Medical document presentation support device, medical document presentation support system, medical document presentation support method, and program

Abstract

This invention provides a medical document presentation support device, a medical document presentation support system, a medical document presentation support method, and a program that can present test results related to medical document data generated from source data. [Solution] A medical document presentation support device comprising: an examination request data acquisition unit that acquires examination request data including original data including input information, SOAP format medical document data generated based on the original data, and criterion information regarding whether or not it is based on the input information; and an examination result data output request unit that requests the examination result data to be output from the examination request data to an examination result data output device that outputs examination result data related to the medical document data based on a trained model that has been trained to output output data corresponding to the input data.

Description

Technical Field

[0001] The present disclosure relates to a medical document presentation support device, a medical document presentation support system, a medical document presentation support method, and a program.

Background Art

[0002] In the utilization of generative AI (Artificial Intelligence), an event called hallucination becomes a problem. Hallucination is an event in which generative AI probabilistically generates information based on the content it has learned so far, and generates and outputs information that does not correspond to the current event as if it were true.

[0003] In particular, the occurrence of hallucinations in the medical field has a high requirement to be avoided, and if incorrect medical information that does not rely on facts is generated, it may affect the health of patients. [[ID=ig]] Specifically, when generating a medical document (for example, a medication history) from the conversation content between a patient and a medical staff by a generative AI, if there is no information input with appropriate accuracy, or if some input information is missing, a situation may occur where the generative AI fabricates or analogizes the conversation content of the medical staff or the patient, and creates a medical document based on the fabricated or analogized information.

[0004] [[ID=2k]]Conventionally, as a countermeasure against such hallucinations, the information generated by the generative AI is visually confirmed by a human, and the human checks whether hallucinations have occurred.

[0005] Note that Patent Document 1 discloses a support device for assisting a pharmacist in providing medication guidance to a patient (see Patent Document 1).

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

[0007] However, with the conventional technologies described above, when a human needs to check whether hallucination has occurred, it is necessary to compare the input information with the generated result, which can make the process cumbersome. With the conventional technologies described above, the amount of information that a human needs to check is large, and the quality of the check results depends on the effort of that human, so there is a possibility that checks may be missed.

[0008] This disclosure is made in consideration of these circumstances and can present the results of an inspection regarding the accuracy of medical document data generated from the original data. The objective is to provide a medical document presentation support device, a medical document presentation support system, a medical document presentation support method, and a program. [Means for solving the problem]

[0009] One embodiment is a medical document presentation support device comprising: an examination request data acquisition unit that acquires examination request data including raw data including input information, SOAP-format medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information; and an examination result data output request unit that requests an examination result data output device, which outputs examination result data related to the medical document data based on a trained model that has been trained to output output data corresponding to the input data, to output the examination result data from the examination request data.

[0010] One embodiment is a medical document presentation support system comprising a terminal device having a medical document presentation support device, a presentation request unit that receives presentation requests from a user, and a presentation unit that presents data corresponding to the presentation request.

[0011] One embodiment is a medical document presentation support method comprising: an examination request data acquisition step of acquiring examination request data including raw data including input information, SOAP-format medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information; and an examination result data output request step of requesting an examination result data output device, which outputs examination result data relating to the medical document data based on a trained model that has been trained to output output data corresponding to the input data, to output the examination result data from the examination request data.

[0012] One embodiment is a program that causes a computer to execute an examination request data acquisition step, which acquires examination request data including raw data including input information, SOAP-format medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information; and an examination result data output request step, which requests an examination result data output device, which outputs examination result data relating to the medical document data based on a trained model that has been trained to output output data corresponding to the input data, to output the examination result data from the examination request data. [Effects of the Invention]

[0013] According to this disclosure, a medical document presentation support device, a medical document presentation support system, a medical document presentation support method, and a program can present inspection results regarding the accuracy of medical document data generated from source data. [Brief explanation of the drawing]

[0014] [Figure 1] This figure shows an example of the configuration of a medical document presentation support system according to the embodiment of this disclosure. [Figure 2] This figure shows an example of the functional configuration of a medical document presentation support system according to the present disclosure. [Figure 3A] This figure shows an example of the criteria for inspecting S information according to the embodiments of this disclosure. [Figure 3B]It is a diagram showing an example of a standard for inspection related to P information according to an embodiment of the present disclosure. [Figure 4A] It is a diagram showing an example of the flow of medical document generation processing according to an embodiment of the present disclosure. [Figure 4B] It is a diagram showing an example of the flow of medical document inspection processing according to an embodiment of the present disclosure. [Figure 5A] It is a diagram showing an example of text data related to EP information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 5B] It is a diagram showing an example of text data related to EP information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 5C] It is a diagram showing an example of text data related to EP information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 6] It is a diagram showing an example of text data related to EP information generated as inspection result data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 7A] It is a diagram showing an example of text data related to S information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 7B] It is a diagram showing an example of text data related to S information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 7C] It is a diagram showing an example of text data related to S information generated as inspection request data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 8] It is a diagram showing an example of text data related to S information generated as inspection result data in medical document inspection processing according to an embodiment of the present disclosure. [Figure 9] It is a diagram showing an example of the functional configuration of a medical document presentation support system according to a modified example of an embodiment of the present disclosure.

Modes for Carrying Out the Invention

[0015] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.

[0016] (Embodiment) Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings.

[0017] [Medical Document Presentation Support System] FIG. 1 is a diagram showing an example of the configuration of a medical document presentation support system 1 according to an embodiment of the present disclosure. The medical document presentation support system 1 is a system that supports the presentation of medical document data presented to a pharmacist. The medical document presentation support system 1 includes a user terminal device 2 and a medical document presentation support device 3. Also, FIG. 1 shows a user U1 of the user terminal device 2, a generation AI server 4 used for generating medical document data, and a generation AI server 5 used for inspecting medical document data.

[0018] Here, in the present embodiment, the user U1 is a pharmacist, but as another example, it may be a person who operates the user terminal device 2 on behalf of the pharmacist. Also, in the present embodiment, for the sake of convenience of explanation, a case where the generation AI server 4 and the generation AI server 5 are separate generation AI servers is shown, but they may be a common generation AI server.

[0019] The user terminal device 2 is a terminal device used by the user U1. The user terminal device 2 is a computer device such as a personal computer, a tablet, or a smartphone. The user terminal device 2 includes a display unit 21 and an operation unit 22. The display unit 21 includes, for example, a liquid crystal display and displays various screens. In the following description, the display unit 21 displaying a screen is also referred to as presenting a screen to the user U1. The operation unit 22 includes, for example, a touch panel and detects the operation of the user U1.

[0020] The user terminal device 2 comprises a calculation unit 200 and a storage unit 250. The arithmetic unit 200 includes, for example, a central processing unit (CPU), and operates based on programs and data stored in the storage unit 250, providing various functions. The storage unit 250 is composed of, for example, one or both of a hard disk drive and semiconductor memory (flash memory), RAM (Random Access Memory), ROM (Read Only Memory), etc., and stores various types of information, such as programs and data read by the arithmetic unit 200. The storage unit 250 may be implemented by a storage device such as a cloud server located outside the user terminal device 2. This storage device may be, for example, a virtual storage device.

[0021] The medical document presentation support device 3, the generation AI server 4, and the generation AI server 5 are, for example, cloud servers. The medical document presentation support device 3, the generation AI server 4, and the generation AI server 5 may each be implemented as a single computer or as virtual servers.

[0022] The medical document presentation support device 3 comprises a calculation unit 300 and a storage unit 350. The arithmetic unit 300, for example, includes a CPU and operates based on programs and data stored in the storage unit 350, providing various functions. The storage unit 350 is composed of, for example, one or both of a hard disk drive and semiconductor memory (flash memory), RAM, ROM, etc., and stores various types of information, such as programs and data read by the arithmetic unit 300. The storage unit 350 may also be implemented by a storage device such as a cloud server located outside the medical document presentation support device 3. This storage device may be, for example, a virtual storage device.

[0023] The user terminal device 2, the medical document presentation support device 3, the generation AI server 4, and the generation AI server 5 can communicate with each other via a network. This network may include, for example, the internet, a WAN (Wide Area Network), a LAN (Local Area Network), a public telephone line, provider equipment, a dedicated line, or a wireless base station.

[0024] User U1 operates the control unit 22 of the user terminal device 2 to send presentation request A1 to the medical document presentation support device 3. Request A1 is information requesting the presentation of medical document data corresponding to the original data. In this embodiment, user U1 can specify whether or not an examination is required for the medical document data to be presented. In this embodiment, this specification information is included in presentation request A1.

[0025] As another example, a configuration may be used in which user U1 does not have the function to specify whether or not to perform tests on medical document data. In this case, for example, presentation request A1 is always treated as a request for the test to be performed.

[0026] When the medical document presentation support device 3 receives a presentation request A1 from the user terminal device 2, it sends the generation request data B1 to the generation AI server 4. The generation request data B1 is information that requests the generation AI server 4 to generate and output the generation result data C1, which is medical document data, and is used as input for the generation AI server 4 to generate and output the generation result data C1.

[0027] The medical document presentation support device 3 acquires the generated result data C1 output by the generation AI server 4. The generated result data C1 is data representing a medical document (medical document data). In this embodiment, the generated result data C1 is text data.

[0028] Here, as an example, we will explain the case where the medical document data related to the generated result data C1 is set to "checked". In this case, when the medical document presentation support device 3 receives the generated result data C1 from the generating AI server 4, it sends the examination request data D1 to the generating AI server 5. The inspection request data D1 is information that requests the generating AI server 5 to inspect the generated result data C1, generate an inspection result, and output it. It is used as input for the generating AI server 5 to generate and output the inspection result data E1.

[0029] The medical document presentation support device 3 acquires the test result data E1 output by the generating AI server 5. The test result data E1 is data that shows the test results of a medical document. In this embodiment, the test result data E1 is text data.

[0030] The medical document presentation support device 3 generates result data F1, which is data that has been processed into a format for presentation to the user terminal device 2, from the generated result data C1 and the test result data E1. The medical document presentation support device 3 transmits the result data F1 to the user terminal device 2. The user terminal device 2 receives the result data F1 and displays the result data F1 on the display unit 21.

[0031] On the other hand, as another example, we will explain the case where no tests are performed on the medical document data related to the generated result data C1. In this case, the medical document presentation support device 3 generates result data F1, which is data that has been processed from the generated result data C1 into a format for presentation to the user terminal device 2. The medical document presentation support device 3 transmits the result data F1 to the user terminal device 2. The user terminal device 2 receives the result data F1 and displays the result data F1 on the display unit 21.

[0032] Thus, in this embodiment, the medical document presentation support device 3 has the function of an AI generation agent that generates medical document data using a generation AI, and the function of an AI inspection agent that performs inspections on the medical document data using the generation AI.

[0033] <Functional Configuration of the Medical Document Presentation Support System> Figure 2 shows an example of the functional configuration of the medical document presentation support system 1 according to the present disclosure. The user terminal device 2 comprises a presentation request unit 201 and a presentation unit 202. Each of these functional units is realized, for example, by the CPU in the arithmetic unit 200 loading a program read from the ROM in the storage unit 250 into the RAM in the storage unit 250, and then executing processing according to that program. The presentation request unit 201 transmits presentation request A1 to the medical document presentation support device 3. The display unit 202 displays the result data F1.

[0034] In this embodiment, the presentation request A1 includes the data that will be used to generate the medical document data (for convenience of explanation, this will also be called the source data). The source data includes information referenced when generating medical document data (for convenience of explanation, this will also be called input information). Furthermore, request A1 may include other information that is referenced when generating medical document data. Furthermore, request A1 may include other information that is referenced when examining medical document data. Furthermore, request A1 may include information specifying whether or not an examination is required.

[0035] The medical document presentation support device 3 comprises a presentation request reception unit 30, a generation request data acquisition unit 31, a generation result data output request unit 32, a generation result data acquisition unit 33, an examination viewpoint determination unit 34, an examination request data acquisition unit 35, an examination result data output request unit 36, an examination result data acquisition unit 37, a result data output unit 38, an examination presence / absence determination unit 51, and a standard definition information management unit 52. Each of these functional units is realized, for example, by the CPU in the calculation unit 300 reading a program from the ROM in the storage unit 350, expanding it into the RAM in the storage unit 350, and executing processing according to the program. These functional units may be distributed across multiple servers. In other words, the medical document presentation support device 3 may be configured across multiple devices.

[0036] The presentation request receiving unit 30 receives presentation request A1 from the user terminal device 2. The inspection status determination unit 51 determines whether or not an inspection has been performed on the medical document data based on the presentation request A1. In this embodiment, if there is an inspection of the medical document data, the inspection of the medical document data generated by the generation AI server 4 is performed by the generation AI server 5. On the other hand, in this embodiment, if no inspection is performed on the medical document data, the medical document data generated by the generation AI server 4 will not be inspected. Furthermore, in configurations where inspection of medical document data is always performed, the medical document presentation support device 3 does not need to be equipped with an inspection presence / absence determination unit 51.

[0037] The generation request data acquisition unit 31 acquires the generation request data B1. Here, the generation request data B1 includes, for example, the source data for medical document data. Furthermore, the generation request data B1 may include other information that is referenced when generating medical document data.

[0038] The generation result data output request unit 32 requests the generation AI server 4 to output the generation result data C1 from the generation request data B1.

[0039] The generation result data acquisition unit 33 acquires the generation result data C1 output by the generation AI server 4. The generated result data C1 includes the generated medical document data. In this embodiment, the medical document data is in SOAP format. In SOAP format, S (Subjective Data) represents subjective data, O (Objective Data) represents objective data, A (Assessment) represents analysis and evaluation, and P (Plan) represents the plan. Furthermore, P information includes EP (Educational Plan), CP (Care Plan), and OP (Observational Plan).

[0040] The Standards and Regulations Information Management Unit 52 manages the information that defines the standards and regulations (for convenience of explanation, this will also be called standards and regulations information). In this embodiment, the standard information management unit 52 stores and manages the standard information in the storage unit 350. The standard information management unit 52 may be provided as a separate server from the medical document presentation support device 3. In this case, the standard information management unit 52 does not need to be provided in the medical document presentation support device 3.

[0041] The inspection perspective determination unit 34 determines the inspection perspective that indicates the inspection perspective related to the medical document data. In this embodiment, the inspection perspective includes criteria information regarding whether the medical document data is based on input information (content of the original data). In this embodiment, the inspection perspective determination unit 34 determines the reference information to be used as an inspection perspective from among multiple reference information included in the reference information managed by the reference information management unit 52. This enables dynamic control of the criteria regarding whether the medical document data is based on input information (content of the original data). In this embodiment, the certainty of whether the medical document data is based on the input information (content of the original data) is used as the certainty of the medical document data.

[0042] In this embodiment, we show a case where information (reference information) regarding whether the medical document data is based on the input information (content of the original data) is used as the inspection perspective. However, for example, the inspection perspective may include information on other perspectives related to the inspection, along with the information on the said perspective. The functions of the inspection perspective determination unit 34 may, for example, be included in the inspection request data acquisition unit 35.

[0043] The inspection request data acquisition unit 35 acquires the inspection request data D1. Here, the inspection request data D1 includes the original data, the medical document data generated from the original data, and the criteria information regarding whether the medical document data is based on the input information (content of the original data). In this embodiment, the inspection request data acquisition unit 35 includes the result (reference information) determined by the inspection viewpoint determination unit 34 in the inspection request data D1.

[0044] The inspection result data output request unit 36 ​​requests the generating AI server 5 to output inspection result data E1 from the inspection request data D1.

[0045] The inspection result data acquisition unit 37 acquires the inspection result data E1 output by the generating AI server 5. In this embodiment, the test result data E1 includes the test results for each item in SOAP format, and also includes information regarding the accuracy of the contents of the medical document data (accuracy information).

[0046] If there is an inspection of the medical document data, processing to obtain the results of said inspection is performed (processing performed by the inspection perspective determination unit 34, the inspection request data acquisition unit 35, the inspection result data output request unit 36, and the inspection result data acquisition unit 37). On the other hand, if there is no inspection of medical document data, the processing to obtain the results of said inspection (processing performed by the inspection perspective determination unit 34, the inspection request data acquisition unit 35, the inspection result data output request unit 36, and the inspection result data acquisition unit 37) is not performed.

[0047] The result data output unit 38 outputs the data generated based on the generated result data C1 and the test result data E1 as result data F1 if there is an examination related to the medical document data. As an example, the result data output unit 38 converts the generated result data C1 and the inspection result data E1 into result data F1 and outputs it to the user terminal device 2. Here, the result data output unit 38 does not necessarily have to output the generated result data C1 and the inspection result data E1 together as result data F1 to the user terminal device 2. For example, the result data F1 may be divided into multiple data sets and output to the user terminal device 2.

[0048] As a specific example, the result data output unit 38 may output separately to the user terminal device 2 the data converted from the generated result data C1 (referred to as the first result data F1A for convenience of explanation) and the data converted from the inspection result data E1 (referred to as the second result data F1B for convenience of explanation). In this case, for example, when the generated result data C1 is acquired, the result data output unit 38 may output the first result data F1A converted from the generated result data C1 to the user terminal device 2, and then, when the inspection result data E1 is obtained, it may output the second result data F1B converted from the inspection result data E1 to the user terminal device 2. In this case, result data F1 can be considered to consist of result data A F1A and result data B F1B.

[0049] On the other hand, if there are no tests related to the medical document data, the result data output unit 38 outputs the data generated based on the generated result data C1 as result data F1. As an example, the result data output unit 38 converts the generated result data C1 into result data F1 and outputs it to the user terminal device 2.

[0050] The generation AI server 4 has a trained model of generation AI, and outputs output data according to the input data using this trained model. In this embodiment, when generation request data B1 is input to the generation AI server 4, it outputs generation result data C1 that shows the result of generating medical document data. The generation AI server 5 has a trained model of generation AI, and outputs output data according to the input data using this trained model. In this embodiment, when the generation AI server 5 receives the inspection request data D1, it outputs inspection result data E1 which shows the inspection result of the medical document data.

[0051] Here, generative AI refers to AI technology that generates new content based on diverse input modalities (text, images, audio, etc.), such as text generation AI based on large language models (LLMs) including GPT (Generative Pretrained Transformer). In addition, various publicly available AI generation services may be used as either or both of the AI ​​generation server 4 and AI generation server 5.

[0052] Generally speaking, in a learning model, the input data used during training is different from the input data used during inference processing (i.e., the processing in which the Generative AI Server 4 or Generative AI Server 5 outputs the inference result data).

[0053] Furthermore, in the training of generative AI, supervised learning is performed using, for example, a vast amount of text from the internet as training data. Reinforcement learning may also be combined with supervised learning in the training of generative AI. The vast amount of text used to train generative AI includes texts about medical information and other topics. Generative AI is an example of a pre-trained model that has been trained to produce output data corresponding to the input data. Furthermore, the generation AI server 4 is an example of a device for outputting generated result data. Furthermore, the generating AI server 5 is an example of a device for outputting inspection result data.

[0054] <Overview of the AI ​​examination agent in the medical document presentation support device> In this embodiment, after inputting some input information into the generating AI to generate medical document data in SOAP format, the AI ​​inspection agent (a function of the medical document presentation support device 3) is used to check whether the generated result is based on the input information. In this embodiment, at least one of the following items from S information and P information (which includes EP information, CP information, and OP information, for example, EP information) is subject to inspection.

[0055] In this embodiment, the AI ​​inspection agent is configured as an application (e.g., a server application) that inputs information into a generating AI service to obtain the intended output. In this embodiment, the AI ​​inspection agent is instructed to input the input information (content of the original data) and the generated result (medical document data) when SOAP-formatted medical document data is generated, and to inspect (confirm) whether the medical document data contains any content that is not based on the input information.

[0056] In this process, the AI ​​inspection agent dynamically changes the criteria for whether or not the input information is based on each item in the SOAP format. Here, the criteria used to determine whether something is based on input information are, for example, the criteria used to determine whether something is based on input information (which may also be called a determination condition).

[0057] As a concrete example, since S information is subjective information from the patient, the system may be configured to determine that the information is based on the input information only if there is information in the input information that directly matches the content described in the medical document data. Furthermore, regarding S information, even if there is no information in the input information that directly matches the content described in the medical document data, the system may be configured to determine that the information is based on the input information if there is information that allows for the inference of similar content. As a concrete example, since EP information is the content of instructions given to patients by healthcare professionals (such as pharmacists), even if there is no information in the input data that directly matches the content described in the medical document data, if there is information that allows for the inference of similar content, the system may be configured to determine that the information is based on the input data.

[0058] Furthermore, the system may be configured to dynamically change the criteria for whether or not the input information is used, depending on the completeness of the input information. In this embodiment, the level of completeness of the input information is generally defined as follows: The level of completeness of input information for the first type refers to the degree to which the information used to generate SOAP-formatted medical document data includes the content of conversations between the patient and healthcare professionals. The level of completeness of input information for the second type refers to the level of completeness in the information used to generate SOAP-format medical document data, specifically, the level of completeness in which only the utterances of the healthcare professional are included in the dialogue between the patient and the healthcare professional. The level of completeness of input information for the third type refers to the level of completeness in the information used to generate SOAP-formatted medical document data, where it includes only the content of the conversation between the patient and healthcare professional summarized by the pharmacist. The content of the summary may, for example, be entered by voice input by the pharmacist.

[0059] As a concrete example, consider a case where the input information consists only of the pharmacist's spoken portion of a conversation between a patient and a pharmacist during medication guidance. In other words, while S information is subjective information from the patient, the input information does not include patient speech information, so it is not possible to confirm information that directly matches the content described in the medical document data from the input information. Therefore, even if the input information does not directly match the content described in the medical document data, if there is information that allows similar content to be inferred from the flow of the medical professional's speech (for example, affirmation or denial in response to the medical professional's question), the system may be configured to determine that the information is based on the input information.

[0060] The test results obtained by the AI ​​testing agent may include information with a degree of certainty such as "completely wrong" because it is not based on the input information, or information with a degree of certainty such as "there is no information that clearly bases the input information, but there is information that can be inferred, so it cannot be definitively said to be wrong." The test results may be output in a structured format, such as JSON.

[0061] Here, there are no particular limitations on the accuracy information, but for example, three levels of accuracy information such as (accuracy information A), (accuracy information B), and (accuracy information C) may be used. (Accuracy Information A) is information that can be determined to be entirely based on the input information. (Accuracy Information B) is information that cannot be judged to be entirely based on the input information, but can be judged to be based on the input information with a certain degree of accuracy by analogy with related information. (Accuracy Information C) indicates that the information cannot be determined to be based on the input information.

[0062] <Examples of user interfaces, etc.> In this embodiment, the user interface (UI) or application interaction that displays the information output by the AI ​​examination agent (function of the medical document presentation support device 3) to medical professionals on an application equipped with a medical information recording function is configured as follows. In other words, in the UI or application interaction, for each SOAP-formatted item, processing such as annotation is performed on the screen UI according to whether the medical document data is based on the input information and the accuracy of that information. This configuration enhances visibility and readability, allowing user U1 to instantly judge, for example, items that are not based on input information and have low accuracy. Furthermore, the configuration allows user U1 to check the information that forms the basis of their judgment for each SOAP-formatted item with minimal operation, and enables healthcare professionals to gradually delve deeper and confirm whether the generated results are accurate. In this embodiment, such an application is an application (client application) of the user terminal device 2.

[0063] <Example of standard / regulation information> Refer to Figures 3A and 3B to illustrate examples of criteria for determining whether or not something is based on input information. Figures 3A and 3B show standard specification information that defines multiple standards. In this embodiment, each of the multiple criteria is identified by the degree of completeness of the SOAP-formatted items (S, EP, CP, OP in this embodiment) and the input information (content of the source data).

[0064] Figure 3A shows an example of the criteria for inspection of S information according to the embodiment of this disclosure. Table T1 shows an example of the criteria for testing S information. In this embodiment, the information in table T1 is managed by the standard definition information management unit 52.

[0065] Regarding Figure 3A, we will now explain the S information among the SOAP format items. The level of completeness of the input information can be categorized as follows: completeness includes "the content of the conversation between the patient and the healthcare professional"; completeness includes "only the content spoken by the healthcare professional from the conversation between the patient and the healthcare professional"; and completeness includes "a summary of the conversation between the patient and the healthcare professional by a pharmacist".

[0066] Regarding S information, if the level of completeness is such that "the content of the dialogue between the patient and healthcare worker" is present, the criteria for determining whether it is based on input information are either "there is direct content in the input information that matches the S information as a statement made by the patient" or "even if there is no direct content that matches the S information, there is information that allows the patient's statement to be inferred from the pharmacist's utterances."

[0067] Regarding S information, if the level of completeness is such that "only the content of the healthcare worker's statements from the conversation between the patient and the healthcare worker is available," the criterion for determining whether it is based on input information is "even if there is no direct content in the input information that matches the S information, there is information that allows the patient's statements to be inferred from the pharmacist's statements."

[0068] Regarding S information, if the information is comprehensive enough to include a summary of the conversation between the patient and healthcare professional, the criterion for determining whether it is based on the input information is that "even if there is no direct content in the input information that matches the S information, there is information in the pharmacist's summary that allows us to infer that it is based on the patient's statements."

[0069] Figure 3B shows an example of the criteria for inspection of P information according to the embodiment of this disclosure. Table T2 shows an example of criteria for testing P information. In this embodiment, the information in table T2 is managed by the standard definition information management unit 52.

[0070] Regarding Figure 3B, we will now explain the P information among the SOAP format items. P information includes EP information, CP information, and OP information. The level of completeness of the input information can be categorized as follows: completeness includes "the content of the conversation between the patient and the healthcare professional"; completeness includes "only the content spoken by the healthcare professional from the conversation between the patient and the healthcare professional"; and completeness includes "a summary of the conversation between the patient and the healthcare professional by a pharmacist".

[0071] Regarding EP information, CP information, and OP information, if the level of completeness is such that "the content of the dialogue between the patient and healthcare worker" is present, the criteria for determining whether it is based on input information are: "there is direct content in the input information that matches the target information (in this case, EP information, CP information, or OP information)" or "even if there is no direct content in the input information, there is information that allows for inference."

[0072] Regarding EP information, CP information, and OP information, if the completeness is such that "only the content of the medical professional's utterances from the patient-medical professional's dialogue is present," the criteria for determining whether it is based on input information are either "there is direct content in the input information that matches the target information (in this case, EP information, CP information, or OP information)" or "even if there is no direct content in the input information, there is information that allows for inference."

[0073] For both EP information and CP information, if the level of completeness is such that "the pharmacist has summarized the content of the conversation between the patient and healthcare professional," then the criterion used is "there is direct content in the input information that matches the target information (in this case, EP information or CP information)."

[0074] Regarding OP information, if the information is comprehensive enough to include a summary of the conversation between the patient and healthcare professional by the pharmacist, the criteria used will be either "there is direct content in the input information that matches the OP information" or "even if there is no direct content in the input information, there is information that allows for inference."

[0075] In this embodiment, the inspection viewpoint determination unit 34 determines to adopt one set of reference information based on the input information, using the reference information managed by the reference information management unit 52 (for example, the information in tables T1 and T2 shown in Figures 3A and 3B).

[0076] As another example, the inspection perspective determination unit 34 may determine a single criterion from the predetermined information based on a trained model that has been trained to output a single criterion from the predetermined information when the predetermined information is input. The predetermined information may include, for example, information that can identify items in SOAP format and information that can identify the degree of completeness of the input information.

[0077] [Example of medical document presentation support processing] Referring to Figures 4A and 4B, the medical document presentation support process, which is a process that supports the presentation of medical documents by the medical document presentation support system 1, will be explained. In this embodiment, the medical document presentation support process includes a medical document generation process and a medical document inspection process.

[0078] Figure 4A is a diagram showing an example of the flow of medical document generation processing according to the embodiment of this disclosure. Step S1: The presentation request receiving unit 30 receives presentation request A1 from the user terminal device 2. Here, user U1 operates the operation unit 22 of the user terminal device 2 to send presentation request A1 to the medical document presentation support device 3. The operation by the operation unit 22 is, for example, a predetermined operation (such as tapping or clicking) performed on a button that instructs the presentation of a medical document displayed on the display unit 21.

[0079] Request A1 includes information requesting the presentation of medical document data. User U1 instructs the user terminal device 2 to generate medical document data regarding the desired medical content through the application's UI. The input from user U1 includes the source data used to generate the medical document data. Based on the content of this source data, the completeness of the input information can be determined. For example, user U1 may explicitly specify the completeness of the input information, and the content of that specification may be included in presentation request A1. Furthermore, in this embodiment, the input from user U1 may include information specifying whether or not the inspection should be performed (whether or not an inspection should be performed).

[0080] Step S2: The generation request data acquisition unit 31 acquires generation request data B1, which includes an instruction to generate medical document data. In this embodiment, the generation request data acquisition unit 31 generates text data as generation request data B1. This text data serves as a prompt to give instructions (or commands) to the text generation AI.

[0081] Step S3: The generation result data output request unit 32 requests the generation AI server 4 to output the generation result data C1 from the generation request data B1. Here, the generation result data output request unit 32 transmits the generation request data B1 generated by the generation request data acquisition unit 31 to the generation AI server 4. In this embodiment, the medical document presentation support device 3 issues instructions for generating medical document data via a predetermined application (backend application). The generation AI server 4 generates the generation result data C1 from the generation request data B1 based on the generation AI.

[0082] Step S4: The generation result data acquisition unit 33 acquires the generation result data C1 output by the generation AI server 4.

[0083] Figure 4B shows an example of the flow of medical document inspection processing according to an embodiment of this disclosure. In this embodiment, the processing shown in Figure 4A is performed first, followed by the processing shown in Figure 4B.

[0084] Step S11: The inspection presence / absence determination unit 51 determines whether an inspection is required. If there is no inspection, the process proceeds to step S15. If there is an inspection, the process proceeds to step S12.

[0085] Step S12: The inspection request data acquisition unit 35 acquires inspection request data D1 which includes the standard information determined by the inspection viewpoint determination unit 34 as an inspection viewpoint. In this embodiment, the inspection request data acquisition unit 35 includes the reference information determined by the inspection viewpoint determination unit 34 in the text data generated as inspection request data D1.

[0086] Here, the medical document presentation support device 3, using an AI examination agent, dynamically determines the perspectives to be examined based on the characteristics of the input information that formed the basis of the medical document generation, and each item in the SOAP format. The test request data D1 includes the test instructions, SOAP-formatted medical document data, the source data from which the medical document data was generated (source data including input information), and reference information. In this embodiment, the inspection request data acquisition unit 35 generates text data as inspection request data D1. This text data serves as a prompt to give instructions (or commands) to the text generation AI.

[0087] Step S13: The inspection result data output request unit 36 ​​requests the generating AI server 5 to output inspection result data E1 from the inspection request data D1. Here, the inspection result data output request unit 36 ​​sends the inspection request data D1 generated by the inspection request data acquisition unit 35 to the generating AI server 5. The generation AI server 5 generates inspection result data E1 from the inspection request data D1 based on the generation AI.

[0088] Step S14: The inspection result data acquisition unit 37 acquires the inspection result data E1 output by the generating AI server 5. The test result data E1 includes the test results for each item in SOAP format, as well as the accuracy of the information described.

[0089] Step S15: The result data output unit 38 converts the generated result data C1 and the test result data E1 into result data F1. This conversion may be performed together, for example, for the generated result data C1 and the test result data E1, or it may be performed at different times. The result data F1 is data in a format suitable for display on the display unit 21 of the user terminal device 2. The process of converting the generated result data C1 and the inspection result data E1 into result data F1 may include processing, editing, or formatting. Furthermore, the result data output unit 38 outputs the result data F1 to the user terminal device 2. The result data F1 includes the generated medical document data, and the test results and information accuracy for each item in said medical document data. As a result, the presentation unit 202 presents the result data F1 in the user terminal device 2. The presentation unit 202 displays the result data F1 on the display unit 21, for example. The presentation unit 202 may present the result data F1 as is, or it may present the result of editing the result data F1. Based on the above, the medical document presentation support system 1 terminates the medical document presentation support process.

[0090] Here, the application on the user terminal device 2 may provide a UI that enhances visibility and readability, for example, by adding annotations to each test result and information accuracy related to the medical document data. In this application, the medical document data, test results, and information accuracy are displayed in an optimal manner as information provided to user U1.

[0091] The inspection criteria (in this embodiment, inspection criteria including reference information) may be specified by user U1. In this case, user U1 inputs the inspection criteria by operating the operation unit 22 of the user terminal device 2. The inspection criteria may be input as text, for example, or selected from a list of options. If the examination viewpoint is specified by user U1, the medical document presentation support device 3 does not need to be equipped with an examination viewpoint determination unit 34. In this case, the examination viewpoint specified by user U1 is used instead of the determination result by the examination viewpoint determination unit 34 in this embodiment. Further, the inspection perspective may be configured by, for example, combining the inspection perspective specified by the user U1 and the inspection perspective determined by the inspection perspective determination unit 34.

[0092] [Input Examples and Output Examples of the Generation AI for Inspection] Examples of input data and output data to the generation AI for inspection (in this embodiment, the generation AI server 5) are shown.

[0093] <Premise> The following examples describe the inspection of EP information and the inspection of S information in a SOAP-formatted medication history. However, CP information and OP information can also be inspected by following the format and structure of this example. Also, this example shows the case where only EP information is inspected independently and the case where only S information is inspected independently. As another example, it is also possible to simultaneously inspect two or more of EP information, CP information, OP information, and S information. That is, regardless of the processing stage or granularity and the number of times, appropriate dynamic control of the input regarding these information can be given to obtain the output in the desired form.

[0094] <Input Example of Inspection of EP Information> FIG. 5A, FIG. 5B, and FIG. 5C are diagrams showing an example of text data regarding EP information generated as inspection request data D1 in the medical document inspection process according to an embodiment of the present disclosure. This example shows the case where there is substantial content of the conversation between the pharmacist and the patient, that is, the case of "the conversation content between the patient and the medical staff". Note that the input texts P1, P2, and P3 shown in FIGS. 5A, 5B, and 5C respectively are shown in multiple drawings, but they are text data grouped together.

[0095] The input text P1 includes, for example, inspection instructions, information sufficiency (a description indicating that it is the content of the conversation between the pharmacist and the patient regarding medication guidance), criteria for whether it is based on the input information, rules at the time of output, and evaluation properties (properties for evaluation). The input text P2 includes, for example, an instruction to convert the output text into a predetermined format (as an example, JSON format). This instruction includes text as an example to be used for the generative AI to imitate when generating the output text. The input text P3 includes, for example, input information (medication guidance content) and EP information of a SOAP-formatted medication history created based on the input information.

[0096] <Example of output of EP information inspection> FIG. 6 is a diagram showing an example of text data regarding EP information generated as inspection result data E1 in the medical document inspection process according to an embodiment of the present disclosure. In this example, examples of outputs for the inputs shown in FIGS. 5A, 5B, and 5C are shown. The output text P11 includes text in a predetermined format (as an example, JSON format).

[0097] <Example of input of S information inspection> FIGS. 7A, 7B, and 7C are diagrams showing an example of text data regarding S information generated as inspection request data D1 in the medical document inspection process according to an embodiment of the present disclosure. In this example, a case of sufficiency where there is only the content of the pharmacist's speech in the conversation between the pharmacist and the patient is shown, that is, a case of "only the speech content of the medical staff among the conversation content between the patient and the medical staff" is shown. Note that the input texts P21, P22, and P23 shown in FIGS. 7A, 7B, and 7C respectively are shown separately in multiple drawings, but they are a single integrated text data.

[0098] The input text P21 includes, for example, inspection instructions, information sufficiency (a description indicating that it is only the pharmacist's speech content among the conversation content of the pharmacist's medication guidance for the patient), criteria for whether it is based on the input information, rules at the time of output, and evaluation properties (properties for evaluation). The input text P22 includes, for example, an instruction to convert the output text into a predetermined format (as an example, the JSON format). This instruction includes text as an example to be used for the generative AI to imitate when generating the output text. The input text P23 includes, for example, input information (medication guidance content) and S information of the SOAP - formatted medical history created based on the input information.

[0099] <Example of the output of the inspection of S information> FIG. 8 is a diagram showing an example of text data related to S information generated as inspection result data E1 in the medical document inspection process according to an embodiment of the present disclosure. In this example, examples of outputs for the inputs shown in FIGS. 7A, 7B, and 7C are shown. The output text P11 includes text in a predetermined format (as an example, the JSON format).

[0100] <Regarding the above two input examples> Input examples related to EP information shown in FIGS. 5A, 5B, and 5C and input examples related to S information shown in FIGS. 7A, 7B, and 7C will be described.

[0101] In the input example for EP information, the parts "using the content of the conversation between the pharmacist and the patient regarding medication guidance as input information" and "this is an EP of the medication history" in "The following is an EP of a SOAP-formatted medication history created using the content of the conversation between the pharmacist and the patient regarding medication guidance as input information," and the part "this EP of the SOAP-formatted medication history" in "Can you extract any content from this SOAP-formatted medication history EP that is not based on the input information, according to the following "criteria for determining whether it is based on input information" and "rules for output"?", and "#Criteria for determining whether it is based on input information" The phrase "EP information is a section where pharmacists describe the content of the guidance they provide to patients during medication counseling. Therefore, even if there is no direct content in the input information that matches the EP information, if there is information that allows for inference, it may be judged as content based on the input information." is the part that dynamically changes the criteria for "based on input information" according to each item in the SOAP format, and also according to the completeness of the input information.

[0102] In the input example for S information, the parts "The following is S in SOAP format of a patient medication history created using only the pharmacist's utterances from the conversation between the pharmacist and the patient regarding medication guidance" and "S in the patient medication history," and the part "S in this SOAP format of a patient medication history" from "Can you extract any content from this SOAP format of a patient medication history that is not based on the input information, according to the following "criteria for determining whether it is based on the input information" and "rules for output"?", and the part "S in this SOAP format of a patient medication history," and "#Criteria for determining whether it is based on the input information" S information is a section where the patient's subjective symptoms and complaints are recorded, without including the pharmacist's subjective thoughts or statements. However, since this input information only contains the pharmacist's utterances and is highly likely to not adequately include the patient's statements, there is a possibility that no direct content matching the S information exists in the input information. Therefore, even if no direct content matching the S information exists in the input information, if there is information that allows the patient's statements to be inferred from the pharmacist's utterances, it may be judged as content based on the input information. However, since this input information consists only of the pharmacist's utterances and is highly likely to not adequately include the patient's statements, there is a possibility that no direct content matching the S information exists in the input information. Therefore, even if no direct content matching the S information exists in the input information, if there is information that allows the patient's statements to be inferred from the pharmacist's utterances, it may be judged as content based on the input information. This section concerns how the criteria for "based on input information" are dynamically changed according to each item in the SOAP format, and how the criteria for "based on input information" are also dynamically changed according to the completeness of the input information.

[0103] As can be seen by comparing the two examples above (an example concerning EP information and an example concerning S information), the content of the inspection instructions and the "based on input information" criteria are dynamically changed depending on which SOAP format item the inspection is for and the completeness of the input information.

[0104] In this embodiment, the process of inputting inspection request data D1 into the generation AI and having the generation AI generate inspection result data E1 (for convenience of explanation, this is also called the inspection result data generation process) is shown as generating the inspection result data E1 all at once. However, as another example, the inspection result data E1 may be generated in multiple stages, such as two stages. For example, the accuracy of the output text may be higher if the text is divided into two or more parts and input sequentially into the generation AI rather than inputting a long text all at once.

[0105] For example, in a two-stage process, the first response data may be obtained from the generating AI server 5 by outputting the first examination request data D1A to the generating AI server 5 in the first step, and then the second response data may be obtained from the generating AI server 5 by outputting the second examination request data D1B to the generating AI server 5 in response to that response data in the second step. Furthermore, the process of generating test result data may be divided into three or more stages. Whether the inspection result data generation process is performed in a single step or in multiple stages may be appropriately selected depending on the required accuracy of the inspection result data E1 and the processing time.

[0106] In this embodiment, an example was described in which the generated result data C1 and the test result data E1 are converted into result data F1 by the medical document presentation support device 3 and presented to the user terminal device 2. As another example, the generated result data C1 and the test result data E1 may be transmitted from the medical document presentation support device 3 to the user terminal device 2, and the generated result data C1 and the test result data E1 may be converted into result data F1 by the user terminal device 2.

[0107] Furthermore, in this embodiment, an example of a case in which the medical document presentation support device 3 generates the examination request data D1 was described. As another example, the inspection request data D1 may be generated based on the actions of user U1. For example, user U1 may operate the operation unit 22 of the user terminal device 2 to specify the information to be included in the inspection request data D1.

[0108] Furthermore, in this embodiment, an example was described in which the test result data E1 generated by the generation AI server 5 is output to the medical document presentation support device 3. As another example, the generating AI server 5 may output the generated test result data E1 directly to the user terminal device 2 without going through the medical document presentation support device 3.

[0109] [Variations in system configuration] In this embodiment, an example has been described in which the user terminal device 2 and the medical document presentation support device 3 are provided as separate devices in the medical document presentation support system 1, but the invention is not limited to this example. As another example, the user terminal device 2 and the medical document presentation support device 3 may be an integrated device. In other words, instead of having the user terminal device 2, the functions that the user terminal device 2 has may be provided in the medical document presentation support device 3.

[0110] Figure 9 shows an example of the functional configuration of a medical document presentation support system 1a according to a modified embodiment of the present disclosure. In this modified example, the medical document presentation support system in which the functional unit of the user terminal device 2 is provided in the medical document presentation support device 3 is referred to as the medical document presentation support system 1a. The medical document presentation support system 1a includes a medical document presentation support device 3a.

[0111] The medical document presentation support device 3a is, for example, a computer. The medical document presentation support device 3a comprises a display unit and an operation unit. The display unit comprises, for example, a liquid crystal display and displays various screens. The operation unit comprises, for example, a mouse, keyboard, or touch panel and detects operations of user U1.

[0112] The medical document presentation support device 3a comprises a presentation request unit 201, a presentation unit 202, a presentation request reception unit 30, a generation request data acquisition unit 31, a generation result data output request unit 32, a generation result data acquisition unit 33, an examination perspective determination unit 34, an examination request data acquisition unit 35, an examination result data output request unit 36, an examination result data acquisition unit 37, a result data output unit 38, an examination presence / absence determination unit 51, and a standard provision information management unit 52.

[0113] The functions of the presentation request unit 201 and the presentation unit 202 are the same as those of the presentation request unit 201 and presentation unit 202 provided in the user terminal device 2 shown in Figure 1. In the example in Figure 9, the same reference numerals are used for functional parts that are the same as those in Figure 1.

[0114] [Regarding the above embodiments] As described above, in the medical document presentation support system 1 according to this embodiment, the medical document presentation support device 3 can present the results of an inspection regarding the accuracy of medical document data generated from the original data, thereby supporting the user (for example, a pharmacist) in presenting the medical document data. The user can then understand the accuracy of the medical document data. Thus, in this embodiment, for example, when medical document data is presented, the results of the examination of the medical document data, taking into account the accuracy of the medical document data, can be presented, thereby supporting user U1 (e.g., a pharmacist).

[0115] In this embodiment, for example, by utilizing information explicitly entered by user U1, the accuracy of the inspection can be improved by dynamically controlling and inputting inspection instructions, which include criterion information regarding whether or not they are based on the input information, into the generating AI.

[0116] In this embodiment, for example, by configuring the medical document presentation support device 3 to process the generated medical document data and the test results generated by the AI ​​asynchronously, the latency of each response can be improved. In this embodiment, for example, in the medical document presentation support device 3, by deliberately providing the generating AI with only somewhat abstracted examination instructions and entrusting the examination accuracy to the generating AI, it is possible to construct a novel examination mechanism of the AI ​​model that has high generalization performance and can improve its accuracy depending on the tuning of the examination instructions.

[0117] In this embodiment, for example, by having user U1 explicitly specify the level of completeness of the input information, the accuracy of dynamic control on the medical document presentation support device 3 and the accuracy of inspection by the generating AI can be improved.

[0118] In this embodiment, for example, the generated medical document data and the test results generated by the AI ​​can be displayed on the UI in an optimal format, thereby improving visibility and readability for user U1 on the UI. In this embodiment, for example, in the user terminal device 2, instead of processing all information such as medical document data and test results generated by the AI ​​synchronously, appropriate processing (asynchronous processing) is performed for each piece of information as it becomes available to be displayed on the UI. This makes it possible to improve the reliability of medical care utilizing the AI ​​without compromising the pharmacist's application experience.

[0119] As an example configuration, the medical document presentation support device 3 includes an examination request data acquisition unit 35 and an examination result data output request unit 36. The inspection request data acquisition unit 35 acquires inspection request data D1 which includes original data containing input information, SOAP-formatted medical document data generated based on the original data, and criterion information regarding whether or not it is based on input information. The test result data output request unit 36 ​​requests the test result data output device (in this embodiment, the generating AI server 5), which outputs test result data E1 related to medical document data based on a trained model (in this embodiment, the trained model of the generating AI server 5) that has been trained to output output data corresponding to the input data, to output test result data E1 from the test request data D1. Therefore, the medical document presentation support device 3 can provide support for the presentation of medical document data by generating test results regarding the accuracy of the medical document data.

[0120] In this embodiment, we have described an example where the trained model of the generation AI server 5 is a generation AI, but the embodiment is not limited to this. The trained model may be a trained model based on a machine learning model other than a generative AI, as long as it is trained to output inspection result data E1 when inspection request data D1 is input.

[0121] Furthermore, while this embodiment describes an example where the trained model of the generation AI server 4 is a generation AI, it is not limited to this example. The trained model may be a trained model based on a machine learning model other than a generative AI, as long as it is trained to output the generated result data C1 when the generation request data B1 is input.

[0122] As an example configuration, the medical document presentation support device 3 further includes a standard information management unit 52 and an examination viewpoint determination unit 34. The standard information management unit 52 manages standard information that defines standard information (in this embodiment, the information shown in Figures 3A and 3B). The inspection perspective determination unit 34 determines the standard information based on the standard provision information. Therefore, the medical document presentation support device 3 can switch the standard information to be used, for example, for each item in the SOAP format and according to the completeness of the input information.

[0123] As an example configuration, the medical document presentation support device 3 has the following configuration. The test result data E1 includes the test results for each item in SOAP format, as well as the accuracy of the test results. Therefore, the medical document presentation support device 3 can display the accuracy of the information contained in the medical document data for each item in SOAP format.

[0124] As one example configuration, the medical document presentation support device 3 further includes a generated result data output request unit 32. The generated result data output request unit 32 requests the generation of medical document data. Therefore, when generating medical document data, the medical document presentation support device 3 can generate test results for the generated medical document data. In other words, the medical document presentation support device 3 can realize a series of operations, including the generation of medical document data and the testing of the generated medical document data.

[0125] As one example configuration, the medical document presentation support device 3 further includes an examination determination unit 51. The inspection presence / absence determination unit 51 determines whether or not an inspection is performed. The inspection result data output request unit 36 ​​requests the output of inspection result data E1 if the inspection presence / absence determination unit 51 determines that an inspection is required, and does not request the output of inspection result data E1 if the inspection presence / absence determination unit 51 determines that no inspection is required. Therefore, the medical document presentation support device 3 can switch whether or not to generate test results in response to instructions from user U1 (for example, a pharmacist).

[0126] As an example configuration, the medical document presentation support device 3 further includes an examination result data acquisition unit 37 and a result data output unit 38. The inspection result data acquisition unit 37 acquires the inspection result data E1 output by the inspection result data output device. The results data output unit 38 outputs results data including the test result data E1 and medical document data. Therefore, the medical document presentation support device 3 can, for example, present medical document data and test results to user U1 (e.g., a pharmacist). In the result data F1, for example, the test result data E1 and the medical document data may be treated as a single data set, or they may be treated as separate data sets.

[0127] As an example configuration, the medical document presentation support device 3 has the following configuration. The process of displaying medical document data and the process of displaying test result data E1 are performed asynchronously. Therefore, the medical document presentation support device 3 can present these pieces of information appropriately by, for example, performing a process to present the medical document data when it is acquired, and then performing a process to present the test result data E1 when it is acquired.

[0128] As an example configuration, the medical document presentation support system 1 comprises a medical document presentation support device 3 and a terminal device (in this embodiment, a user terminal device 2). The terminal device comprises a presentation request unit 201 that receives a presentation request A1 from user U1, and a presentation unit 202 that presents data corresponding to presentation request A1. Therefore, the medical document presentation support system 1 can provide support for the presentation of medical document data by generating test results regarding the accuracy of the medical document data.

[0129] In this embodiment, the medical document presentation support device 3 is shown to have the function of an AI generation agent that generates medical document data using a generation AI, and the function of an AI inspection agent that performs inspections on the medical document data using the generation AI, but it is not limited to this.

[0130] As another example, a medical document presentation device may be configured that has the functionality of an AI examination agent but does not have the functionality of an AI generation agent. In this case, the medical document presentation device does not include, for example, the functional units related to the process of causing the generation AI server 4 to generate medical document data (generation request data acquisition unit 31, generation result data output request unit 32, generation result data acquisition unit 33) among the functional units shown in Figure 2. Also, in this case, the generation AI server 4 is not used. Furthermore, for example, if presentation request A1 always includes an instruction for an examination, the medical document presentation device does not need to be equipped with an examination presence / absence determination unit 51. In this configuration, for example, the presentation request A1 output from the presentation request unit 201 includes the original data, the medical document data generated from the original data, and an instruction to present the examination results of the medical document data. Here, the medical document data corresponding to the original data may be generated by any method. In the medical document presentation device, when such a presentation request A1 is received, the examination perspective determination unit 34 determines the examination perspective based on the information contained in the presentation request A1, the examination request data acquisition unit 35 acquires the examination request data D1, the examination result data output request unit 36 ​​requests the generation AI server 5 to output the examination result data E1, the examination result data acquisition unit 37 acquires the examination result data E1 from the generation AI server 5, and the result data output unit 38 outputs the result data F1 based on the examination result data E1 to the presentation unit 202.

[0131] For example, a computer program to implement the functions of each of the above-mentioned devices may be recorded on a computer-readable recording medium, and the program recorded on this recording medium may be loaded into a computer system and executed. The term "computer system" here may include hardware such as an operating system and peripheral devices. Furthermore, "computer-readable recording media" refers to writable non-volatile memory such as flexible disks, magneto-optical disks, ROMs, and flash memory, portable media such as DVDs (Digital Versatile Discs), and storage devices such as hard disks built into computer systems. Furthermore, "computer-readable recording media" also includes volatile memory (such as DRAM (Dynamic Random Access Memory)) within a computer system that acts as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line, which retains the program for a certain period of time.

[0132] Furthermore, the above program may be transmitted from a computer system that stores the program in a memory device or the like to another computer system via a transmission medium or by transmission waves within the transmission medium. Here, the "transmission medium" for transmitting the program refers to a medium that has the function of transmitting information, such as a network (communication network) like the Internet or a communication line (communication line) like a telephone line. Furthermore, the above program may be intended to implement some of the functions described above. It may also be a so-called differential file (differential program) that can implement the aforementioned functions in combination with programs already recorded in the computer system.

[0133] While embodiments of this disclosure have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and may include design changes and the like that do not depart from the gist of this disclosure.

[0134] <Note> [Configuration Example 1] An inspection request data acquisition unit acquires inspection request data which includes raw data containing input information, SOAP-format medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information. An inspection result data output device that outputs inspection result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, includes an inspection result data output request unit that requests the output of the inspection result data from the inspection request data, A medical document presentation support device equipped with the following features.

[0135] [Configuration Example 2] moreover, A standard information management unit that manages standard information that defines the aforementioned standard information, An inspection perspective determination unit that determines the standard information based on the standard information, Equipped with, A medical document presentation support device as described in [Configuration Example 1].

[0136] [Configuration Example 3] The aforementioned test result data includes the test results for each item in SOAP format and the accuracy of the information of the test results. A medical document presentation support device as described in [Configuration Example 1] or [Configuration Example 2].

[0137] [Configuration Example 4] moreover, The system includes a generation result data output request unit that requests the generation of the aforementioned medical document data. A medical document presentation support device as described in any one of the following items: [Configuration Example 1] to [Configuration Example 3].

[0138] [Configuration Example 5] moreover, It is equipped with an inspection presence / absence determination unit that determines whether or not an inspection is performed, The inspection result data output request unit requests the output of the inspection result data when the inspection presence / absence determination unit determines that the inspection has taken place, and does not request the output of the inspection result data when the inspection presence / absence determination unit determines that the inspection has not taken place. A medical document presentation support device as described in any one of the following items: [Configuration Example 1] to [Configuration Example 4].

[0139] [Configuration Example 6] moreover, An inspection result data acquisition unit that acquires the inspection result data output by the inspection result data output device, A result data output unit that outputs result data including the aforementioned test result data and the aforementioned medical document data, A medical document presentation support device according to any one of [Configuration Example 1] to [Configuration Example 5], comprising the above.

[0140] [Configuration Example 7] The process of presenting the aforementioned medical document data and the process of presenting the aforementioned test result data are performed asynchronously. A medical document presentation support device as described in any one of the items from [Configuration Example 1] to [Configuration Example 6].

[0141] Here, we can provide a medical document presentation support system equipped with a medical document presentation support device. [Configuration Example 8] A medical document presentation support device described in any one of the [Configuration Example 1] to [Configuration Example 7], A terminal device comprising: a presentation request unit that receives presentation requests from users; and a presentation unit that presents data in response to the presentation requests. A medical document presentation support system equipped with the following features.

[0142] Furthermore, we can provide a method for processing performed in a medical document presentation support device. [Configuration Example 9] A test request data acquisition step that acquires test request data including source data containing input information, SOAP-formatted medical document data generated based on the source data, and criterion information regarding whether or not it is based on the input information, A test result data output request step requests a test result data output device, which outputs test result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, to output the test result data from the test request data. A method for supporting the presentation of medical documents that has the following characteristics.

[0143] Furthermore, we can provide a program (computer program) for realizing the processing performed in the medical document presentation support device. [Configuration Example 10] On the computer, A test request data acquisition step that acquires test request data including source data containing input information, SOAP-formatted medical document data generated based on the source data, and criterion information regarding whether or not it is based on the input information, A test result data output request step requests a test result data output device, which outputs test result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, to output the test result data from the test request data. A program to execute. [Explanation of Symbols]

[0144] 1, 1a...Medical document presentation support system, 2...User terminal device, 3, 3a...Medical document presentation support device, 4...Generating AI server (medical document generation), 5...Generating AI server (examination), 21...Display unit, 22...Operation unit, 30...Presentation request reception unit, 31...Generating request data acquisition unit, 32...Generating result data output request unit, 33...Generating result data acquisition unit, 34...Examination perspective determination unit, 35...Examination request data acquisition unit, 36...Examination result data output request unit, 37...Examination result data Acquisition unit, 38... Result data output unit, 51... Inspection presence / absence determination unit, 52... Standard specification information management unit, 200, 300... Calculation unit, 201... Presentation request unit, 202... Presentation unit, 250, 350... Storage unit, A1... Presentation request, B1... Generation request data, C1... Generation result data, D1... Inspection request data, E1... Inspection result data, F1... Result data, P1~P3, P21~P23... Input text, P11, P31... Output text, T1, T2... Table, U1... User

Claims

1. An inspection request data acquisition unit acquires inspection request data which includes raw data containing input information, SOAP-format medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information. An inspection result data output device that outputs inspection result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, includes an inspection result data output request unit that requests the output of the inspection result data from the inspection request data, A medical document presentation support device equipped with the following features.

2. moreover, A standard information management unit that manages standard information that defines the aforementioned standard information, An inspection perspective determination unit that determines the standard information based on the standard information, Equipped with, A medical document presentation support device according to claim 1.

3. The aforementioned test result data includes the test results for each item in SOAP format and the accuracy of the information of the test results. A medical document presentation support device according to claim 1.

4. moreover, The system includes a generation result data output request unit that requests the generation of the aforementioned medical document data. A medical document presentation support device according to claim 1.

5. moreover, It is equipped with an inspection presence / absence determination unit that determines whether or not an inspection is performed, The inspection result data output request unit requests the output of the inspection result data when the inspection presence / absence determination unit determines that the inspection has taken place, and does not request the output of the inspection result data when the inspection presence / absence determination unit determines that the inspection has not taken place. A medical document presentation support device according to claim 1.

6. moreover, An inspection result data acquisition unit that acquires the inspection result data output by the inspection result data output device, A result data output unit that outputs result data including the aforementioned test result data and the aforementioned medical document data, A medical document presentation support device according to claim 1, comprising:

7. The process of presenting the aforementioned medical document data and the process of presenting the aforementioned test result data are performed asynchronously. A medical document presentation support device according to claim 1.

8. A medical document presentation support device according to any one of claims 1 to 7, A terminal device comprising: a presentation request unit that receives presentation requests from users; and a presentation unit that presents data in response to the presentation requests. A medical document presentation support system equipped with the following features.

9. A step to acquire test request data, which includes raw data containing input information, SOAP-formatted medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information, A test result data output request step requests a test result data output device, which outputs test result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, to output the test result data from the test request data. A method for supporting the presentation of medical documents that has the following characteristics.

10. On the computer, A step to acquire test request data, which includes raw data containing input information, SOAP-formatted medical document data generated based on the raw data, and criterion information regarding whether or not it is based on the input information, A test result data output request step requests a test result data output device, which outputs test result data relating to medical document data based on a trained model that has been trained to output output data corresponding to input data, to output the test result data from the test request data. A program to execute.

Images