by searching the drug database using the search query. The information providing device 1 then provides information on the identified drugs to the user U. With the information providing device 1 configured in this way, the user U can easily identify drugs suitable for his or her health condition, even if he or she does not have specialized knowledge about drugs.

[0054] The present invention has been described above using embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments, and various modifications and changes are possible within the scope of the gist of the present invention. For example, all or part of the device can be configured by functionally or physically distributing or integrating any unit. Furthermore, new embodiments resulting from any combination of multiple embodiments are also included in the embodiments of the present invention. The effects of the new embodiments resulting from the combination also have the effects of the original embodiments.

[0055] For example, in the above description, the first large-scale language model, the second large-scale language model, and the third large-scale language model are each a different large-scale language model, but the information providing device 1 may use a large-scale language model that has the functions of all or some of these large-scale language models. [Explanation of symbols]

[0056] 1 Information provision device 2. User terminal 3. Paper information server 4. LLM Server for Dissertation Structure 5 Query Creation LLM Server 6 Drug Search LLM Server 11 Communications Department 12 Storage section 13 Control Unit 121 Drug Database 131 Information Acquisition Department 132 Database Creation Department 133 Query Creation Section 134 Drug Specification Department 135 Output section

Claims

1. an information acquisition unit that acquires health information indicating the health condition of the user; a query generation unit that generates a search query for a drug database to search for a drug suitable for the user by inputting a first instruction text together with the health information into a first large-scale language model; a drug identification unit that identifies one or more drugs corresponding to the health information by searching the drug database using the search query; an output unit that outputs drug information regarding the drug identified by the drug identification unit; An information providing device having the above.

2. the query creation unit creates the search query including the age of the user and the symptoms of the user as keywords by inputting the first instruction text, which instructs the creation of the search query including keywords related to the attributes of the user and the symptoms of the user, into the first large-scale language model; The information providing device according to claim 1 .

3. the information acquisition unit further acquires paper information; the information providing device further includes a database creation unit that inputs the second instruction text together with the article information into a second large-scale language model, and creates structured article data in which the article information is structured, thereby creating the drug database. The information providing device according to claim 1 .

4. the database creation unit creates the drug database by inputting the second instruction text, which instructs the extraction of drug names and keywords related to symptoms from the paper information, into the second large-scale language model. The information providing device according to claim 3 .

5. the drug identification unit identifies the article information associated with the words included in the search query in the drug database, and inputs the identified article information, the health information, and third instruction text instructing to output the drug information into a third large-scale language model, thereby identifying a drug name and a reason for recommending the drug, which are output from the third large-scale language model; The output unit outputs the drug information including the drug name and reason identified by the drug identification unit. The information providing device according to claim 3 .

6. In the drug database, each of the plurality of drugs is associated with other drugs that cannot be used in combination, The output unit outputs the names of other drugs that cannot be used in combination with the name of the drug identified by the drug identification unit. The information providing device according to claim 1 .

7. In the drug database, each of a plurality of drugs is associated with a country in which the drug is approved; The output unit outputs information indicating a country in which the drug is approved together with the name of the drug identified by the drug identification unit. The information providing device according to claim 1 .

8. The computer executes acquiring health information indicative of a health condition of a user; inputting the first instructional text along with the health information into a first large-scale language model to generate a search query for a medication database to search for medications appropriate for the user; searching the medication database using the search query to identify one or more medications corresponding to the health information; outputting drug information regarding the identified drug; An information provision method having the above.

9. On the computer, acquiring health information indicative of a health condition of a user; inputting the first instructional text along with the health information into a first large-scale language model to generate a search query for a medication database to search for medications appropriate for the user; searching the medication database using the search query to identify one or more medications corresponding to the health information; outputting drug information regarding the identified drug; A program to execute.

Images