PROGRAM, INFORMATION PROCESSING APPARATUS, METHOD AND SYSTEM

Abstract

Provided is a technology that enables easy construction of a system for obtaining information for diagnosing patients. [Solution] The program causes a processor 29 to output disease candidates, etc. based on case information. Memories 25, 26 store a second program that receives as input case information and a text or vector expression that is a prerequisite for outputting an evaluation of a disease name, etc., and outputs disease candidates, etc. based on the case information and an evaluation of the disease candidates, etc. When the program accepts input of case information from a user, it inputs the case information, etc. into the second program, and causes the second program to output disease candidates, etc. based on the case information and an evaluation of the disease candidates, etc.

Description

[Technical field]

[0001] The present disclosure relates to a program, an information processing device, a method, and a system. [Background technology]

[0002] Patent Document 1 discloses a technique for assisting doctors in diagnosing patients with medical conditions.

[0003] Patent Document 1 discloses a technique relating to a diagnostic support method, which includes the steps of determining a number of factors having values ​​that can be used to classify the patient and a number of abstraction levels between the factor values ​​and raw data, generating a neural network according to the determined number of levels, acquiring raw data related to each of a number of factors from a number of sources, processing the raw data of the factors using the neural network to obtain values ​​of the factors, classifying the patient using the factor values, comparing classification information of the patient with equivalent information of other subjects, and outputting information for use in diagnosing the patient based on the comparison. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2018-067303 A Summary of the Invention [Problem to be solved by the invention]

[0005] The technology described in Patent Document 1 obtains information to be used in diagnosing patients by identifying and inputting various factors related to the patient, constructing a so-called neural network based on these factors, and classifying the patient. Therefore, it takes a lot of time and effort to construct such a diagnostic support method.

[0006] Therefore, the present disclosure has been made to solve the above-mentioned problem of high effort, and its purpose is to provide a technology for a system that obtains information for diagnosing patients, in which input is information from medical records and prompts are used to control a machine learning model in natural language or vector expressions, thereby maintaining sufficient accuracy.

[0007] In one example, the machine learning model receives the contents of a patient's medical record along with a prompt, and outputs a graded representation of the certainty of the disease and treatment recommendations along with their explanation. In this case, the accuracy of the output of the machine learning model is maintained by controlling the output of the machine learning model by specifying the list of candidates to be output in the prompt and the output form of the certainty / recommendation for each candidate. This makes it possible to automatically improve the accuracy of the prompts to be input to the machine learning model through trial and error based on the large amount of electronic medical record information currently available and the graded representation of the certainty of the doctor's diagnosis or the graded representation of the treatment recommendation for that information, without the need to proactively collect new patient data.

[0008] As a further embodiment of the present invention, based on electronic medical record information, a machine learning model (large-scale language model) is used to create a sentence that specifies the date of outcome occurrence such as positive findings, negative findings, test date, treatment method, treatment start date, definitive diagnosis date, cure, death, hospitalization, etc., the date of outcome occurrence, the period taken until the outcome occurrence, and the index date, and replaces the sentence with the relative period taken from that date until the occurrence, and further creates pseudo electronic medical record information from the converted sentence that matches positive findings, negative findings, etc., test findings, outcomes, diagnosis names, diagnosis dates, etc., thereby performing data augmentation, and while avoiding the problems of the Personal Information Protection Act, it is possible to obtain a large amount of text and correct answer data and train a diagnostic support machine learning model. This also makes it possible to support the diagnosis of rare diseases.

[0009] For example, the normal value for a test such as TSH varies from facility to facility, ranging from 0.5 to 5. In order to eliminate such differences between facilities, a value such as TSH 10 can be converted to TSH elevated, and a value of TSH 100 or higher can be converted to TSH markedly elevated, etc. This type of conversion can absorb the differences between facilities. [Means for solving the problem]

[0010] A program for operating a computer having a processor and a memory. The program causes the processor to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information. The memory stores a second program that receives as input the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of the disease name, the treatment name, and the test name, and outputs at least one of the disease candidate, the treatment candidate, and the test candidate based on the case information, and an evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate. The program includes a first step of receiving input of case information from a user, a seventh step of inputting the case information received in the first step and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector representations, or at least one of a treatment name related to the case information, a strong applicable condition, a weak applicable condition, and a contraindication condition related to the treatment name, or at least one of a test name related to the case information, a strong applicable condition, a weak applicable condition, and a contraindication condition for the test related to the test name, which are text or vector representations, into a second program, and a second step of causing the second program to output at least one of the disease candidate, treatment candidate, and test candidate corresponding to the information and the text or vector representation, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate that is the output of the second step, and the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate. In the second step, the at least one of the disease candidate, treatment candidate, and test candidate, and the likelihood or recommendation level of the at least one of the disease candidate, treatment candidate, or test candidate are output as the evaluation. Effect of the Invention

[0011] According to the present disclosure, it is possible to provide a technology that maintains sufficient accuracy as a system for obtaining information for diagnosing patients by using medical record information as input and controlling a machine learning model in natural language or vector expressions using prompts. [Brief description of the drawings]

[0012] [Figure 1] 1 is a diagram showing an overall configuration of a system according to an embodiment; [Diagram 2] FIG. 2 is a diagram illustrating a functional configuration of a terminal device according to an embodiment. [Diagram 3] FIG. 2 is a diagram illustrating a functional configuration of a server according to an embodiment. [Figure 4] 11 is a flowchart illustrating an example of a processing flow in a system according to an embodiment. [Diagram 5] 13 is a flowchart showing another example of the processing flow in the system according to an embodiment. [Figure 6] 13 is a flowchart showing yet another example of the processing flow in the system according to an embodiment. [Figure 7] FIG. 11 is a schematic diagram illustrating an example of a screen displayed on a terminal device according to an embodiment. [Figure 8] FIG. 11 is a schematic diagram illustrating another example of a screen displayed on a terminal device according to an embodiment. [Figure 9] FIG. 11 is a schematic diagram illustrating yet another example of a screen displayed on the terminal device according to an embodiment. [Figure 10] FIG. 11 is a schematic diagram illustrating yet another example of a screen displayed on the terminal device according to an embodiment. [Figure 11] FIG. 11 is a schematic diagram illustrating yet another example of a screen displayed on the terminal device according to an embodiment. [Figure 12] FIG. 11 is a diagram illustrating an example of a method for generating pseudo case information in a system according to an embodiment. [Figure 13] FIG. 13 is a diagram showing an example of a method for generating pseudo case information in a system according to an embodiment, and is a diagram showing a part of FIG. 12 in detail. [Figure 14]FIG. 2 is a diagram illustrating an example of a program installed in a system according to an embodiment. [Figure 15] 14 is a diagram showing an example of a program installed in the system according to an embodiment. FIG. [Figure 16] 15 is a diagram showing an example of a program installed in the system according to an embodiment. FIG. [Figure 17] 16 is a diagram showing an example of a program installed in the system according to an embodiment. FIG. [Figure 18] 17 is a diagram showing an example of a program installed in the system according to an embodiment. FIG. [Figure 19] 18 is a diagram showing an example of a program installed in the system according to an embodiment. FIG. [Figure 20] 19 is a diagram showing an example of a program installed in the system according to an embodiment, and FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of the present disclosure will be described with reference to the drawings. In all the drawings explaining the embodiment, the same reference numerals are given to common components, and repeated explanations are omitted. Note that the following embodiment does not unduly limit the contents of the present disclosure described in the claims. In addition, not all of the components shown in the embodiment are essential components of the present disclosure. In addition, each figure is a schematic diagram and is not necessarily illustrated strictly.

[0014] In the following description, a "processor" refers to one or more processors. The at least one processor is typically a microprocessor such as a CPU (Central Processing Unit), but may be another type of processor such as a GPU (Graphics Processing Unit). The at least one processor may be a single-core or multi-core.

[0015] Furthermore, the at least one processor may be a processor in the broad sense, such as a hardware circuit (for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC)) that performs part or all of the processing.

[0016] In the following explanation, information that gives an output for an input may be described using an expression such as "xxx table", but this information may be data of any structure or a learning model such as a neural network that generates an output for an input. Therefore, the "xxx table" may be called "xxx information".

[0017] Furthermore, in the following description, the configuration of each table is an example, and one table may be divided into two or more tables, or all or part of two or more tables may be one table.

[0018] In addition, in the following explanation, the processing may be described with the "program" as the subject, but since the program is executed by a processor to perform a specified processing step by appropriately using a memory unit and / or an interface unit, etc., the subject of the processing may be the processor (or a device such as a controller having the processor).

[0019] The program may be installed in a device such as a computer, or may be, for example, in a program distribution server or a computer-readable (e.g., non-transitory) recording medium. In the following description, two or more programs may be realized as one program, or one program may be realized as two or more programs.

[0020] The functions performed by the components described herein may be implemented in circuitry or processing circuitry, including general purpose processors, application specific processors, integrated circuits, ASICs, CPUs, conventional circuits, and / or combinations thereof, programmed to perform the functions described. Processors include transistors and other circuits and are considered to be circuitry or processing circuitry. A processor may be a programmed processor that executes a program stored in a memory.

[0021] In this specification, a circuitry, unit, or means is hardware that is programmed to realize or performs the described functions, which may be any hardware disclosed in this specification or any hardware known to be programmed to realize or perform the described functions.

[0022] If the hardware is a processor considered to be a type of circuitry, the circuitry, means, or unit is a combination of the hardware and software used to configure the hardware and / or processor.

[0023] In the following description, identification numbers are used as identification information for various objects, but other types of identification information (for example, identifiers including alphabetic characters or symbols) may be used instead.

[0024] In addition, in the following description, when describing elements of the same type without distinguishing between them, reference signs (or common signs among the reference signs) may be used, and when describing elements of the same type with distinction between them, the identification numbers (or reference signs) of the elements may be used.

[0025] In the following description, the control lines and information lines are those that are considered necessary for the description, and not all control lines and information lines in the product are necessarily shown. All components may be connected to each other.

[0026] <0 System Overview> The system according to the present disclosure is a system that, when a medical professional such as a doctor inputs medical information such as case information and medical notes, outputs disease candidates, treatment candidates, or test candidates based on this medical information.

[0027] Here, case information refers to information including at least the current symptoms of a specific patient (not limited to a single person), and preferably includes the patient's age, sex, chief complaint heard by the medical professional, current medical history (history of present illness) and past medical history / family history, findings at first visit and test values ​​for the patient, patient outcomes, findings from imaging tests, and audio information. Case information includes information interviewed by a medical professional in person, information entered by the patient on a medical questionnaire, and audio recordings.

[0028] In order to enter case information into the system, it needs to be converted into electronic data (e.g., text data). For example, when a medical professional interviews a patient face-to-face, the details of the interview that the medical professional manually enters into a free-entry field in the electronic medical record are included in the digitized case information. In addition, the details entered by the patient into the medical questionnaire prior to the medical professional's interview also need to be digitized. In recent years, systems have become available that use tablet devices and other devices to electronically input answers to medical questions corresponding to medical questions, known as electronic medical questionnaires, and the answers to the questions on electronic medical questionnaires have already been digitized.

[0029] The system according to the present disclosure uses a machine learning model that receives medical information, typically case information, as input and outputs disease candidates, treatment candidates, or test candidates based on this medical information, and an evaluation of the disease candidates, etc. This machine learning model may be a general-purpose large-scale language model, typically ChatGPT, or may be a machine learning model tuned for the system according to the present disclosure.

[0030] In order to tune a machine learning model for the system according to the present disclosure, it is preferable to use data such as a list of diseases, findings regarding the diagnosis, exclusion, or evaluation value of each disease (including case information, interview information, and some processed texts in which the interpretation is changed for positive / negative tests), textbook information on cases, and case information as pre-learning data or teacher data for fine tuning and instruction tuning, and perform machine learning using this data. Then, case information, etc. is input into the trained machine learning model, and the evaluation of disease candidates, treatment candidates, or test candidates and these disease candidates, etc. as the output result, is compared with the evaluation of disease candidates, treatment candidates, or test candidates and these disease candidates, etc., which is the doctor's answer when the same input is given, and the doctor's answer is given to the machine learning model as the correct answer data, thereby tuning the machine learning model.

[0031] When inputting medical information into a machine learning model, it is preferable to add some text (called a prompt) to the case information and input the case information + prompt to the machine learning model. Details of the prompt will be described later, but basically it includes the specification of the input contents for the machine learning model, output examples, and textbook information for each case.

[0032] Therefore, the system disclosed herein is a system that generates a machine learning model that creates an individualized doctor's manual for each situation corresponding to a case, and outputs information from this machine learning model to assist medical professionals in making diagnoses.

[0033] <One embodiment> <1 Overall system configuration> FIG. 1 is a diagram showing the overall configuration of a medical support system (hereinafter simply referred to as the "system") 1 of this embodiment. As shown in FIG. 1, the system 1 includes a plurality of terminal devices (terminal devices 10A and 10B are shown in FIG. 1. Hereinafter, they may be collectively referred to as "terminal devices 10") and a server 20. The terminal devices 10 and the server 20 are connected to each other so as to be able to communicate with each other via a network 80. The network 80 is configured as a wired or wireless network. In this embodiment, the server 20 is a server having a function as a Web server (including a cloud server), and exchanges information with the terminal device 10 through Web pages. In addition, a Web page browser for viewing Web pages is installed in the terminal device 10, but a dedicated application for providing the services of the server 20 may be installed and configured to be viewable through the dedicated application.

[0034] The terminal device 10 is a device operated by a medical professional or an administrator of the system 1. Here, the concept of a medical professional includes doctors, nurses, medical technicians, etc. In the following description, the medical professional includes the administrator of the system 1, except when a distinction is made between the medical professional and the administrator of the system 1.

[0035] A medical professional inputs medical information including case information using the terminal device 10, receives information on disease candidates, treatment candidates, or test candidates based on the input medical information from the server 20, and views the information on the terminal device 10. In addition, a medical professional inputs various information using the terminal device 10, and the input information is processed in the server 20 described below or in an external server 40 outside the system 1, and the processing results are displayed on the terminal device 10.

[0036] The server 20 is managed by an administrator of the system 1 of this embodiment, and the stored contents are appropriately modified / added / deleted by medical staff who are users of the terminal devices 10.

[0037] The server 20 is a computer connected to a network 80. The server 20 includes a communication IF 22, an input / output IF 23, a memory 25, a storage 26, and a processor 29.

[0038] The communication IF 22 is an interface for inputting and outputting signals so that the server 20 can communicate with an external device. The input / output IF 23 functions as an interface with an input device for receiving input operations from a user and an output device for presenting information to a user. The memory 25 is for temporarily storing programs and data processed by the programs, etc., and is a volatile memory such as a DRAM (Dynamic Random Access Memory). The storage 26 is a storage device for saving data, such as a flash memory or a HDD (Hard Disc Drive). The processor 29 is hardware for executing an instruction set described in a program, and is composed of an arithmetic unit, a register, a peripheral circuit, etc.

[0039] <1.1 Functional configuration of the terminal device 10> FIG. 2 is a block diagram showing a functional configuration of the terminal device 10 constituting the system 1 of the present embodiment. As shown in FIG. 2, the terminal device 10 includes a plurality of antennas (antenna 111, antenna 112), wireless communication units (first wireless communication unit 121, second wireless communication unit 122) corresponding to the respective antennas, an operation reception unit 130 (including a keyboard 131 and a mouse 132), a voice processing unit 140, a microphone 141, a speaker 142, a display 150, a storage unit 170, and a control unit 180. The terminal device 10 also has functions and configurations (for example, a battery for storing power, a power supply circuit for controlling the supply of power from the battery to each circuit, etc.) that are not particularly shown in FIG. 3. As shown in FIG. 2, each block included in the terminal device 10 is electrically connected by a bus or the like.

[0040] The antenna 111 emits a signal generated by the terminal device 10 as a radio wave. The antenna 111 also receives a radio wave from space and provides the received signal to the first wireless communication unit 121.

[0041] The antenna 112 radiates a signal generated by the terminal device 10 as a radio wave. The antenna 112 also receives a radio wave from space and provides the received signal to the second radio communication unit 122.

[0042] The first wireless communication unit 121 performs modulation / demodulation processing and the like for transmitting and receiving signals via the antenna 111 so that the terminal device 10 can communicate with other wireless devices. The second wireless communication unit 122 performs modulation / demodulation processing and the like for transmitting and receiving signals via the antenna 112 so that the terminal device 10 can communicate with other wireless devices. The first wireless communication unit 121 and the second wireless communication unit 122 are communication modules including a tuner, a Received Signal Strength Indicator (RSSI) calculation circuit, a Cyclic Redundancy Check (CRC) calculation circuit, a high-frequency circuit, and the like. The first wireless communication unit 121 and the second wireless communication unit 122 perform modulation / demodulation and frequency conversion of wireless signals transmitted and received by the terminal device 10, and provide the received signals to the control unit 180.

[0043] The operation reception unit 130 has a mechanism for receiving input operations from a user. Specifically, the operation reception unit 130 includes a keyboard 131 and a mouse 132. Note that the operation reception unit 130 may be configured as a touch screen that detects the user's contact position on the touch panel, for example, by using a capacitive touch panel.

[0044] The keyboard 131 accepts input operations by the user of the terminal device 10. The keyboard 131 is a device for inputting characters, and outputs input character information to the control unit 180 as an input signal.

[0045] The mouse 132 accepts input operations by the user of the terminal device 10. The mouse 132 is a pointing device for selecting an object displayed on the display 150, and outputs position information of an object selected on the screen and information indicating that a button has been pressed to the control unit 180 as input signals.

[0046] The audio processing unit 140 modulates and demodulates an audio signal. The audio processing unit 140 modulates a signal provided from the microphone 141 and provides the modulated signal to the control unit 180. The audio processing unit 140 also provides the audio signal to the speaker 142. The audio processing unit 140 is realized by, for example, a processor for audio processing. The microphone 141 accepts audio input and provides an audio signal corresponding to the audio input to the audio processing unit 140. The speaker 142 converts the audio signal provided from the audio processing unit 140 into audio and outputs the audio to the outside of the terminal device 10.

[0047] Display 150 displays data such as images, videos, and text under the control of control unit 180. Display 150 is realized by, for example, an LCD (Liquid Crystal Display) or an organic EL (Electro-Luminescence) display.

[0048] Storage unit 170 is configured with, for example, a flash memory, and stores data and programs used by terminal device 10. In one aspect, storage unit 170 stores user information 1701.

[0049] The user information 1701 is information about a user who uses the terminal device 10 to receive information about disease candidates, treatment candidates, and examination candidates based on input medical information, which is a function of the system 1.

[0050] The control unit 180 reads a program stored in the storage unit 170 and executes instructions included in the program to control the operation of the terminal device 10. The control unit 180 is, for example, an application that is pre-installed in the terminal device 10. The control unit 180 operates according to the program to fulfill the functions of an input operation reception unit 1801, a transmission / reception unit 1802, a data processing unit 1803, and a notification control unit 1804.

[0051] The input operation receiving unit 1801 performs processing for receiving input operations by a user via an input device such as the keyboard 131 .

[0052] The transmitting / receiving unit 1802 performs processing for the terminal device 10 to transmit and receive data to and from an external device such as the server 20 in accordance with a communication protocol.

[0053] The data processing unit 1803 performs calculations on the data that the terminal device 10 accepts as input in accordance with a program, and outputs the calculation results to a memory or the like.

[0054] The notification control unit 1804 performs processing to present information to the user. The notification control unit 1804 performs processing to display a display image on the display 150, processing to output sound to the speaker 142, and the like.

[0055] <1.2 Functional configuration of server 20> Fig. 3 is a diagram showing an example of the functional configuration of the server 20. As shown in Fig. 4, the server 20 exerts the functions of a communication unit 201, a storage unit 202, and a control unit 203.

[0056] The communication unit 201 performs processing for the server 20 to communicate with external devices.

[0057] The memory unit 202 includes, for example, prompt data 2022, teacher data 2023, and a machine learning model 2024.

[0058] The prompt data 2022 is data used when a prompt generation module 2035 (described later) generates a prompt for inputting to the machine learning model 2024 based on case information input by the user.

[0059] The prompt data 2022 is mainly composed of text data, but the data format is diverse. Specifically, various existing data formats such as JSON (JavaScript Object Notation) format, markup format, and vector expression format compatible with text information can be applied. In addition, it is also possible to convert a text-based prompt into a vector expression format in a large-scale language model and use this as input data for the machine learning model 2024. This method means that an embedding vector is provided as a prompt in a model such as a large-scale language model, and it is expected that this will realize an efficient and effective learning process and output generation process. Hereinafter, the text included in the prompt data 2022 includes a vector expression format compatible with this text information, and the description will proceed assuming that the term "text" also includes a vector expression format corresponding to this text.

[0060] As already explained, the text included in the prompt data 2022 includes input specifications for the machine learning model 2024, output examples, and textbook information for each disease, test, and treatment.

[0061] The text included in the prompt data 2022 may include at least one of a list of disease names, a list of treatment names, and a list of test names. The text included in the prompt data 2022 may also include at least one of a disease name, a treatment name, and a test name. The text included in the prompt data 2022 may also include at least one of a description of a disease, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease.

[0062] Here, the text regarding findings for diagnosing a disease may include at least one of findings that increase the probability of a diagnosis of the disease when a positive diagnosis is made for the disease, findings that decrease the probability of a diagnosis of the disease when a positive diagnosis is made for the disease, findings that decrease the probability of a diagnosis of the disease when an abnormal positive diagnosis is made for the disease, and findings of a definitive diagnosis when the diagnosis of the disease is 100%.

[0063] In addition, the text regarding findings for ruling out a diagnosis of a disease may include at least one of the following findings: findings that increase the probability of a diagnosis of the disease when a negative diagnosis for the disease is made; findings that decrease the probability of a diagnosis of the disease when a negative diagnosis for the disease is made; and findings that are useful in ruling out a diagnosis of the disease when it is 100% ruled out.

[0064] Additionally, the list of treatments or text for the treatment may include at least one of strong indications, weak indications, and contraindications for the treatment associated with the treatment.

[0065] The list of test names or the text about the test names may include at least one of strong indications, weak indications, and contraindications about the test related to the test name.

[0066] The teacher data 2023 is teacher data that serves as learning data when the machine learning model 2024 is trained. The teacher data 2023 has text data included in the above-mentioned prompt data 2022, and further includes disease candidates, treatment candidates, or test candidates corresponding to the case information, and evaluations of these disease candidates, treatment candidates, or test candidates. The evaluations include the likelihood of the disease candidates corresponding to the likelihood of the inference results of the machine learning model. The teacher data 2023 further includes data such as a list of diseases, findings regarding the diagnosis, exclusion, or evaluation of each disease (including case information and interview information, some of which are processed texts that change the interpretation of test positive / negative, etc.), textbook information about the cases, and case information.

[0067] As already explained, the teacher data 2023 may include electronic medical record information stored in an electronic medical record system (not shown). As an example, the contents of medical records of 30 cases of patients with a certain disease A and 30 cases of patients who were diagnosed with a disease other than disease A are collected.

[0068] The teacher data 2023 is based on the electronic medical record information stored in the electronic medical record system (not shown) and includes a large amount of pseudo electronic medical record information that includes positive findings, negative findings, examination dates, treatment methods, treatment start dates, definitive diagnosis dates, outcome occurrence dates such as cure, death, hospitalization, etc., the period taken until the outcome occurred, and index dates, and is replaced with the relative period taken from that date until the occurrence, using a machine learning model (large-scale language model) to create sentences based on the electronic medical record information stored in the electronic medical record system (not shown) and the positive findings, negative findings, etc. are matched. The large amount of pseudo electronic medical record information may be the machine learning model 2024 of the system 1 of this embodiment described later, or may be a machine learning model trained to generate pseudo electronic medical record information.

[0069] This makes it possible to avoid issues related to the Personal Information Protection Act, while obtaining large amounts of text, large amounts of pseudo-cases, and correct answer data to train the machine learning model 2024. As a result, it will also be possible to assist in the diagnosis of rare diseases.

[0070] The teacher data 2023 may be generated by a generative model generation module 2033 described later, or may be generated by a system other than the system 1 of this embodiment. In this case, the teacher data 2023 is transferred from the system that generated the teacher data 2023 to the system 1 of this embodiment, and the machine learning model 2024 is trained using the teacher data 2023.

[0071] One form of the machine learning model 2024 is a large-scale language model that, when a natural language or a vector is input, outputs a natural language or a vector for the natural language, as typified by ChatGPT. In particular, in the machine learning model 2024 of this embodiment, machine learning is performed using the teacher data 2023. Then, case information and the like are input to the trained machine learning model 2024, and the evaluation of the disease candidate, treatment candidate, or examination candidate, etc. as the output result is compared with the evaluation of the disease candidate, treatment candidate, or examination candidate, etc., which is the answer result of the doctor when the same input is given, and the doctor's answer is given to the machine learning model 2024 as the correct answer data, thereby performing tuning.

[0072] In particular, the machine learning model 2024 of this embodiment is trained to output, as an evaluation, disease candidates, treatment candidates, or test candidates, and the likelihood or recommendation level of these disease candidates, treatment candidates, or test candidates, in natural language and a vector expression that can be converted into natural language. Such likelihood or recommendation level may be displayed as a sentence such as a comment.

[0073] Similarly, the machine learning model 2024 of this embodiment is trained to receive input of case information and a disease candidate, and output an evaluation of the likelihood of the disease candidate (preferably on a scale of about 2 to 20). Similarly, the machine learning model 2024 of this embodiment is trained to receive input of case information and the treatment or test candidate, and output an evaluation of a recommendation for the treatment or test candidate (preferably on a scale of about 2 to 20), and an explanation of the evaluation.

[0074] The chat function, the machine learning model 2024, and the large-scale language model themselves are publicly known technologies, so a detailed description of their specific configurations and generation methods will be omitted here. However, the learning method of the machine learning model 2024 will be described later in detail.

[0075] The control unit 203 is realized by the processor 29 reading an application program 2021 stored in the storage unit 202 and executing instructions included in the application program 2021. The control unit 203 operates in accordance with the application program 2021 to perform functions indicated as a reception control module 2031, a transmission control module 2032, a generative model generation module 2033, a case information input acceptance module 2034, a prompt generation module 2035, a generative model input / output module 2036, and a presentation control module 2037.

[0076] The reception control module 2031 controls the process in which the server 20 receives a signal from an external device in accordance with a communication protocol.

[0077] The transmission control module 2032 controls the process in which the server 20 transmits signals to external devices in accordance with a communication protocol.

[0078] The generative model generation module 2033 generates a machine learning model 2024 based on the teacher data 2023 stored in the memory unit 202 of the server 20.

[0079] An example of the machine learning model 2024 generated by the generative model generation module 2033 is a machine learning model that collects, for example, 30 cases of medical records of patients with a certain disease A and 30 cases of medical records of patients diagnosed with a disease other than disease A, receives each of them together with a prompt, and outputs a graded representation of the certainty of the disease and a treatment recommendation together with an explanation. At this time, the accuracy of the output of the machine learning model 2024 is maintained by controlling the output of the machine learning model 2024 by instructing a list of candidates to be output in the prompt and the output form of the certainty / recommendation for each. This makes it possible to automatically improve the accuracy of the prompt to be input to the machine learning model through trial and error based on the currently available large amount of electronic medical record information and the graded representation of the certainty of the doctor's diagnosis for that, or the graded representation of the treatment recommendation, without having to proactively collect new patient data.

[0080] The generative model generation module 2033 may generate the pseudo electronic medical record information described above. Based on the electronic medical record information stored in the electronic medical record system (not shown), a machine learning model (large-scale language model) is used to create sentences that specify positive findings, negative findings, examination dates, treatment methods, treatment start dates, definitive diagnosis dates, outcome occurrence dates such as cure, death, hospitalization, etc., the period until the outcome occurred, and index dates, and replace them with the relative period from that date until the occurrence, and a large amount of pseudo electronic medical record information that matches the positive findings, negative findings, etc. is generated.

[0081] In generating this pseudo electronic medical record information, if the electronic medical record information contains personal information of the patient, the generative model generation module 2033 masks out part of the personal information or replaces it with personal information different from the personal information contained in the electronic medical record information. The electronic medical record information may contain personal information such as the patient's name, date of birth, contact information, and very rare diseases. The Personal Information Protection Act strictly prescribes the management method for such personal information. Therefore, if such restricted electronic medical record information is included in the teacher data 2023, it is necessary to handle it strictly and there is a risk of personal information leakage. Therefore, in the system 1 according to the present disclosure, at least a part of the case information is anonymized or modified, thereby protecting personal information and improving the convenience of data utilization.

[0082] Specifically, the actual electronic medical record information is modified by increasing or decreasing some of the numbers by a specified value, or by swapping the input contents of the same input items with those of other patients. In addition, personal information is made anonymous by masking proper nouns. At the same time, morphological analysis is performed on words, and very rare words are masked.

[0083] However, when generating pseudo-electronic medical record information, it is desirable to create pseudo-electronic medical record information based on evidence in order to maintain a certain level of medical validity. For example, fever should remain positive, headache should remain positive, test values ​​should vary within ±1 / 2 of a standard deviation, the date of discharge should vary within ±1 day, the date of re-admission should also vary within ±1 day, and the date of death should also vary within ±1 day.

[0084] In addition, the pseudo medical record information generated by the generative model generation module 2033 needs to be quality-controlled, but rather than just controlling the output with a prompt, it is also possible to generate a large amount of pseudo electronic medical record information and, based on the generated data from the products that remain within the above-mentioned fluctuations, fine-tune the generative model generation module 2033 to improve the accuracy of the generative model generation module 2033.

[0085] The case information input accepting module 2034 accepts medical information including case information input from the terminal device 10 or the like, and temporarily stores it in the storage unit 202. Preferably, the case information input accepting module 2034 accepts various information input via the input device 13 and the voice processing unit 140 of the terminal device 10, and temporarily stores it in the storage unit 202.

[0086] The case information accepted by the case information input acceptance module 2034 includes the chief complaint or disease name from the patient related to the case information, and further includes at least one of the patient's current medical history, past medical history, findings at the first visit, and test results for the patient.

[0087] When the case information input acceptance module 2034 accepts medical information including case information, the prompt generation module 2035 generates a prompt, which is text data to be input to the machine learning model 2024, based on the prompt data 2022 stored in the memory unit 202, and temporarily stores the generated prompt in the memory unit 202.

[0088] The prompt generated by the prompt generation module 2035 is the same as the text stored in the prompt data 2022. The text contained in the prompt data 2022 has already been described, so a description thereof will be omitted here.

[0089] In addition, the prompt generation module 2035 generates search words from medical information including case information accepted by the case information input acceptance module 2034, uses these search words to perform a search in the external server 40, which is an external search database, generates a prompt, which is text data to be input into the machine learning model 2024, from the search results of this external search database, and temporarily stores the generated prompt in the memory unit 202.

[0090] Any method may be used to generate search words from medical information. One example is a method of extracting noun words by morphologically analyzing medical information and using the extracted words as search words. When performing morphological analysis, it is preferable to prepare a medical dictionary (not shown) containing medical terms used by medical professionals in the storage unit 202 and perform morphological analysis using the medical dictionary. On the other hand, keeping in mind that medical information including case information is input by medical professionals, it is expected that desired search results can be obtained without using a medical dictionary, since the noun words obtained by performing morphological analysis are also preferable words for searching using a search database. Therefore, a medical dictionary is not an essential component. Also, as a method of morphological analysis, a method of extracting words using a Japanese input system such as MS IME can be suitably applied. Medical terms are terms used in medical institutions to accurately describe medical terms, such as "medical history," "current medical history," "medication history," "social history," "disease name," and "medication name."

[0091] As an example of such an external search database, a database called a knowledge graph can be suitably applied. A knowledge graph is a knowledge network that systematically links various knowledge (= knowledge) and represents it in a graph structure. It is said that the use of a knowledge graph enables data integration and integration, knowledge discovery, and advanced analysis, and also leads to process optimization and decision-making support (Reference: https: / / www.rd.ntt / se / media / article / 0025.html). As a method of expressing a knowledge graph, it is possible to clearly extract and express the site, test name, finding name, positive or negative, or disease / pathological condition and causation of disease pathological condition, such as pneumonia → respiratory failure, such as "test name = findings / positive @ site" (e.g. CT = bleeding / positive @ cerebellum).

[0092] This method of inputting search results from a search database into the machine learning model 2024 is called RAG (Retrieval-augmented Generation). RAG is a method for improving the quality of answers generated by the machine learning model 2024 by grounding the model in an external knowledge source to supplement the internal representation of knowledge held by the machine learning model 2024.

[0093] Furthermore, the text generated by the prompt generation module 2035 and used as a premise for the machine learning model 2024 to output an evaluation includes an output display example of the likelihood evaluation. Similarly, the text generated by the prompt generation module 2035 and used as a premise for the machine learning model 2024 to output an evaluation includes a state corresponding to each stage of the likelihood evaluation. Furthermore, the text generated by the prompt generation module 2035 and used as a premise for the machine learning model 2024 to output an evaluation includes an output display example of an explanation of the likelihood.

[0094] Here, the text or vector expression on which the machine learning model 2024 outputs an evaluation includes at least one of a list of disease names, a list of treatment names, and a list of test names. Alternatively, the text or vector expression on which the machine learning model 2024 outputs an evaluation includes at least one of a disease name, a treatment name, and a test name.

[0095] In addition, the text or vector representation generated by the prompt generation module 2035, on which the machine learning model 2024 outputs an evaluation, includes a text or vector representation of at least one of a description of a disease, findings for diagnosing the disease, and findings for ruling out a diagnosis of the disease.

[0096] Here, findings for diagnosing a disease include at least one of findings that increase the probability of a diagnosis of the disease when a positive diagnosis is made for the disease, findings that decrease the probability of a diagnosis of the disease when a positive diagnosis is made for the disease, findings that decrease the probability of a diagnosis of the disease when an abnormal positive diagnosis is made for the disease, and findings for a definitive diagnosis when the diagnosis of the disease is 100%.

[0097] In addition, findings for excluding a diagnosis of a disease include at least one of findings that increase the probability of a diagnosis of the disease when a negative diagnosis for the disease is made, findings that decrease the probability of a diagnosis of the disease when a negative diagnosis for the disease is made, and findings that are useful in excluding a diagnosis of the disease when it is 100% ruled out.

[0098] Additionally, the list of treatments or the text or vector representation for the treatment may include at least one of strong indications, weak indications, and contraindications for the treatment associated with the treatment.

[0099] Further, the list of test names or the text or vector representation of the test names includes at least one of strong applicable conditions, weak applicable conditions, and contraindication conditions for the test associated with the test name.

[0100] In addition, the text or vector representation generated by the prompt generation module 2035 and used as a basis for the machine learning model 2024 to output an evaluation includes display examples of the recommended evaluation. Similarly, the text or vector representation generated by the prompt generation module 2035 and used as a basis for the machine learning model 2024 to output an evaluation includes states corresponding to each stage of the recommended evaluation.

[0101] The generation model input / output module 2036 inputs medical information including case information accepted by the case information input acceptance module 2034, and prompts generated by the prompt generation module 2035, etc. into the machine learning model 2024, and causes this machine learning model 2024 to output disease candidates, treatment candidates, or test candidates corresponding to the case information input to the machine learning model 2024, as well as evaluations of these disease candidates, treatment candidates, or test candidates, and comments on the evaluations.

[0102] The evaluation output by the machine learning model 2024 includes an evaluation of the likelihood (preferably on a scale of about 2 to 20) of the disease candidate that is the output of the machine learning model 2024. Similarly, the evaluation output by the machine learning model 2024 includes an evaluation of a recommendation (preferably on a scale of about 2 to 20) of the treatment candidate or test candidate that is the output of the machine learning model 2024.

[0103] In addition, expressing the recommendation and certainty in a graded scale is useful for communication because the reasons for the graded scale can be clearly defined and clarified, and the evidence can be shown. For example, it is intuitively easier to understand if migraine is expressed in stars, with a rating of ★★★★☆, than if the probability of having a migraine is expressed as 80%.

[0104] At the same time, it is possible to clearly define each stage, such as the findings for the 5-star condition: ★★★★★ are "pulsatile, unilateral, vomiting / nausea during headache attacks, migraine history, visual field disorder during headaches, and a frequency of about once a week to several times a year," and the findings for the 4-star condition: ★★★★☆ are "pulsatile, unilateral, vomiting / nausea during headache attacks, rest during headaches, migraine history, visual field disorder during headaches, and a frequency of about once a week to several times a year," and it is possible to express the basis for the reason why the stage was evaluated. This makes it clear why the output was that way as an output representation of the large-scale language model.

[0105] In addition, for patients suspected of having migraine, structuring and inputting medical record information such as "Headache quality: pulsatile, headache location: unilateral, symptoms during headache attack: vomiting, nausea, medical history: history of migraine, complications during headache: visual field disorder, headache frequency: once a week to several times a year, TSH elevated" will standardize the input and stabilize the accuracy of the output, as well as improve the user's understanding of the machine learning results.

[0106] In this way, by structuring the medical record information into expected findings / outcomes that are important for diagnostic / treatment decisions, and then entering this structured content into a prompt, the stage expression output by the generative AI can be compared with an explanation of the basis for that stage expression, making it possible to manage the accuracy of the generative AI.

[0107] In addition, by designing the items of the structured data created in this way to be included in the input as prompts such as "findings that increase the probability of diagnosis of a disease when a positive or negative diagnosis is made for a disease" or "findings that decrease the probability of diagnosis of a disease when a positive or negative diagnosis is made for a disease" or "findings that are useful for excluding a disease" in the machine learning model 2024, expressions of findings related to the diagnosis, exclusion, or evaluation of each disease to be written in the prompt are automatically created based on sentences structured by the large-scale language model from the electronic medical record information. In this way, by automatically creating evaluation criteria for each disease and selecting prompts that match the doctor's evaluation, it is possible to create prompts that include evaluation criteria for each disease that can obtain highly accurate results.

[0108] While this example describes the generation of prompts for diagnosis, it is also possible to generate prompts for recommending treatments and examinations by automatically creating evaluation criteria for each treatment or examination from electronic medical record information and the doctor's evaluation of the recommended treatment or examination.

[0109] When inputting medical information into the machine learning model 2024, if the case information accepted by the case information input acceptance module 2034 contains personal information of the patient, the generation model input / output module 2036 masks part of the personal information or replaces it with personal information different from the personal information accepted by the case information input acceptance module 2034. The case information may contain personal information such as the patient's name, date of birth, contact information, and very rare disease. The Personal Information Protection Act strictly prescribes management methods for such personal information. Therefore, inputting such restricted case information into the machine learning model 2024 requires strict handling and carries the risk of personal information leakage. Therefore, in the system 1 according to the present disclosure, at least a part of the case information is anonymized or modified, thereby protecting personal information and improving the convenience of data utilization.

[0110] Specifically, the actual case information is altered by increasing or decreasing some of the numbers by a specified value, or by swapping the input contents of the same input items with the case information of other patients. In addition, personal information is made anonymous by masking proper nouns. At the same time, morphological analysis is performed on words, and very rare words are masked.

[0111] Furthermore, the generation model input / output module 2036 may cause the machine learning model 2024 to output disease candidates, treatment candidates, or test candidates corresponding to the case information input to the machine learning model 2024, and evaluations of these disease candidates, treatment candidates, or test candidates, without inputting the prompt generated by the prompt generation module 2035 to the machine learning model 2024. In other words, without inputting a prompt, the machine learning model 2024 may be fine-tuned so that case information is input and disease candidates or test / treatment candidates, evaluations therefor, and an explanation of the evaluations are output from the machine learning model 2024, whereby disease candidates, treatment candidates, or test candidates corresponding to the case information input to the machine learning model 2024, and evaluations of these disease candidates, treatment candidates, or test candidates, and an explanation of the evaluations are output from the machine learning model 2024.

[0112] It is also possible to fine-tune the machine learning model 2024 using data generated by the generative model generation module 2033. In one example, when the generative model generation module 2033 fine-tunes the machine learning model 2024, the generative model generation module 2033 may also input a prompt generated by the prompt generation module 2025 to the machine learning model 2024. Also, similar to the case of generating pseudo electronic medical record information, teacher data for fine tuning may be generated by a machine learning model separate from the machine learning model 2024.

[0113] As an example, the teacher data for fine tuning includes case information as input, and a recommendation and a graded notation of likelihood as output, and an explanation based on the graded notation. If the recommendation and likelihood described in the generated output match a recommendation and likelihood separately given by a doctor, the explanatory expression included in the output is likely to match the doctor's opinion, which is effective in improving the accuracy of pre-learning of the machine learning model 2024. Therefore, a large amount of teacher data for fine tuning may be generated by a generative model, and data that matches or is close to the doctor's graded notation may be selected from the created data and used for pre-learning of the machine learning model 2024.

[0114] In summary, these multiple generative AIs create structured case texts, pseudo case texts, and prompts that serve as evaluation criteria from case texts and the evaluations of the AI ​​and doctors, and also create correct answer data for fine-tuning with explanatory text for the evaluation criteria, and texts for pre-learning of the generative AI. In this way, it is revolutionary that correct answer data and machine learning models for AI are automatically created based on case texts and the staged evaluations of the AI ​​and doctors.

[0115] The presentation control unit 2037, in cooperation with the transmission control module 2032, transmits the disease candidates, treatment candidates, or test candidates corresponding to the case information input to the machine learning model 2024 output by the machine learning model 2024, and the evaluations of these disease candidates, treatment candidates, or test candidates, to the terminal device 10 operated by a medical professional, and displays them on the display 241 of the terminal device 10.

[0116] <2 Example of operation> An example of the operation of the server 20 will now be described.

[0117] FIG. 4 is a flowchart showing an example of the operation of the server 20. FIG. 4 is a diagram showing an example of the operation of generating teacher data 2023 for training a machine learning model 2024. Case information in an electronic medical record is converted into structured data (pairs of organized item names and values) with matching outcomes and findings using a large-scale language model, and then a large amount of pseudo-case sentences with matching or similar outcomes and findings are created from the structured data. Structured data has the advantage that it is easy to check whether it is anonymized, and the machine learning model is easy to understand.

[0118] 4, the control unit 203 obtains electronic medical record information of an electronic medical record system outside the system 1 or prepared in the storage unit 202. Specifically, for example, the control unit 203 obtains electronic medical record information of an electronic medical record system outside the system 1 or prepared in the storage unit 202 by the generative model generation module 2033. An example of electronic medical record information is shown in FIG.

[0119] In step S401, the control unit 203 structures the findings assumptions important for diagnosis for the electronic medical record information acquired in step S400. Specifically, for example, the control unit 203 structures the findings assumptions important for diagnosis for the electronic medical record information acquired in step S400 by the generative model generation module 2033. More specifically, the control unit 203 creates a sentence in which the findings assumptions / outcomes important for diagnosis / treatment judgments are structured by AI using the generative model generation module 2033. An example of structuring the findings assumptions is shown in the left part of FIG. 13. As shown in FIG. 13, the electronic medical record information acquired in step S400 is converted into structured data, and the outcomes are further extracted and converted into structured data. At this time, natural language processing can be performed on the creation of the structured data. In one example, the medical record sentence may be pre-processed to convert the test values ​​into abstractions such as normal, abnormal, high value, low value, etc. Also, structuring may be performed using a large-scale language model, and the structuring items may be specified by prompts, or a reference value for determining whether a test value is increased or decreased may be passed. One example may be written as follows: "For structuring, extract the chief complaint, duration of the chief complaint, severity of the chief complaint (in the case of pain, it is expressed as 1 to 10), complications, quality of the chief complaint, and medical history. Follow the example below. Chief complaint: headache. Duration of the chief complaint: 3 days, Severity of the chief complaint: 5, TSH increased."

[0120] In steps S402 to S403, the control unit 203 generates pairs of a large amount of (pseudo electronic) medical record information and outcomes, likelihood of disease name, and examination and treatment recommendations based on the structured data generated in step S401. Specifically, for example, the control unit 203 creates a large amount of (pseudo electronic) medical record information based on the structured data generated in step S401 using the generative model generation module 2033, and pairs of outcomes and likelihood of disease name and examination and treatment recommendations are assigned by the generative model generation module 2033 or a doctor. An example of the generated pseudo electronic medical record information is shown in the right part of FIG. 13. The findings and outcomes of these pseudo electronic medical record information are consistent, and some of the personal information, etc. has been replaced.

[0121] In step S404, the control unit 203 generates the machine learning model 2024 using pairs of the (pseudo electronic) medical record information, outcome, certainty of disease name, and examination and treatment recommendation generated in steps S402 to S403, or fine-tunes the machine learning model 2024 using pairs of the (pseudo electronic) medical record information, outcome, certainty of disease name, and examination and treatment recommendation generated in steps S402 to S403. Specifically, for example, the control unit 203 generates the machine learning model 2024 by the generative model generation module 2033 using pairs of the (pseudo electronic) medical record information, outcome, certainty of disease name, and examination and treatment recommendation generated in steps S402 to S403, or fine-tunes the machine learning model 2024 using pairs of the (pseudo electronic) medical record information, outcome, certainty of disease name, and examination and treatment recommendation generated in steps S402 to S403.

[0122] Fig. 5 is a flowchart showing another example of the operation of the server 20. Fig. 5 is a diagram showing an example of the operation of the prompt generation module 2035 of the control unit 203 to generate a prompt.

[0123] In step S500, the control unit 203 generates a wide variety of prompts based on expected findings that are important for an ideal diagnosis from the electronic medical record information and output examples of the machine learning model 2024. Specifically, for example, the control unit 203 generates a wide variety of prompts using the prompt generation module 2035 based on expected findings that are important for an ideal diagnosis from the electronic medical record information and output examples of the machine learning model 2024.

[0124] Next, in step S501, the control unit 203 verifies the accuracy of the prompt generated in step S500. Specifically, for example, the control unit 203 verifies the accuracy of the prompt generated in step S500 by the prompt generation module 2035. More specifically, the control unit 203 inputs the prompt generated in step S500 to the machine learning model 2024 by the prompt generation module 2035 to obtain an output, and performs accuracy verification for each prompt. Note that this accuracy verification may be performed by a medical professional.

[0125] Then, in step S502, the control unit 203 selects a prompt with high accuracy based on the verification result in step S501, or selects a large-scale language model capable of creating a prompt with high accuracy. Specifically, for example, the control unit 203 selects a prompt with high accuracy based on the verification result in step S501 by the prompt generation module 2035, or selects a large-scale language model capable of creating a prompt with high accuracy. This allows a prompt with high diagnostic performance / recommendation evaluation performance to be created. In the verification process, a large number of prompts and a large number of medical record information are applied to the machine learning model 2024, and it is possible to obtain output text including a large number of evaluations and explanations of the evaluations. Of these, output texts whose evaluations match the doctor's evaluation are selected, and the output texts are used as learning data for the machine learning model 2024. The sentences of the evaluation explanations included in the output texts that match the doctor's evaluation are likely to be true, and the accuracy can be improved by using highly truthful texts for learning the machine learning model 2024.

[0126] Furthermore, based on the "findings useful for diagnosis" obtained in step S500 and described in the prompt, the "expected findings important for diagnosis" of the structured text generated by the generative model generation module 2033 are updated, and the consistency between the structured text output from the generative model generation module 2033 and the machine learning model 2024 is improved, and as a result, the accuracy of the entire system can be improved. In one example of a specific implementation method, a list of "expected findings important for diagnosis" is input to the generative model generation module 2033 as a prompt, and the output of the generative model generation module 2033 is controlled. Note that, in the output of S502, it is useful to select information or explanatory text that is consistent with the clinician's diagnosis and input it as text for pre-training of the large-scale language model in step S404 of FIG. 4 in order to improve the accuracy of the large-scale language model in S404.

[0127] Fig. 6 is a flowchart showing an example of the operation of the server 20. Fig. 6 is a diagram showing an example of the operation, which starts with a medical worker inputting medical information including case information via the terminal device 10, inputting the medical information and a prompt to the machine learning model 2024, and outputting disease candidates, treatment candidates, or test candidates corresponding to the input case information from the machine learning model 2024, and evaluations of the disease candidates, treatment candidates, or test candidates.

[0128] In step S600, the control unit 203 accepts medical information including medical history information input by a medical professional via the terminal device 10. Specifically, for example, the control unit 203 accepts medical information including medical history information input by a medical professional via the terminal device 10 by the case information input acceptance module 2034. Then, the case information input acceptance module 2034 temporarily stores the accepted medical information in the storage unit 202.

[0129] In step S601, the control unit 203 generates a prompt to be input to the machine learning model 2024 based on the medical information accepted in step S450. Specifically, for example, the control unit 203 generates a prompt to be input to the machine learning model 2024 based on the medical information accepted in step S450 by the prompt generation module 2035. Note that in step S601, the prompt generation module 2035 may obtain the prompt using an external search database or the like.

[0130] In step S602, the control unit 203 inputs the medical information accepted in step S600 and the prompt generated in step S601, etc., to the machine learning model 2024. Specifically, for example, the control unit 203 inputs the medical information accepted in step S450 and the prompt generated in step S451, etc., to the machine learning model 2024 via the generation model input / output module 2036.

[0131] In step S603, the control unit 203 accepts an output from the machine learning model 2024 which is a result of inputting medical information, etc. to the machine learning model 2024 in step S452. Specifically, for example, the control unit 203 accepts, via the generation model input / output module 2036, an output from the machine learning model 2024 which is a result of inputting medical information, etc. to the machine learning model 2024 in step S452.

[0132] In step S604, the control unit 203 generates a display control signal for displaying the output from the machine learning model 2024 accepted in step S603 on the display 150 of the terminal device 10, and sends this display control signal to the terminal device 10. Specifically, for example, the control unit 203 generates a display control signal for displaying the output from the machine learning model 2024 accepted in step S453 on the display 150 of the terminal device 10 by the presentation control unit 2037, and sends this display control signal to the terminal device 10.

[0133] In step S605, the control unit 203 verifies the accuracy of the output of step S604, and selects an output that can be determined to be correct and an input corresponding to this output (this input can also be determined to be correct). Specifically, for example, the control unit 203 verifies the accuracy of the output of step S604 using the generative model input / output module 2036, and selects an output that can be determined to be correct and an input corresponding to this output (this input can also be determined to be correct). After that, in step S607, the correct input is used as the input for step S404 in FIG. 4.

[0134] <3 Screen example> Hereinafter, an example of a screen output to the terminal device 10 will be described with reference to FIGS.

[0135] 7 to 10 are example screens showing an example of medical information and prompts, etc., input by the control unit 203 of the server 20 to the machine learning model 2024 in the system 1 of this embodiment. The medical information, etc. input to the machine learning model 2024 is also displayed on the display 150 of the terminal device 10 by the presentation control unit 2037 of the control unit 203 of the server 20, and the example screens in Figs. 5 to 8 are example screens displayed on the display 150 of the terminal device 10.

[0136] The medical information, etc. input to the machine learning model 2024 includes the specification of the input contents for the machine learning model 2024, an output example, and textbook information for each case. The output example includes disease candidates, treatment candidates, or test candidates based on the case information, and further includes evaluations of these disease candidates, treatment candidates, or test candidates. In the example screen, the evaluation is output on a 5-point scale, with the number of stars representing the evaluation points. As already described, the textbook information may include search results from the external server 40, which is a search database. In the example screen, the textbook information includes the diagnosis, exclusion, or findings regarding each evaluation point for each disease.

[0137] 11 is a screen example showing an example of an output of the machine learning model 2024 in the system 1 of this embodiment. As already described, the output from the machine learning model 2024 is also displayed on the display 150 of the terminal device 10 by the presentation control unit 2037.

[0138] 7 to 11 show inputs and outputs to the machine learning model 2024 in a markup format, and FIG. 11 shows text data in this markup format converted into a table format.

[0139] <4. Effects of one embodiment> As described in detail above, according to the system 1 of this embodiment, as a system for obtaining information for diagnosing a patient, it is possible to provide a technology that maintains sufficient accuracy by using medical record information as input and controlling a machine learning model in natural language or vector expressions using prompts.

[0140] <5 Notes> In addition, the above-described embodiments are described in detail to clearly explain the present disclosure, and are not necessarily limited to those including all of the described configurations. In addition, some of the configurations of each embodiment can be added to, deleted from, or replaced with other configurations.

[0141] As an example, in the embodiment described above, the server 20 has the machine learning model 2024, but the external server 40 in Fig. 1 may have the machine learning model. In this case, the learning of the machine learning model may be mainly performed by the external server 40 (i.e., from the operation of collecting teacher data to the operation of learning the machine learning model based on this teacher data), or the server 20 may learn the machine learning model in the external server 40 (i.e., from the operation of collecting teacher data to the operation of learning the machine learning model based on this teacher data).

[0142] In the above embodiment, at least one of disease candidates, treatment candidates, and test candidates based on case information input by a user and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates are obtained using a machine learning model, but such output may be obtained using other methods other than the machine learning model, for example, a rule-based database, or from a dedicated application (second program) stored in memory. An example of such a dedicated application will be described with reference to Figures 14 to 20.

[0143] For example, given the following text: Positive symptoms of cold (100%): fever, cough, phlegm, runny nose, or sore throat Positive symptoms of cold (80%): sore throat, runny nose, cough Positive findings of cold (80%): Cough Positive symptoms of cold (30%): fever Positive symptoms of cold (10%): chills, muscle pain, fatigue, nausea Positive findings that rule out a cold (90%): Influenza positive, X-ray = pulmonary infiltrative shadow 14 to 19 are stored in the storage unit 202 of the server 20, and data is extracted from the above text and input into this program, and it is possible to obtain an output in which the post-diagnosis probability is displayed as follows: up to 10%, ☆, 10-30%, ☆☆, 40-60%, ☆☆☆, 70-90%, ☆☆☆☆, and 90% or more, ☆☆☆☆☆, as shown in Figure 20. Such ☆ display is one example of the evaluation.

[0144] The program shown in Figures 14 to 19 extracts findings related to the diagnosis of cold and their frequency from text, and estimates the probability of diagnosing cold based on observed symptoms. First, the following cold symptom list text is input into the program. "Positive findings in 100% of cases of the common cold: fever, cough, phlegm, runny nose, or sore throat. Positive findings in 80% of cases of the common cold: sore throat, runny nose, cough. Positive findings in 30% of cases of the common cold: fever. Positive findings in 10% of cases of the common cold: chills, muscle pain, fatigue, nausea. Positive findings that rule out the common cold in 90% of cases: positive for influenza, X-ray = pulmonary infiltrate shadow."

[0145] The program then inputs text describing the patient's observed symptoms and extracts the symptoms from the text. "The patient presented with a suspected cold and presented with symptoms of cough, phlegm, and runny nose."

[0146] The programs shown in Figures 14 to 19 analyze the contents of the findings list text using regular expressions and store the frequency of occurrence of each symptom in dictionary format. Next, symptoms that match those extracted in advance are extracted from the patient's symptom text. Bayesian estimation is performed based on the obtained symptoms and prior probability, and the posterior probability of having a cold is calculated. The diagnostic probability is evaluated with a number of stars (☆ to ☆☆☆☆☆), and the certainty of the diagnosis is displayed.

[0147] In addition, the above-mentioned configurations, functions, processing units, processing means, etc. may be realized in part or in whole by hardware, for example, by designing them as integrated circuits. The present invention can also be realized by software program code that realizes the functions of the embodiments. In this case, a storage medium on which the program code is recorded is provided to a computer, and a processor included in the computer reads the program code stored in the storage medium. In this case, the program code itself read from the storage medium realizes the functions of the above-mentioned embodiments, and the program code itself and the storage medium storing it constitute the present invention. Examples of storage media for supplying such program code include flexible disks, CD-ROMs, DVD-ROMs, hard disks, SSDs, optical disks, magneto-optical disks, CD-Rs, magnetic tapes, non-volatile memory cards, ROMs, etc.

[0148] Furthermore, the program code for realizing the functions described in this embodiment can be implemented in a wide range of program or script languages, such as assembler, C / C++, perl, Shell, PHP, Java (registered trademark), and the like.

[0149] Furthermore, the program code of the software that realizes the functions of the embodiments may be distributed over a network and stored in a storage means such as a computer's hard disk or memory, or in a storage medium such as a CD-RW or CD-R, and the processor of the computer may read out and execute the program code stored in the storage means or storage medium.

[0150] The matters described in the above embodiments will be supplemented below. (Appendix 1) A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory storing a second program that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the processor to execute a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or a treatment name related to the case information, a state of strong application, a state of weak application, and a state of weak application related to the treatment name, a seventh step of inputting into a second program at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is output from the second step, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and in the second step, outputting as the evaluation, the certainty or recommendation level of the at least one of the disease candidate, the treatment candidate, and the test candidate. (Appendix 2) A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the processor to perform a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or a treatment name related to the case information, a state of strong application, a state of weak application related to the treatment name, a seventh step of inputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and a text or vector representation, and an evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, which are output from the machine learning model; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate, which are output from the second step, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, and in which in the second step, the certainty or recommendation level of at least one of the disease candidate, the treatment candidate, and the test candidate, is output as the evaluation. (Appendix 3) A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the processor to perform a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or and an eighth step of inputting at least one of a strong indication, a weak indication, and a contraindication for the examination into the machine learning model; a fourth step of outputting from the machine learning model at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector expression input into the machine learning model, and an evaluation of the disease candidate, the treatment candidate, and the test candidate; and a fifth step of outputting information for presenting to a user the disease candidate, the treatment candidate, and the test candidate, which are the output of the fourth step, and the evaluation of the disease candidate, the treatment candidate, and the test candidate. In the fourth step, the machine learning model outputs, as the evaluation, the likelihood or recommendation level of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation level of the disease candidate, the treatment candidate, and the test candidate.and a machine learning model that is trained by fine tuning to output at least one of the test candidates. (Appendix 4) The program described in Appendix 2, wherein the case information accepted in the first step includes a chief complaint or disease name from the patient related to the case information, and further includes at least one of the patient's current medical history, medical history, findings, treatment history, and test findings for the patient, and the machine learning model is a machine learning model trained to execute the second step and the third step for the input of the seventh step, and in the second step, outputs, as an evaluation, at least one of disease candidates, treatment candidates, and test candidates, and the certainty or recommendation level for at least one of the disease candidates, treatment candidates, or test candidates in natural language. (Appendix 5) In the seventh step, the program described in Appendix 4 structures expected findings that are important for diagnostic / treatment decisions based on the electronic medical record information as the case information accepted in the first step, and inputs the structured content into a machine learning model. (Appendix 6) The program described in Appendix 2, further comprising a processor executing a sixth step of searching an external search database for information related to the case information and obtaining or generating a text or vector representation from the search results, and further comprising a seventh step of inputting, in addition to the case information, the text or a vector representation convertible to text that is the prerequisite for outputting the evaluation generated in the sixth step into a machine learning model. (Appendix 7) 7. The program of claim 6, wherein in a sixth step, the program searches a knowledge graph database as an external search database. (Appendix 8) The program described in Appendix 4, wherein in the seventh step, the text or vector representation serving as a premise for outputting the evaluation includes at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or a treatment name related to the case information, and at least one of a strongly applicable condition, a weakly applicable condition, and a contraindication condition related to the treatment name, or a test name related to the case information, and at least one of a strongly applicable condition, a weakly applicable condition, and a contraindication condition for the test related to the test name. (Appendix 9) 7. The program of claim 6, wherein in a sixth step, the text or vector representation includes a text or vector representation of at least one of a description of the disease, findings for diagnosing the disease, and findings for ruling out the diagnosis of the disease. (Appendix 10) The program of Appendix 9, wherein in the sixth step, the findings for diagnosing the disease include at least one of the following findings: a finding that increases the probability of a diagnosis of the disease when a positive judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when a positive judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when an abnormal positive judgment is made for the disease, and a definitive diagnosis finding when the diagnosis of the disease is 100%. (Appendix 11) The program described in Appendix 9, wherein in the sixth step, the findings for excluding a diagnosis of the disease include at least one of a finding that increases the probability of a diagnosis of the disease when a negative judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when a negative judgment is made for the disease, and a finding that is useful for excluding a diagnosis of the disease when it is 100% ruled out. (Appendix 12) The program of claim 9, wherein in a sixth step, the text or vector representation of the list of treatment names includes at least one of strong indications, weak indications, and contraindications for the treatment associated with the treatment name. (Appendix 13) The program of claim 9, wherein in a sixth step, the text or vector representation of the list of test names includes at least one of a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the test related to the test name. (Appendix 14) The program of claim 2, wherein the evaluation presented in the third step includes an evaluation of likelihood for the disease candidate as an output of the machine learning model. (Appendix 15) The program according to claim 6, wherein the text or vector representation that is the premise for outputting the evaluation input in the seventh step includes an output display example of the likelihood or recommendation evaluation. (Appendix 16) The program described in Appendix 6, in which the text or vector representation that is the premise for outputting the evaluation input in the seventh step contains states corresponding to each stage of the evaluation of likelihood or recommendation level. (Appendix 17) The program according to claim 6, wherein the text or vector representation that is the premise for outputting the evaluation input in the seventh step includes an output display example of the likelihood explanation. (Appendix 18) The program of Appendix 2, wherein the evaluation presented in the third step includes an evaluation of a recommendation for a treatment or test candidate that is an output of the machine learning model. (Appendix 19) The program of claim 6, wherein the text or vector representation that is the basis for outputting the evaluation input in the seventh step includes display examples of the recommended evaluation. (Appendix 20) A program as described in Appendix 6, in which the text or vector representation that is the premise for outputting the evaluation input in the seventh step contains states that correspond to each stage of the recommendation evaluation. (Appendix 21) The machine learning model is a machine learning model trained to create pseudo-case information from case information in electronic medical records, and the pseudo-case information is created by generating structured data that matches findings and outcomes from patient medical record information and inputting the structured data into the machine learning model, as described in Appendix 4. (Appendix 22) The program described in Appendix 4, wherein the machine learning model is trained to output one of an assessment of the likelihood or strength of recommendation for a disease candidate, or an assessment of a recommendation for a treatment or test candidate or an explanation of the strength of recommendation, or an explanation of the strength of recommendation. (Appendix 23) A program as described in Appendix 2, which in a third step outputs information to present to a user one of a disease candidate, a treatment candidate, or a test candidate, and an evaluation of the one of the disease candidate, treatment candidate, or test candidate, as its likelihood or degree of recommendation. (Appendix 24) A program as described in Appendix 23, which in a third step outputs information for presenting to a user one of a disease candidate, a treatment candidate, or a test candidate, and an evaluation of the disease candidate, treatment candidate, or test candidate, in the form of a graded expression of its likelihood or degree of recommendation. (Appendix 25) The program described in Appendix 24, in a third step, outputs information for presenting to the user the likelihood or degree of recommendation of one of the disease candidates, treatment candidates, or test candidates as an evaluation of the disease candidate, treatment candidate, or test candidate in a graded format, and also outputs information for presenting to the user the basis for the graded format. (Appendix 26) The program described in Appendix 25, in which the text or vector representation that is input in the seventh step and is the premise for outputting the evaluation describes a state corresponding to each stage of the evaluation of the likelihood or degree of recommendation, and in the third step, information is output for presenting to the user one of the disease candidates, treatment candidates, or test candidates, and the likelihood or degree of recommendation as an evaluation of this one of the disease candidates, treatment candidates, or test candidates in a staged expression, and information is output for presenting to the user the basis for the staged expression. (Appendix 27) In a medical support system comprising a computer and a server, which outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, a program for operating a computer comprising a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the server comprising a second processor and a second memory, the second memory storing a second program for receiving case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a disease name, treatment name, and test name, and outputting at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the first processor to perform a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name, treatment name, and test name, among the disease name, disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, a seventh step of inputting at least one of the case information, or at least one of the treatment name, strong applicable state, weak applicable state, and contraindication state for the treatment name, or at least one of the test name, strong applicable state, weak applicable state, and contraindication state for the test related to the test name, into a second program; a second step of outputting from the second program at least one of the disease candidate, treatment candidate, and test candidate corresponding to the case information and the text or vector representation inputted to the second program, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate outputted from the second step, and the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, and in the second step, outputs as the evaluation, the certainty or recommendation degree of the at least one of the disease candidate, treatment candidate, and test candidate, and the at least one of the disease candidate, treatment candidate, or test candidate. (Appendix 28) In a medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, a program for operating a computer comprising a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the server comprising a second processor and a second memory, the second memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the first processor to perform a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, a seventh step of inputting at least one of the case information, a treatment name, a strongly applicable state, a weakly applicable state, and a contraindicated state for the treatment name, or at least one of the test name, a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name, into a machine learning model; a second step of outputting from the machine learning model at least one of the disease candidate, treatment candidate, and test candidate corresponding to the case information and the text or vector representation inputted to the machine learning model, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate outputted from the second step, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, (Appendix 29) In a medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, a program for operating a computer comprising a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on case information, the server comprising a second processor and a second memory, the second memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the program causing the first processor to perform a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for ruling out the diagnosis of the disease, which are text or vector expressions. an eighth step of inputting into the machine learning model at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of a test name related to the case information, and a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name; and at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the machine learning model, and and a fifth step of outputting information for presenting to a user at least one of the disease candidate, treatment candidate, and test candidate, which are the output of the fourth step, and an evaluation of at least one of the disease candidate, treatment candidate, and test candidate. In the fourth step, at least one of the disease candidate, treatment candidate, and test candidate, and a likelihood or recommendation level of at least one of the disease candidate, treatment candidate, or test candidate are output as the evaluation. The machine learning model outputs the input case information, a list of disease names, a list of treatment names,and a machine learning model that has been trained by fine tuning to output at least one of a disease candidate, a treatment candidate, and a test candidate for at least one of the lists of test names. (Appendix 30) An information processing device including a processor and a memory, the information processing device outputs at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the memory storing a second program that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a disease name, a treatment name, and a test name, and outputs at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, and an evaluation for at least one of the disease candidate, the treatment candidate, and the test candidate, the first step of accepting an input of case information from a user, and an evaluation for at least one of the disease candidate, the treatment candidate, and the test candidate based on the case information, the first step of accepting an input of case information from a user, the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing a disease, and a finding for excluding a diagnosis of a disease, or a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or a seventh step of inputting into a second program at least one of the case information input into the second program, and a condition in which the test related to the disease name, treatment name, and test name is strongly applicable, a condition in which the test related to the test name is weakly applicable, and a condition in which the test is contraindicated; a second step of causing the second program to output at least one of disease candidates, treatment candidates, and test candidates corresponding to the case information input into the second program and a text or vector expression that is a premise for outputting an evaluation of at least one of the disease name, treatment name, and test name, and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates; and a third step of outputting information for presenting to a user the at least one of the disease candidates, treatment candidates, and test candidates that are output from the second step, and the evaluation of at least one of the disease candidates, treatment candidates, and test candidates, and in the second step, outputs as an evaluation, a likelihood or a degree of recommendation of at least one of the disease candidates, treatment candidates, or test candidates. (Appendix 31) An information processing device including a processor and a memory, the information processing device outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory stores case information and a text or vector expression serving as a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the processor includes a first step of accepting input of case information from a user, and a machine learning model that accepts the case information accepted in the first step and outputs, as text or vector expressions, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or an information processing device which executes a seventh step of inputting at least one of the case name, the condition for which the test related to the test name is strongly applicable, the condition for which the test related to the test name is weakly applicable, and the condition for which the test related to the test name is contraindicated into a machine learning model; a second step of causing the machine learning model to output at least one of disease candidates, treatment candidates, and test candidates corresponding to the case information inputted into the machine learning model and the text or vector expression that is the premise for outputting an evaluation for at least one of the disease name list, the treatment name list, and the test name list, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates; and a third step of outputting information for presenting to a user the at least one of the disease candidates, treatment candidates, and test candidates that are the output of the second step, and the evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and in the second step, outputs as an evaluation, a likelihood or a recommendation level for at least one of the disease candidates, treatment candidates, or test candidates. (Appendix 32) An information processing device including a processor and a memory, the information processing device outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory stores case information and a text or vector expression serving as a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates, the processor includes a first step of accepting input of case information from a user, and a text or vector expression serving as a premise for outputting the evaluation of at least one of the list of disease names, the list of treatment names, and the list of test names, the first step of accepting input of case information from a user, and at least one of a disease name related to the case information, a description of a disease related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or and an eighth step of inputting into the machine learning model at least one of a test name related to the test name, and a state in which the test related to the test name is strongly applicable, a state in which the test related to the test name is weakly applicable, and a state in which the test related to the test name is contraindicated; a fourth step of outputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the machine learning model, and an evaluation of the disease candidate, treatment candidate, and test candidate; and a fifth step of outputting information for presenting to a user the disease candidate, treatment candidate, and test candidate, which are the output of the fourth step, and the evaluation of the disease candidate, treatment candidate, and test candidate. In the fourth step, the machine learning model outputs, as the evaluation, the likelihood or recommendation level of the disease candidate, treatment candidate, and test candidate, and the likelihood or recommendation level of the disease candidate, treatment candidate, and test candidate for the input case information, and at least one of the disease name list, treatment name list, and test name list.and an information processing device, the information processing device being a machine learning model that has been trained by fine tuning to output at least one of the test candidates. (Appendix 33) In a medical support system including an information processing device and a server, which outputs disease candidates, treatment candidates, or test candidates based on case information, the information processing device includes a first processor and a first memory, and the information processing device outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information. The server includes a second processor and a second memory, and the second memory stores a second program that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name, treatment name, and test name, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates. The first processor performs a first step of accepting input of case information from a user, and outputs the case information accepted in the first step, and at least one of the case information-related disease name, disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or a seventh step of inputting into a second program at least one of a treatment name to be treated and a strong applicable state, a weak applicable state, and a contraindicated state for the treatment name, or at least one of a test name related to the case information and a strong applicable state, a weak applicable state, and a contraindicated state for the test related to the test name; a second step of outputting from the second program at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the second program, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is the output of the second step, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and in the second step, outputting as the evaluation the likelihood or recommendation level of the at least one of the disease candidate, the treatment candidate, and the test candidate. (Appendix 34) In a medical support system including an information processing device and a server, the system outputs disease candidates, treatment candidates, or test candidates based on case information. The information processing device includes a first processor and a first memory, and the information processing device outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information. The server includes a second processor and a second memory, and the second memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates. The first processor performs a first step of accepting input of case information from a user, and outputs the case information accepted in the first step and at least one of the case information, which are text or vector expressions, of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, or and a seventh step of inputting into a machine learning model at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, and a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation inputted to the machine learning model, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is outputted from the second step, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and in the second step, outputting a likelihood or a recommendation level of the at least one of the disease candidate, the treatment candidate, and the test candidate, as the evaluation. (Appendix 35) In a medical support system comprising an information processing device and a server, which outputs disease candidates, treatment candidates, or test candidates based on case information, the information processing device comprises a first processor and a first memory, and the information processing device outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information. The server comprises a second processor and a second memory, and the second memory stores a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name list, treatment name list, and test name list, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates. The first processor performs a first step of accepting input of case information from a user, and a second step of processing the case information received in the first step, and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name list, treatment name list, and test name list, and outputs at least one of the disease candidates, treatment candidates, and test candidates related to the case information. an eighth step of inputting at least one of the case information, or at least one of the treatment name, the strongly applicable state, the weakly applicable state, and the contraindicated state for the treatment name, or at least one of the test name, the strongly applicable state, the weakly applicable state, and the contraindicated state for the test name, into the machine learning model; and at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation inputted into the machine learning model, and at least one of the disease candidates, the treatment candidates, and the test candidates. and a fifth step of outputting information for presenting to a user at least one of the disease candidate, treatment candidate, and test candidate, which is the output of the fourth step, and the evaluation of at least one of the disease candidate, treatment candidate, and test candidate. In the fourth step, at least one of the disease candidate, treatment candidate, and test candidate, and the likelihood or recommendation level of at least one of the disease candidate, treatment candidate, or test candidate are output as the evaluation. The machine learning model outputs the input case information, the list of disease names,An information processing device that outputs at least one of a disease candidate, a treatment candidate, and a test candidate for at least one of a list of treatment names and a list of test names, and a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate. (Appendix 36) A method executed by a computer having a processor and a memory, the computer outputs at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the memory storing a second program for receiving as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of the disease name, the treatment name, and the test name, and outputting at least one of the disease candidate, the treatment candidate, and the test candidate based on the case information, and an evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, the processor performing a first step of accepting input of case information from a user, and inputting the case information accepted in the first step, and at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, which are text or vector expressions, or the treatment name related to the case information, the strongly applicable condition, the weakly applicable condition, and the contraindicated condition related to the treatment name, a seventh step of inputting at least one of the conditions of the case, or at least one of the test name related to the case information, and the conditions in which the test related to the test name is strongly applicable, weakly applicable, and contraindicated, into a second program; a second step of outputting from the second program at least one of the disease candidate, treatment candidate, and test candidate corresponding to the case information and the text or vector representation inputted into the second program, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate outputted from the second step, and the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, (Appendix 37) A method executed by a computer having a processor and a memory, the computer outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, the processor including: a first step of accepting input of case information from a user; and inputting the case information accepted in the first step, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindication related to the treatment name, which are text or vector expressions. a seventh step of inputting at least one of a state in which the test related to the case information is a condition in which the test is strongly applicable, a state in which the test related to the test name is a condition in which the test is strongly applicable, a state in which the test is weakly applicable, and a state in which the test related to the test name is a contraindication to the case information into a machine learning model; a second step of outputting from the machine learning model at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input to the machine learning model, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is the output of the second step, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, (Appendix 38) A method executed by a computer having a processor and a memory, the computer outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, the memory storing a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates, the processor including a first step of accepting input of case information from a user, and a text or vector expression that is a premise for outputting the case information and an evaluation of at least one of the list of disease names, the list of treatment names, and the list of test names, the first step including a disease name related to the case information, a disease description related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or The method includes an eighth step of inputting into the machine learning model at least one of a test name related to the case information, and a state in which the test related to the test name is strongly applicable, a state in which the test is weakly applicable, and a state in which the test is contraindicated, into the machine learning model; a fourth step of outputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the machine learning model, and an evaluation of the disease candidate, treatment candidate, and test candidate; and a fifth step of outputting information for presenting to a user the disease candidate, treatment candidate, and test candidate, which are the output of the fourth step, and the evaluation of the disease candidate, treatment candidate, and test candidate. In the fourth step, the machine learning model outputs, as the evaluation, the likelihood or recommendation level of the disease candidate, treatment candidate, and test candidate, and ...and a machine learning model trained by fine-tuning to output at least one of the test candidates. (Appendix 39) A method executed by a computer having a first processor and a first memory in a medical support system including a computer and a server, the computer outputs at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the server including a second processor and a second memory, the second memory storing a second program that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a disease name, a treatment name, and a test name, and outputs at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, and an evaluation for at least one of the disease candidate, the treatment candidate, and the test candidate, the first processor including a first step of accepting input of case information from a user, and inputting the case information accepted in the first step and at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or a seventh step of inputting into a second program at least one of a treatment name related to the case information, and a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of a test name related to the case information, and a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name; a second step of outputting from the second program at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the second program, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is the output of the second step, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and outputting, as the evaluation, the certainty or the degree of recommendation of the at least one of the disease candidate, the treatment candidate, and the test candidate in the second step. (Appendix 40) A method executed by a computer having a first processor and a first memory in a medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, wherein the computer outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and the server comprises a second processor and a second memory, and the second memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and the first processor includes a first step of accepting input of case information from a user, and a second step of accepting the case information accepted in the first step and at least one of the case information, which are text or vector expressions, of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for ruling out the diagnosis of the disease. a seventh step of inputting into a machine learning model one or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of a test name related to the case information, and a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name; a second step of outputting from the machine learning model at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation inputted to the machine learning model, and an evaluation of at least one of the disease candidate, treatment candidate, and test candidate; and a third step of outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate outputted from the second step, and an evaluation of at least one of the disease candidate, treatment candidate, and test candidate, (Appendix 41) A method executed by a computer having a first processor and a first memory in a medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, wherein the computer outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and the server comprises a second processor and a second memory, and the second memory stores a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and the first processor includes a first step of accepting input of case information from a user, and a text or vector expression that is a premise for outputting an evaluation for at least one of the list of disease names, a list of treatment names, and a list of test names, the disease names related to the disease names, findings for diagnosing the diseases, and The method includes an eighth step of inputting at least one of findings for excluding a diagnosis of a disease, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, and a strongly applicable state, a weakly applicable state, and a contraindication state for the test related to the test name, into a machine learning model; a fourth step of outputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the machine learning model, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a fifth step of outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate that is output from the fourth step, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and in the fourth step, the certainty or recommendation level of the at least one of the disease candidate, the treatment candidate, and the test candidate is output as the evaluation, and the machine learning model isA method in which a machine learning model is trained by fine tuning to output at least one of a disease candidate, a treatment candidate, and a test candidate for input case information and at least one of a list of disease names, a list of treatment names, and a list of test names. (Appendix 42) A system that outputs at least one of disease candidates, treatment candidates, and test candidates based on case information and has a memory, wherein the memory stores a second program that receives as input case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates, and includes a means for accepting input of case information from a user, and case information accepted by the means for accepting input of case information, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or at least one of a treatment name, a strong applicable state, a weak applicable state, and a contraindicated state related to the treatment name, which are related to the case information, or a test name related to the case information, and a strong applicable state for the test related to the test name, a means for inputting at least one of an applicable condition, a weakly applicable condition, and a contraindicated condition into a second program; a means for causing the second program to output at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input into the second program, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate; and a means for outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate, which is output from the means for outputting the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, wherein the means for outputting the evaluation outputs as the evaluation the at least one of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation level of the at least one of the disease candidate, the treatment candidate, or the test candidate. (Appendix 43) A system that outputs at least one of disease candidates, treatment candidates, and test candidates based on case information and has a memory, wherein the memory stores a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and further includes a means for accepting input of case information from a user, and case information accepted by the means for accepting input of case information, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or a test name related to the case information, a means for inputting at least one of a strong applicable state, a weak applicable state, and a contraindicated state into a machine learning model; a means for causing the machine learning model to output at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and a text or vector representation input into the machine learning model, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate; and a means for outputting information for presenting to a user the at least one of the disease candidate, treatment candidate, and test candidate that is output from the means for outputting the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, and the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, wherein the means for outputting the evaluation outputs as an evaluation the likelihood or recommendation level of the at least one of the disease candidate, treatment candidate, or test candidate. (Appendix 44) A system that outputs at least one of disease candidates, treatment candidates, and test candidates based on case information and has a memory, wherein the memory stores case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation of at least one of the disease candidates, treatment candidates, and test candidates, and a machine learning model that receives as input case information from a user, and the case information received by the means for receiving input of case information and the text or vector expression that is a premise for outputting an evaluation of at least one of the list of disease names, the list of treatment names, and the list of test names, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, or at least one of a treatment name, a strong applicable state, a weak applicable state, and a contraindicated state related to the treatment name, or a test name related to the case information, and a strong applicable state for the test related to the test name. and a means for outputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation inputted to the machine learning model, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and a means for outputting information for presenting the at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate to a user, wherein the means for outputting the evaluation outputs, as the evaluation, the at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation degree of the at least one of the disease candidate, the treatment candidate, or the test candidate, and the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate, and at least one of the disease candidate, the treatment candidate, and the test candidate for the inputted case information, and at least one of the disease name list, the treatment name list, and the test name list. (Appendix 45) A medical support system comprising a computer and a server, which outputs at least one of disease candidates, treatment candidates, and test candidates based on case information, wherein the computer outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and the server comprises a second memory, and the second memory stores a second program that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name, treatment name, and test name, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and the computer has a means for accepting input of case information from a user, and receives the case information accepted by the means for accepting input of case information, and at least one of the disease name, disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector expressions, or the treatment name, strong applicable condition, weak applicable condition, and contraindicated condition related to the treatment name, which are text or vector expressions, a means for inputting at least one of the conditions of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation inputted to the second program, and an evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, which is output from the means for outputting the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, and the evaluation of the at least one of the disease candidate, treatment candidate, and test candidate, to a user, wherein the means for outputting the evaluation outputs, as the evaluation, a likelihood or a degree of recommendation of the at least one of the disease candidate, treatment candidate, and test candidate, and the at least one of the disease candidate, treatment candidate, or test candidate. (Appendix 46) A medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, wherein the computer outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and the server comprises a second memory, and the second memory stores a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and the computer has a means for accepting input of case information from a user, and receives the case information accepted by the means for accepting input of case information, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, which are text or vector expressions, or a treatment name, a state for which a treatment name is strongly applicable, a state for which a treatment name is weakly applicable, and a contraindication related to the treatment name, which are text or vector expressions, a means for inputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector representation input to the machine learning model, and an evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate from the machine learning model; and a means for outputting information for presenting to a user the at least one of the disease candidate, the treatment candidate, and the test candidate, which is output from the means for outputting the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of the at least one of the disease candidate, the treatment candidate, and the test candidate, wherein the means for outputting the evaluation outputs, as the evaluation, a likelihood or a degree of recommendation of the at least one of the disease candidate, the treatment candidate, and the test candidate. (Appendix 47) A medical support system comprising a computer and a server, and outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, wherein the computer outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and the server comprises a second memory, and the second memory stores a machine learning model that receives as input case information and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name list, treatment name list, and test name list, and outputs at least one of the disease candidates, treatment candidates, and test candidates based on the case information, and an evaluation for at least one of the disease candidates, treatment candidates, and test candidates, and the computer has a means for accepting input of case information from a user, and a text or vector expression that is a premise for outputting an evaluation for at least one of the disease name list, treatment name list, and test name list, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing a disease, and findings for excluding a diagnosis of a disease, or a treatment name related to the case information, a means for inputting at least one of a strong applicable state, a weak applicable state, and a contraindication state for a test related to the case information, or at least one of a test name related to the test name and a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name into a machine learning model; a means for outputting at least one of a disease candidate, a treatment candidate, and a test candidate corresponding to the case information and the text or vector expression inputted into the machine learning model, and an evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; and a means for outputting information for presenting to a user an evaluation of at least one of the disease candidate, treatment candidate, and test candidate, and at least one of the disease candidate, treatment candidate, and test candidate, and the means for outputting the evaluation outputs, as an evaluation, at least one of the disease candidate, treatment candidate, and test candidate, and a likelihood or recommendation level of at least one of the disease candidate, treatment candidate, or test candidate, and the machine learning model outputs at least one of the disease candidate, treatment candidate, and test candidate, andThe system is a machine learning model that is trained by fine tuning to output at least one of the disease candidates, the treatment candidates, and the test candidates. [Explanation of symbols]

[0151] 1... system 10... terminal device 10 20... server 25... memory 26... storage 29... processor 40... external server 201... communication unit 202... memory unit 203... control unit 2021... application program 2022... prompt template data 2023... teacher data 2024... machine learning model 2031... reception control module 2032... transmission control module 2033... generative model generation module 2034... case information input acceptance module 2035... prompt generation module 2036... generative model input / output module 2037... presentation control unit

Claims

1. A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; the memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the processor to: A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the second step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the second step, the program outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate.

2. A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the processor to: A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the second step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the second step, the program outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate.

3. A program for operating a computer having a processor and a memory, the program causing the processor to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the processor to: A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a fourth step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the fourth step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the fourth step, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output as the evaluation; The machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the disease name list, the treatment name list, and the test name list, and at least one of the disease candidate, the treatment candidate, and the test candidate.

4. The case information received in the first step includes a chief complaint or a disease name from a patient related to the case information, and further includes at least one of a current medical history, a medical history, a finding, a medical history, and a test result for the patient; The machine learning model is a machine learning model trained to execute the second step and the third step for an input of the seventh step, The program according to claim 2, wherein in the second step, the evaluation includes outputting, in natural language, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate.

5. In the seventh step, the case information accepted in the first step is The program according to claim 4, which structures expected findings that are important for diagnostic / treatment decisions based on electronic medical record information, and inputs the structured content into the machine learning model.

6. The program further causes the processor to performing a sixth step of searching an external search database for information regarding the case information and obtaining or generating the text or vector representation from the search results; The program according to claim 2 , further comprising: inputting, into the machine learning model, in addition to the case information, in the seventh step, the text or the vector expression that can be converted into the text, which is a prerequisite for outputting the evaluation generated in the sixth step.

7. 7. The program according to claim 6, wherein in the sixth step, a knowledge graph database is searched as the external search database.

8. 5. The program of claim 4, wherein in the seventh step, the text or vector expression serving as a premise for outputting the evaluation includes at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or a treatment name related to the case information, and at least one of a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or a test name related to the case information, and at least one of a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the test related to the test name.

9. The program of claim 6, wherein in the sixth step, the text or vector representation includes the text or vector representation of at least one of a description of a disease, findings for diagnosing the disease, and findings for excluding a diagnosis of the disease.

10. The program of claim 9, wherein in the sixth step, the findings for diagnosing the disease include at least one of the following: a finding that increases the probability of a diagnosis of the disease when a positive judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when a positive judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when an abnormal positive judgment is made for the disease, and a definitive diagnosis finding when the diagnosis of the disease is 100%.

11. The program of claim 9, wherein in the sixth step, the findings for excluding a diagnosis of the disease include at least one of the following: a finding that increases the probability of a diagnosis of the disease when a negative judgment is made for the disease, a finding that decreases the probability of a diagnosis of the disease when a negative judgment is made for the disease, and a finding that is useful for the exclusion when a diagnosis of the disease is ruled out 100%.

12. The program of claim 9, wherein in the sixth step, the text or vector representation of the list of treatment names includes at least one of a strong indication, a weak indication, and a contraindication for the treatment associated with the treatment name.

13. The program according to claim 9 , wherein in the sixth step, the text or vector representation of the list of test names includes at least one of a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name.

14. The program of claim 2 , wherein the evaluation presented in the third step includes an evaluation of likelihood for the disease candidate as an output of the machine learning model.

15. The program according to claim 6 , wherein the text or the vector expression that is the premise for outputting the evaluation input in the seventh step includes an output display example of an evaluation of likelihood or degree of recommendation.

16. The program described in claim 6, wherein the text or the vector expression that is the premise for outputting the evaluation input in the seventh step contains states corresponding to each stage of the evaluation of likelihood or recommendation level.

17. 7. The program according to claim 6, wherein the text or the vector expression that is the premise for outputting the evaluation input in the seventh step includes an output display example of an explanation of likelihood.

18. The method of claim 2 , wherein the evaluation presented in the third step includes the evaluation of a recommendation for the treatment or test candidate that is an output of the machine learning model.

19. 7. The program according to claim 6, wherein the text or the vector expression that is the basis for outputting the evaluation input in the seventh step includes a display example of the recommended evaluation.

20. 7. The program according to claim 6, wherein the text or the vector expression that is the basis for outputting the evaluation input in the seventh step describes a state that corresponds to each stage of the recommended evaluation.

21. The machine learning model is a machine learning model trained to create pseudo case information from the case information in the electronic medical record, The program according to claim 4 , wherein the pseudo-case information is created by generating structured data that matches findings and outcomes from patient chart information, and inputting the structured data into the machine learning model.

22. The program of claim 4 , wherein the machine learning model is trained to output one of an evaluation of the likelihood or level of recommendation for the disease candidate, or an evaluation of a recommendation for a treatment candidate or test candidate or an explanation of the level of recommendation, or an explanation of the level of recommendation.

23. In the third step, The program according to claim 2, which outputs information for presenting to the user any one of the disease candidates, treatment candidates, or test candidates, and an evaluation of any one of the disease candidates, treatment candidates, or test candidates, including its likelihood or degree of recommendation.

24. In the third step, The program according to claim 23, which outputs information for presenting to the user one of the disease candidates, treatment candidates, or test candidates, and an evaluation of the one of the disease candidates, treatment candidates, or test candidates, in a graded format, indicating the likelihood or degree of recommendation of the one of the disease candidates, treatment candidates, or test candidates.

25. In the third step, The program described in claim 24, which outputs information for presenting to the user the likelihood or degree of recommendation of any one of the disease candidates, treatment candidates, or test candidates as an evaluation of the disease candidates, treatment candidates, or test candidates in a graded manner, and outputs information for presenting to the user the basis for the graded expression.

26. In the seventh step, a text or vector expression serving as a premise for outputting the evaluation is input, and a state corresponding to each stage of the evaluation of the likelihood or the degree of recommendation is described; In the third step, The program described in claim 25, which outputs information for presenting to the user the likelihood or degree of recommendation of any one of the disease candidates, treatment candidates, or test candidates as an evaluation of the disease candidates, treatment candidates, or test candidates in a graded manner, and outputs information for presenting to the user the basis for the graded expression.

27. A medical support system including a computer and a server, the system outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the system comprising: a program for operating a computer including a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on the case information; the server includes a second processor and a second memory, and the second memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the first processor to A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the second step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the second step, the program outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate.

28. A medical support system including a computer and a server, the system outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the system comprising: a program for operating a computer including a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on the case information; the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the first processor to A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the second step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the second step, the program outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate.

29. A medical support system including a computer and a server, the system outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the system comprising: a program for operating a computer including a first processor and a first memory, the program causing the first processor to output at least one of disease candidates, treatment candidates, and test candidates based on the case information; the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The program causes the first processor to A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a fourth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the fourth step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the fourth step, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output as the evaluation; The machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the disease name list, the treatment name list, and the test name list, and at least one of the disease candidate, the treatment candidate, and the test candidate.

30. An information processing device including a processor and a memory, the information processing device being configured to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; the memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidate, the treatment candidate, and the test candidate corresponding to the case information input to the second program and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of the disease name, the treatment name, and the test name, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the information processing device outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

31. An information processing device including a processor and a memory, the information processing device being configured to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting, from the machine learning model, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information input to the machine learning model and a text or vector expression that is a premise for outputting an evaluation of at least one of the disease name list, the treatment name list, and the test name list, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the information processing device outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

32. An information processing device including a processor and a memory, the information processing device being configured to output at least one of a disease candidate, a treatment candidate, and a test candidate based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a fourth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the fourth step, and information for presenting to the user the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; In the fourth step, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output; The information processing device, wherein the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the disease name list, the treatment name list, and the test name list, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

33. A medical support system including an information processing device and a server, the system outputting a disease candidate, a treatment candidate, or a test candidate based on case information, the information processing device including a first processor and a first memory, the information processing device outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second processor and a second memory, and the second memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the information processing device outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

34. A medical support system including an information processing device and a server, the system outputting a disease candidate, a treatment candidate, or a test candidate based on case information, the information processing device including a first processor and a first memory, the information processing device outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the information processing device outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

35. A medical support system including an information processing device and a server, the system outputting a disease candidate, a treatment candidate, or a test candidate based on case information, the information processing device including a first processor and a first memory, the information processing device outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a fourth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output of the fourth step, and information for presenting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate to the user; In the fourth step, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output as the evaluation; The information processing device, wherein the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the disease name list, the treatment name list, and the test name list, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

36. 1. A method implemented by a computer having a processor and a memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on case information; the memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the method outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation for at least one of the disease candidate, the treatment candidate, or the test candidate.

37. 1. A method implemented by a computer having a processor and a memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the method outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation for at least one of the disease candidate, the treatment candidate, or the test candidate.

38. 1. A method implemented by a computer having a processor and a memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on case information; The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The processor, A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a fourth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the fourth step, and information for presenting to the user the evaluation of the disease candidates, the treatment candidates, and the test candidates; In the fourth step, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output as the evaluation; A method in which the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the list of disease names, the list of treatment names, and the list of test names, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

39. A method for a medical support system including a computer and a server, the method including outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the method being executed by the computer including a first processor and a first memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on the case information, the server includes a second processor and a second memory, and the second memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the second program, the case information being the text or vector representation, at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable condition, a weakly applicable condition, and a contraindicated condition for the examination related to the examination name; a second step of outputting, from the second program, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the method outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation for at least one of the disease candidate, the treatment candidate, or the test candidate.

40. A method for a medical support system including a computer and a server, the method including outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the method being executed by the computer including a first processor and a first memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on the case information, the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; a seventh step of inputting the case information accepted in the first step into the machine learning model, the case information being the text or vector representation, which are at least one of a disease name related to the case information, a disease description related to the disease name, a finding for diagnosing the disease, and a finding for excluding the diagnosis of the disease, or at least one of a treatment name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state related to the treatment name, or at least one of a test name related to the case information, a strong applicable state, a weak applicable state, and a contraindication state for the test related to the test name; a second step of outputting from the machine learning model at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a third step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the second step, and information for presenting the evaluation of the disease candidates, the treatment candidates, and the test candidates to the user; In the second step, the method outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or recommendation for at least one of the disease candidate, the treatment candidate, or the test candidate.

41. A method for a medical support system including a computer and a server, the method including outputting at least one of disease candidates, treatment candidates, and test candidates based on case information, the method being executed by the computer including a first processor and a first memory, the computer outputting at least one of disease candidates, treatment candidates, and test candidates based on the case information, the server includes a second processor and a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The first processor, A first step of accepting input of the case information from a user; an eighth step of inputting the case information accepted in the first step into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a fourth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; a fifth step of outputting at least one of the disease candidates, the treatment candidates, and the test candidates, which are the output of the fourth step, and information for presenting to the user the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; In the fourth step, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate are output as the evaluation; A method in which the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the list of disease names, the list of treatment names, and the list of test names, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

42. 1. A system for outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the system comprising: the memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information into the second program, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector representations, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of a test name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name; a means for causing the second program to output at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; and a means for outputting information for presenting to the user at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output from the means for outputting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; The system wherein the means for outputting the evaluation outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

43. 1. A system for outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the system comprising: The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector representations, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state for the examination related to the examination name, a means for outputting, from the machine learning model, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; and a means for outputting information for presenting to the user at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output from the means for outputting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; The system wherein the means for outputting the evaluation outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

44. 1. A system for outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the system comprising: The memory stores a machine learning model that receives as input the case information and a text or vector expression that is a premise for outputting an evaluation of at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or at least one of the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or at least one of the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a means for outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; and means for outputting information for presenting to the user at least one of the disease candidates, the treatment candidates, and the test candidates, and the evaluation of the disease candidates, the treatment candidates, and the test candidates; The means for outputting the evaluation outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate; The system, wherein the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the list of disease names, the list of treatment names, and the list of test names, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

45. A medical support system comprising a computer and a server, the system outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the computer outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second memory, and the second memory stores a second program that receives the case information and a text or vector expression that is a prerequisite for outputting an evaluation of at least one of a disease name, a treatment name, and a test name, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; The computer includes: means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information into the second program, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector representations, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of a test name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state for the test related to the test name; a means for causing the second program to output at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the second program, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; and a means for outputting information for presenting to the user at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output from the means for outputting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; The system wherein the means for outputting the evaluation outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

46. A medical support system comprising a computer and a server, the system outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the computer outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The computer includes: means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information, and at least one of a disease name related to the case information, a disease description related to the disease name, findings for diagnosing the disease, and findings for excluding the diagnosis of the disease, which are text or vector representations, or at least one of a treatment name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state related to the treatment name, or at least one of an examination name related to the case information, a strongly applicable state, a weakly applicable state, and a contraindicated state for the examination related to the examination name, a means for outputting, from the machine learning model, at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of the at least one of the disease candidates, the treatment candidates, and the test candidates; and a means for outputting information for presenting to the user at least one of the disease candidate, the treatment candidate, and the test candidate, which is an output from the means for outputting the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate, and the evaluation of at least one of the disease candidate, the treatment candidate, and the test candidate; The system wherein the means for outputting the evaluation outputs as the evaluation at least one of the disease candidate, the treatment candidate, and the test candidate, and the likelihood or recommendation degree for at least one of the disease candidate, the treatment candidate, or the test candidate.

47. A medical support system comprising a computer and a server, the system outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on case information, the computer outputting at least one of a disease candidate, a treatment candidate, and a test candidate based on the case information, the server includes a second memory, and the second memory stores a machine learning model that receives the case information and a text or vector expression that is a premise for outputting an evaluation for at least one of a list of disease names, a list of treatment names, and a list of test names, and outputs at least one of the disease candidates, the treatment candidates, and the test candidates based on the case information, and the evaluation for at least one of the disease candidates, the treatment candidates, and the test candidates; The computer includes: means for accepting input of the case information from a user; a means for inputting the case information accepted by the means for accepting input of the case information into the machine learning model, the case information being a text or vector expression on which an evaluation of at least one of the disease name list, the treatment name list, and the test name list is output, the text or vector expression being a premise for outputting an evaluation of at least one of the disease name related to the case information, the disease description related to the disease name, the findings for diagnosing the disease, and the findings for excluding the diagnosis of the disease, or at least one of the treatment name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state related to the treatment name, or at least one of the test name related to the case information, the strong applicable state, the weak applicable state, and the contraindication state for the test related to the test name; a means for outputting at least one of the disease candidates, the treatment candidates, and the test candidates corresponding to the case information and the text or vector representation input to the machine learning model, and the evaluation of at least one of the disease candidates, the treatment candidates, and the test candidates; and means for outputting information for presenting to the user at least one of the disease candidates, the treatment candidates, and the test candidates, and the evaluation of the disease candidates, the treatment candidates, and the test candidates; The means for outputting the evaluation outputs, as the evaluation, at least one of the disease candidate, the treatment candidate, and the test candidate, and a likelihood or a recommendation level for at least one of the disease candidate, the treatment candidate, or the test candidate; The system, wherein the machine learning model is a machine learning model that has been trained by fine tuning to output at least one of the disease candidate, the treatment candidate, and the test candidate for the input case information and at least one of the list of disease names, the list of treatment names, and the list of test names, and to output at least one of the disease candidate, the treatment candidate, and the test candidate.

Images