Processing device, processing program, and processing method

Abstract

Processing device, processing program, and method capable of generating medical documents more efficiently A processing method is provided. [Solution] A type to be created is determined from medical documents classified based on multiple types, a first screen is output containing first medical information, which is medical information obtained based on the determined type from multiple medical information associated with one or more subjects, editing instructions are input from the user for the medical information output on the first screen, and processing is performed to output a second screen containing a medical document generated based on at least the determined type and second medical information, which is medical information edited based on the instruction input.

Description

[Technical Field]

[0001] The present disclosure relates to a processing device, a processing program, and a processing method capable of generating medical documents. [Background technology]

[0002] Systems for improving the work efficiency of doctors have been considered. Patent Document 1 describes an electronic medical record system that includes a first processing unit that displays information about a plurality of items that are registered in advance in association with vaccinations on an input support screen, and a second processing unit that records information about one or more of the vaccination items displayed on the input support screen in an electronic medical record based on a user operation on the input support screen. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2023-54254 Summary of the Invention [Problem to be solved by the invention]

[0004] In light of the above-described techniques, an object of the present disclosure is to provide a processing device, a processing program, and a processing method that are capable of generating medical documents more efficiently. [Means for solving the problem]

[0005] According to one aspect of the present disclosure, there is provided a processing device having at least one processor, wherein the at least one processor is configured to execute processing to determine a type to be created from medical documents classified based on a plurality of types, output a first screen including first medical information which is medical information obtained based on the determined type from a plurality of medical information associated with one or more subjects, accept editing instruction input from a user for the medical information output on the first screen, and output a second screen including a medical document generated based on at least the determined type and second medical information which is medical information edited based on the instruction input.

[0006] According to one aspect of the present disclosure, there is provided a processing program for causing a computer having at least one processor to function as follows: determine a type to be created from medical documents classified based on a plurality of types; output a first screen including first medical information, which is medical information obtained based on the determined type from a plurality of medical information associated with one or more subjects; accept editing instructions from a user for the medical information output on the first screen; and output a second screen including a medical document generated based on at least the determined type and second medical information, which is medical information edited based on the instruction input.

[0007] According to one aspect of the present disclosure, there is provided a processing method executed by at least one processor in a computer having the at least one processor, the processing method including the steps of: determining a type to be created from medical documents classified based on a plurality of types; outputting a first screen including first medical information, which is medical information obtained based on the determined type from a plurality of medical information associated with one or more subjects; accepting editing instruction input from a user for the medical information output on the first screen; and outputting a second screen including a medical document generated based on at least the determined type and second medical information, which is medical information edited based on the instruction input. [Effects of the Invention]

[0008] According to the present disclosure, it is possible to provide a processing device, a processing program, and a processing method that can generate medical documents more efficiently.

[0009] It should be noted that the above effects are merely illustrative for the sake of convenience and are not limiting. In addition to or instead of the above effects, any effect described in this disclosure or an effect obvious to a person skilled in the art may be achieved. [Brief explanation of the drawings]

[0010] [Figure 1A] FIG. 1A is a diagram showing an overview of processing executed by a processing system 1 according to an embodiment of the present disclosure. [Figure 1B] FIG. 1B is a block diagram showing the configuration of a processing system 1 according to an embodiment of the present disclosure. [Figure 2] FIG. 2 is a block diagram showing a configuration of the processing device 100 according to an embodiment of the present disclosure. [Figure 3A] FIG. 3A is a diagram conceptually illustrating a subject management table stored in the information providing device 200 according to an embodiment of the present disclosure. [Figure 3B]FIG. 3B is a diagram conceptually illustrating a template management table stored in the processing device 100 according to an embodiment of the present disclosure. [Figure 3C] FIG. 3C is a diagram illustrating an example of a user attribute management table stored in the processing device 100 according to an embodiment of the present disclosure. [Figure 4A] FIG. 4A is a diagram showing a processing flow executed in the processing device 100 according to an embodiment of the present disclosure. [Figure 4B] FIG. 4B is a diagram showing a processing flow executed in the processing device 100 according to an embodiment of the present disclosure. [Figure 4C] FIG. 4C is a diagram showing a processing flow executed in the processing device 100 according to an embodiment of the present disclosure. [Figure 5] FIG. 5 is a diagram illustrating an example of medical information according to an embodiment of the present disclosure. [Figure 6] FIG. 6 is a diagram showing an example of template information processed by the processing device 100 according to an embodiment of the present disclosure. [Figure 7] FIG. 7 is a diagram showing an example of query information processed by the processing device 100 according to an embodiment of the present disclosure. [Figure 8] FIG. 8 is a diagram showing an example of a medical document output by the processing device 100 according to an embodiment of the present disclosure. [Figure 9A] FIG. 9A is a diagram showing an example of a screen output from the processing device 100 according to an embodiment of the present disclosure. [Figure 9B] FIG. 9B is a diagram showing an example of a screen output from the processing device 100 according to an embodiment of the present disclosure. [Figure 9C] FIG. 9C is a diagram showing an example of a screen output from the processing device 100 according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0011] 1. Overview of Processing System 1 1A is a diagram illustrating an overview of processing executed by a processing system 1 according to an embodiment of the present disclosure. According to FIG. 1A, the processing system 1 according to the present disclosure is used to generate a medical document based on medical information acquired from an information providing device such as a medical device, an interview device, an electronic medical record device, a PACS (Picture Archiving and Communication Systems), a medical staff terminal device usable by a medical staff member, a subject terminal device usable by a subject, or a database device in which information input to these devices is recorded.

[0012] The processing system 1 acquires various information such as a subject's attribute information, hospital visit information, medical interview information, findings information, diagnosis information, or discharge information from at least one of a medical device, an interview device, an electronic medical record device, a PACS, a medical staff terminal device, a subject terminal device, and a database device that function as an information providing device. Then, the processing system 1 extracts medical information (first medical information) necessary to generate a desired medical document from the acquired information. This extraction process is performed, for example, using a first trained model. Once the medical information is extracted, a medical document is generated using the extracted medical information. This generation process is performed, for example, using a second trained model.

[0013] That is, the processing system 1 can perform processes related to the extraction of medical information required for generating medical documents and processes related to the generation of medical documents using the extracted medical information. For example, medical professionals need to input various information and write sentences when generating medical documents, but by using the processing system 1, medical documents can be generated more efficiently.

[0014] The processing system 1 according to the present disclosure is typically used in medical institutions. Examples of such facilities include medical institutions, welfare facility management organizations, nursing care facility management organizations, various school management organizations, and public facility management organizations, preferably medical institutions. Such medical institutions are required to create various medical documents, such as letters of referral to other medical institutions, permission letters for attendance at school or work, reports reporting the occurrence of a specific disease for a specified period of time for a national or local government, or reports reporting the progress of medical treatment within the institution. Therefore, by using the processing system 1, it is possible to efficiently output such medical documents.

[0015] In this disclosure, a "medical document" may be any document generated using medical information. Therefore, a medical document is not limited to documents generated or used only by medical institutions. Furthermore, a medical document is not limited to documents generated by instructions input by medical professionals, but may also be created by other parties, such as the subject or the subject's relatives. Examples of such medical documents include letters of referral to other medical institutions, permission letters to attend school or work, reports from national or local governments reporting the onset of a specific disease for a specified period of time, or reports reporting the progress of medical treatment within an institution, as well as various other documents such as discharge summaries recording the progress of hospitalization, nursing summaries recording the subject's condition and nursing care at the medical institution, medical fee statements (so-called receipts), detailed descriptions of symptoms related to medical fees, records of surgeries performed on the subject at the medical institution, opinions to be submitted to life insurance management institutions, opinions from the attending physician, visiting nursing instructions, hospital visit reports, special instructions, death certificates, or autopsy reports.

[0016] Furthermore, in the present disclosure, "medical information" may be any information acquired by a medical institution or used by a medical institution. Therefore, it is not limited to information acquired only by a medical institution and used only by a medical institution. Examples of such medical information include attribute information indicating the attributes of the subject, interview information input by the subject or a medical professional, etc., findings information input by a medical professional, etc., diagnosis information indicating the results of a diagnosis by a medical professional, etc., or treatment information indicating the contents of treatment performed by a medical professional, etc. More specifically, examples of medical information include: - Information indicating the subject's attributes, such as the subject's name, date of birth, age, gender, family structure, occupation, means of transportation, address, and contact details; -Various information recorded in medical records Information obtained from various medical devices such as CT devices, MRI devices, X-ray devices, electrocardiogram devices, ultrasound diagnostic devices, endoscope devices, blood pressure monitor devices, or angiography devices; History information showing medical treatment history at medical institutions Chief complaint, medical history, allergy history, medication history, name / site / severity of disease, test details and test findings (tests using the above medical devices, blood sampling, physiological tests, and the dates of these tests), treatment details (surgical procedures, medication administration, and the dates of these treatments), prescription at time of discharge, patient condition (patient's level of independence), degree of dementia, paralysis, limb loss, means of transportation, need for medical management (need for medical treatment, care, and nursing-related services), outcome at time of discharge, medical devices used, rehabilitation details, route of admission (emergency transport or referral), attached documents (test results, etc.), date of hospital visit, date of discharge, and diagnostic information such as progress Examples include:

[0017] Furthermore, in the present disclosure, a "subject" may be any person who is the subject of a medical document generated using the processing system 1, and is not limited to only those with a specific attribute. Such subjects include all types of persons, such as patients, examinees, diagnostic subjects, and healthy individuals. Furthermore, a "user" may be any person who uses the processing device 100 of the processing system 1, and is not limited to only those with a specific attribute. Such users include all types of persons, such as those belonging to the institutions exemplified above, such as medical institution staff (doctors, nurses, laboratory technicians, medical administrators, medical institution staff and managers, emergency personnel, and medical professionals such as those in charge of medical devices and medical equipment), as well as the subject himself / herself.

[0018] FIG. 1B is a block diagram showing a configuration of a processing system 1 according to an embodiment of the present disclosure. According to FIG. 1B, the processing system 1 includes at least a processing device 100 for outputting a medical document based on medical information and an information providing device 200 for providing the medical information to the processing device 100. The processing system 1 may also include, as necessary, a medical device for acquiring measurement data, a model generation device for generating various trained models used to generate the medical document, an output device (e.g., a display device, a printer device, or a terminal device at another institution) to which the medical document is output, or an input device for accepting user instructions. In the processing system 1, various devices, including the processing device 100 and the information providing device 200, are connected to each other so as to be able to communicate with each other via at least one of a wireless network and a wired network.

[0019] 2. Configuration of the Processing Device 100 FIG. 2 is a block diagram showing a configuration of a processing device 100 according to an embodiment of the present disclosure. According to FIG. 2, the processing device 100 includes a processor 111, a memory 112, an input interface 113, an output interface 114, and a communication interface 115. These components are electrically connected to one another via control lines and data lines. It should be noted that the processing device 100 does not need to include all of the components shown in FIG. 2; some components may be omitted, or other components may be added. For example, the processing device 100 may include a battery or the like for driving each component.

[0020] Such a processing device 100 can be suitably applied to any device capable of executing the processes according to the present disclosure, such as a laptop computer, a desktop computer, a smartphone, a tablet terminal, an on-premise server device, or a cloud server device. Furthermore, the processing device 100 can also perform at least a portion of the processes on one or more server devices on a cloud or one or more server devices operated within a specific institution. Therefore, the processing device 100 includes a combination of a laptop computer, a desktop computer, a smartphone, a tablet terminal, an on-premise server device, and a cloud server device.

[0021] In the processing device 100, the processor 111 functions as a control unit that controls the processing device 100 or other components of the processing system 1 based on a program stored in the memory 112. Specifically, the processor 111 executes, based on the program stored in the memory 112, the following processes: "determining a type to be created from medical documents classified based on a plurality of types," "outputting a first screen including first medical information, which is medical information acquired based on the determined type from a plurality of pieces of medical information associated with one or more subjects," "accepting an editing instruction input from a user for the medical information output on the first screen," and "outputting a second screen including a medical document generated based on at least the determined type and second medical information, which is medical information edited based on the instruction input." The processor 111 is mainly composed of one or more CPUs, but may also be combined with a GPU, an FPGA, or the like as appropriate.

[0022] The memory 112 is composed of RAM, ROM, nonvolatile memory, HDD, SSD, etc., and functions as a storage unit. The memory 112 stores instructions and commands for various controls of the processing system 1 according to this embodiment as programs. Specifically, the memory 112 stores programs that the processor 111 executes, such as "a process of determining a type to be created from medical documents classified based on multiple types," "a process of outputting a first screen including first medical information, which is medical information acquired based on the determined type from multiple pieces of medical information associated with one or more subjects," "a process of accepting an editing instruction input from a user for the medical information output on the first screen," and "a process of outputting a second screen including a medical document generated based on at least the determined type and second medical information, which is medical information edited based on the instruction input." In addition to these programs, the memory 112 may also store various information stored in the subject management table, template management table, and user attribute management table (FIGS. 3A to 3C). It should be noted that this information does not need to be stored in the memory 112 installed inside the processing device 100 at all times, but may be stored in a database device connected to the outside via a wireless or wired network. In such a case, therefore, the database device may also be included in the memory 112.

[0023] The input interface 113 functions as an input unit that accepts user input instructions to the processing device 100. Examples of the input interface 113 include various hard keys such as a keyboard and a mouse, and a touch panel that is superimposed on the display of a display device and has an input coordinate system that corresponds to the display coordinate system of the display. In the case of a touch panel, icons corresponding to commands to be input are displayed on the display, and the user selects or specifies each icon by inputting instructions via the touch panel. The method for detecting user input instructions via the touch panel may be any method, such as a capacitive method or a resistive film method. The input interface 113 does not always need to be physically provided in the processing device 100, and may be connected as needed via a wired or wireless network.

[0024] The output interface 114 functions as an output unit for outputting medical documents generated by the processing device 100. An example of the output interface 114 is an interface for connecting to an external device or equipment, such as a display device configured with a liquid crystal panel, an organic EL display, a plasma display, or the like. However, if the processing device 100 itself has a display, the display can function as the output interface. Furthermore, if the processing device 100 is connected to a display device or the like via a communication interface 115, the communication interface 115 can also function as the output interface 114.

[0025] The communication interface 115 functions as a communication unit for transmitting and receiving the above-mentioned medical information and the like to and from the information providing device 200 via a wired or wireless communication network. Examples of the communication interface 115 include various types, such as a connector for wired communication such as USB or SCSI, a transmitting and receiving device for wireless communication such as LTE, wireless LAN, Bluetooth (registered trademark) or infrared, and various connection terminals for printed circuit boards or flexible circuit boards.

[0026] Although the specific configuration of the information providing device 200 will not be described, it has at least a processor, a memory, and a communication interface, and, as necessary, an input interface and an output interface, etc. Examples of such information providing devices 200 include various medical devices such as CT devices, MRI devices, X-ray devices, electrocardiograph devices, ultrasound diagnostic devices, endoscopic devices, sphygmomanometer devices, and angiography devices, control devices for controlling these medical devices, interview devices, electronic medical record devices, PACS, medical staff terminal devices, subject terminal devices, database devices in which information input to these devices is recorded, or combinations of these. The information providing device 200 reads out necessary information in response to a request received from the processing device 100, for example, and provides the information to the processing device 100.

[0027] 3. Information stored in each management table 3A is a diagram conceptually illustrating a subject management table stored in the information providing device 200 according to an embodiment of the present disclosure. The information stored in the subject management table is read from the information providing device 200 in accordance with the progress of processing by the processor 111 of the processing device 100, and is updated and stored in the information providing device 200.

[0028] According to FIG. 3A, the subject management table stores attribute information, medical interview information, findings information, diagnosis information, etc. in association with subject ID information. "Subject ID information" is information unique to each subject for identifying each subject. Subject ID information is generated each time a new subject is registered by the user or the subject himself / herself. "Attribute information" is information indicating the attributes of each subject. The attribute information is input, for example, by the subject or a medical professional. Examples of such attribute information include the subject's name, date of birth, age, sex, family structure, occupation, means of transportation, address, and contact information.

[0029] "Medical interview information" is information input by, for example, a subject or a medical professional via a medical interview device. Examples of such medical interview information include subjective symptoms, medical history, current oral medications, family medical history, presence or absence of allergies and travel destinations, body temperature, presence or absence of underlying diseases, presence or absence of physical findings, vaccination history, and timing of vaccination. This medical interview information is generally input by the subject themselves or by a medical professional asking the subject about it. "Finding information" is information input by a medical professional such as a doctor. Examples of such finding information include information indicating abnormalities obtained through examinations such as digital examination, medical interview, and palpation.

[0030] "Diagnostic information" refers to information that includes at least one of information that assists diagnosis at a medical institution, information indicating the results of a diagnosis, information indicating treatments based on the results of a diagnosis, and information indicating the history of a diagnosis. Examples of such diagnostic information include information obtained from various medical devices such as CT scanners, MRI scanners, X-ray scanners, electrocardiographs, ultrasound diagnostic scanners, endoscopes, blood pressure monitors, and angiographs, the chief complaint, the name of the disease (diagnosis), the location and its severity, the details and findings of tests (tests using the above medical devices, blood sampling, physiological tests, and the dates of these tests), the details of treatment (type and date of surgery, medication administration and duration), prescriptions at the time of discharge, the patient's condition (patient's level of independence), the degree of dementia, paralysis, limb loss, the need for medical management (whether medical treatment, care, and nursing-related services are required), the outcome at the time of discharge, the medical devices used, the details of rehabilitation, the route of admission (emergency transport or referral), the date of admission, the date of discharge, and progress information.

[0031] In the present disclosure, as described above, medical information may include any of attribute information, medical interview information, finding information, and diagnosis information.

[0032] 3B is a diagram illustrating an example of a template management table stored in the processing device 100 according to an embodiment of the present disclosure. Information stored in the template management table is read from the memory 112 in accordance with the progress of processing by the processor 111 of the processing device 100, and is updated and stored at any timing.

[0033] 3B, the template management table stores, for each type of information, template ID information, part ID information, placement information, medical information, etc. "Type information" is information for identifying the type of medical document. Examples of such type information include referral letters to other medical institutions, permission letters permitting attendance at school or work, reports reporting the development of a specific disease for a specified period of time, such as those issued by a national or local government, reports reporting the progress of medical treatment within an institution, discharge summaries recording the progress of hospitalization, nursing summaries recording the patient's condition and nursing care at a medical institution, medical fee statements (so-called receipts), detailed symptom descriptions related to medical fees, records of surgeries performed on a patient at a medical institution, opinions to be submitted to a life insurance management institution, opinions from the attending physician, home nursing instructions, hospital visit reports, special instructions, death certificates, and autopsy reports.

[0034] "Template ID information" is information unique to each template information and is information for identifying each template information. Template ID information is generated each time new template information is created. "Template information" is information used as a template when generating a medical document. An example of such template information is document data of the portion of a medical document excluding the portion containing information unique to each subject, i.e., the portion common to multiple subjects. The data format of such document data may be any format that can be processed by the processing device 100 or a terminal device used by a medical professional.

[0035] "Part ID information" is information unique to each piece of part information and is used to identify each piece of part information. Here, "part information" refers to information about components constituting template information, which serves as a template for a medical document. For example, taking a referral letter as an example, the referral letter is composed of multiple parts, such as a preamble and a concluding part, a section name part, a diagnosis name part, a purpose of referral part, a medical history part, a progress part, and a discharge prescription part. In other words, part information can be, for example, the document data constituting these parts. Furthermore, a section whose content changes depending on the diagnosis information, such as a progress part, may be a combination of multiple more subdivided parts, such as an examination part, a surgery part, and a medication part. The part ID information stores one or more pieces of information for identifying the part information constituting a medical document. For example, if "L1, L2, L3..." is stored as part ID information associated with template ID information "K1," the template information identified by template ID information "K1" is composed of part information identified by L1 to L3, etc. The data format of such document data may be any format that can be processed by the processing device 100 or a terminal device used by a medical professional.

[0036] "Location information" is information that indicates the location in the medical document of part information identified by part ID information. For example, information such as the order in which part information identified by each part ID information is arranged, or coordinates indicating the location in the medical document, is used as location information. Furthermore, this location information is not limited to information indicating the location in the medical document by itself, but may also be information indicating the location in the medical document when combined with diagnostic information.

[0037] In addition, the template management table stores one or more pieces of part information associated with part ID information for each type of medical document. In addition, different template information may be stored by further subdividing each piece of medical information (e.g., diagnosis name). For example, if the type information is "referral letter," it may be further subdivided by diagnosis name (medical information), and different template information may be stored for "disease A" and "disease B" even for referral letters. This makes it possible to use a wider variety of template information.

[0038] "Medical information" includes at least one of the attribute information, interview information, findings information, and diagnosis information shown in Figure 3A. That is, information specifying one or more pieces of medical information required to generate each type of medical document is stored.

[0039] Furthermore, the information stored in the template management table may be stored in the memory 112 of the processing device 100, but may also be stored in other devices such as other processing devices or database devices.

[0040] 3C is a diagram illustrating an example of a user attribute management table stored in the processing device 100 according to an embodiment of the present disclosure. The information stored in the user attribute management table is read from the memory 112 in accordance with the progress of processing by the processor 111 of the processing device 100, and is updated and stored at any timing.

[0041] According to FIG. 3C, the user attribute management table stores user attribute information and type information in association with each other. "User attribute information" is information indicating the attributes of a user. Examples of such user attribute information include medical staff, subjects, and others. The above user attribute information can also be further subdivided and distinguished. In this case, examples of user attributes include doctors, nurses, laboratory technicians, medical administrators, other staff and managers of medical institutions, emergency personnel, personnel in charge of medical devices and medical equipment, subjects, and others.

[0042] "Type information" is information that identifies the type of medical document, similar to the type information shown in FIG. 3B. In FIG. 3C, the type information can include one or more types of type information. That is, one or more types of type information may be stored in association with each attribute information. For example, if type information "J1, J2, J3" is stored as type information in association with user attribute information of "doctor," this indicates that the doctor can select medical documents of the types J1, J2, and J3.

[0043] The information stored in the user attribute management table may be stored in the memory 112 of the processing device 100, or may be stored in another device such as another processing device or database device.

[0044] 3C, a user management table that associates user attribute information with type information and manages it is described as an example, but any table that can manage the type of medical document according to the user is acceptable. For example, it is possible to manage user authority information and type information in association with each other, or to directly associate each user with type information and manage them.

[0045] 3C, the type information associated with the user attribute information may be further subdivided and classified based on the format of the medical document. For example, a type management table may be prepared in which one or more type ID information is associated with type information J1 and one or more type ID information is associated with type information J2. Specifically, for the type of "referral letter" identified by type information J1, the following formats are associated and stored: type ID information indicating "treatment," type ID information indicating "examination," type ID information indicating "rehabilitation," type ID information indicating "follow-up observation," type ID information indicating "examination result report," and progress ID information indicating "progress report."

[0046] In this case, the template management table in Fig. 3B stores various information in association with each type information, but stores various information in association with each format ID information. Also, in this case, the various processes in Fig. 4A and Fig. 4B are each performed based on the type information, but each process is performed based on the format ID information. In this way, a wider variety of medical documents with different formats can be managed.

[0047] Although not specifically shown, in addition to the tables of Figures 3A to 3C, other information may be stored in memory 112, such as a user management table in which information identifying a user, such as each user ID information, is stored in association with user attribute information, etc.

[0048] 4. Processing flow executed by the processing device 100 (1) Processing flow for outputting the first screen containing medical information Fig. 4A is a diagram showing a processing flow executed in the processing device 100 according to an embodiment of the present disclosure. Specifically, Fig. 4A is a diagram showing a series of processing flows from accepting a request from a user to generate a medical document to outputting medical information used to generate the medical document. The processing flow is mainly performed by the processor 111 of the processing device 100 reading and executing a program stored in the memory 112.

[0049] (A) Medical document generation request acceptance process 4A, the processor 111 receives a request for generating a medical document from a user (S111). One example of this process is performed by receiving, via the input interface 113 of the processing device 100, an instruction input from the user to select a subject for whom a medical document is to be generated and an instruction input to execute the generation of the medical document. That is, the generation request includes subject ID information for identifying the selected subject and information indicating a request for the generation of a medical document. Another example of this process is performed by receiving, via the input interface of a user terminal device used by the user, an instruction input from the user to select a subject for whom a medical document is to be generated and an instruction input to execute the generation of the medical document, thereby generating a generation request, and receiving the generation request from the user terminal device via the communication interface 115.

[0050] (B) Medical document type selection process Next, the processor 111 selects the type of medical document to be generated (S112). As an example of this process, the processor 111 receives a user's instruction input regarding the selection of user attribute information via the input interface 113 in accordance with a screen (e.g., FIG. 9A) output via the output interface 114. The processor 111 identifies one or more types of type information associated with the selected user attribute information by referring to the user attribute management table (FIG. 3C). The processor 111 outputs a screen including the identified one or more types of type information via the output interface 114, and receives a user's instruction input regarding the selection of specific type information via the input interface 113. As a result, the processor 111 selects the type identified by the received type information as the type of the medical document to be generated.

[0051] As another example of this process, the processor 111 identifies user attribute information based on pre-input user ID information. The processor 111 refers to the user attribute management table (FIG. 3C) to identify one or more types of information associated with the identified user attribute information. The processor 111 outputs a screen including the identified one or more types of information via the output interface 114, and accepts a user instruction input regarding selection of specific type information via the input interface 113. As a result, the processor 111 selects the type identified by the accepted type information as the type of the medical document to be generated.

[0052] In this way, appropriate types are presented to the user according to the user's attributes, and the user can select the desired type from among them, which allows the user to select the type more efficiently. Note that the process for selecting the type may be performed based on user authority information or information that identifies the user, rather than on user attribute information.

[0053] (C) Processing to read out the subject's medical information Next, the processor 111 reads out the medical information of the subject (S113). As an example of this process, the processor 111 refers to the subject management table and reads out the medical information of the subject based on the subject ID information of S111.

[0054] (D) Extraction and processing of the subject's medical information (D-1) Example of extraction process (part 1) Next, the processor 111 extracts medical information necessary for generating a medical document from the read medical information (S114). As an example of this process, the processor 111 refers to the template management table (FIG. 3B) and identifies medical information associated with the type information identified in S112 based on the type information. The processor 111 then extracts only the identified medical information (i.e., medical information necessary for generating medical information) from the subject's medical information read in S113. For example, the medical information read in S113 includes all of the attribute information, medical interview information, finding information, and diagnosis information (FIG. 3A) associated with the subject ID information. Of these, only admission / discharge information, diagnosis information, medical treatment details, and discharge prescription are necessary for generating a referral letter selected as the type. Therefore, here, only necessary information is extracted from the attribute information, medical interview information, finding information, and diagnosis information. Furthermore, the extracted medical information may be extracted based on information stored as medical interview information, finding information, or diagnosis information. Specifically, information with higher importance or priority may be extracted based on the name of a disease, which is one of the pieces of information stored as medical interview information, finding information, or diagnosis information. For example, if pneumonia is stored as the name of the disease, information on test data or imaging findings that can confirm an inflammatory reaction may be preferentially extracted, or information on antibiotics may be preferentially extracted from information on drug treatment (although saline may be administered during hospitalization, this is of low importance as medical information for pneumonia).

[0055] (D-2) Example of extraction process (part 2) Here, among the attribute information, medical interview information, findings information, and diagnosis information, some information is systematically managed in terms of type and content, while other information is input in free text format, for example. FIG. 5 is a diagram illustrating an example of medical information according to an embodiment of the present disclosure. Specifically, FIG. 5 is a diagram illustrating an example of medical record information 10 stored as one piece of medical information. According to FIG. 5, the medical record information 10 is systematically entered by a doctor in the order of "subjective information," "objective information," "evaluation," and "plan," with the results of medical treatment. Such medical record information 10 is input via an input interface or the like of a medical staff terminal device and stored in association with the subject in a subject management table.

[0056] The medical record information 10 includes information expressed in free text format as "S" (subjective information), such as "The patient has had fever and chills for three days. Medical history includes a right rib fracture, a lumbar compression fracture, high blood pressure, diabetes, and dyslipidemia." The medical record information 10 also includes information expressed in free text format as "O" (objective information), such as "Temperature 38.5°C, oxygen saturation 97%, negative result from influenza rapid diagnostic kit, and abnormal findings from CT scan and blood test." The medical record information 10 also includes information expressed in free text format as "A" (evaluation), such as "Based on the results of the CT scan and blood test, pneumonia was diagnosed." The medical record information 10 also includes information as "P" (plan), such as "ABPC / SBT was administered and hospitalization for follow-up observation was recommended." In other words, the medical record information 10 includes information such as the chief complaint, medical history, name of the disease, test details and test findings, and treatment details.

[0057] Returning to FIG. 4A again, the processor 111 can extract medical information necessary for generating a medical document from the medical record information exemplified in FIG. 5. Such processing is performed, for example, by utilizing a first trained model. The first trained model is a trained model generated by the processor 111 or a processor of a model generation device through machine learning using a predetermined dataset. Specifically, the first trained model uses a combination of pre-prepared training strings (e.g., medical record information similar to the medical record information 10 in FIG. 5) and label information that labels the medical information included in the training strings (e.g., assigning the label "chief complaint" to "fever" and "chills" and assigning the label "hospitalization route" to "emergency transport") to have a learner learn a labeling pattern by machine learning. For example, the machine learning is performed by providing a neural network composed of a combination of neurons with this information combination and repeatedly learning while adjusting the parameters of each neuron so that the output from the neural network is the same as the correct label. Then, the first trained model is acquired through the machine learning.

[0058] The first trained model is not limited to the neural networks exemplified above, but can also be generated using machine learning techniques such as convolutional neural networks, multi-layer Herceptrons (MLP), long short-term memory (LSTM), gated recurrent units (GRUs), graph neural networks (GNNs), and transformers; gradient boosting decision trees (GBDTs) such as LightGBM (Light Gradient Boosting Machine), XGBoost, and CatBoost; ridge regression, logistic regression, support vector regression (SVR), nearest neighbor methods, decision trees, regression trees, and random forests.

[0059] In S114 of FIG. 4A, the processor 111 reads out the first trained model generated as described above from the memory 112 and executes it. Specifically, the processor 111 inputs the subject's medical record information 10 illustrated in FIG. 5 as input information to the first trained model, and acquires medical information for each item as output information (for example, medical information for "fever" and "chills" is acquired for the item "chief complaint", and medical information for "emergency transport" is acquired for the item "hospitalization route"). Then, the processor 111 refers to the template management table (FIG. 3B) and, based on the type information identified in S112, acquires a template corresponding to the type information. The processor 111 identifies the medical information associated with the item. The processor 111 extracts only the identified medical information (i.e., medical information required for generating medical information) from the medical information for each item obtained as output information. For example, the medical information obtained as output information includes various medical information, but in generating the referral letter selected as the type, only admission and discharge information, diagnosis information, medical treatment details, and discharge prescription are required. Therefore, here, only the necessary information is extracted from the various medical information.

[0060] The above description deals with a case in which medical information is extracted using a first trained model obtained by training using a combination of training strings and label information. However, the first trained model is not limited to such trained models. A trained large-scale language model (i.e., a natural language processing model trained using a large amount of text data) can also be used. In this case, various models can be used, such as general-purpose language models such as GPT-3, GPT-4, Claude, PaLM, NeMo LLM, or LLaMA, or medical language models such as GPT-MD and MED-PaLM. However, the trained model is not limited to these, and other models such as BERT, Word2vec, other neural networks, support vector machines, logistic regression, decision trees, random forests, or combinations of these models may also be used. Furthermore, a morphological analysis algorithm using a dictionary of medical information may also be used in combination.

[0061] When a large-scale language model is used as the first trained model, in S114 of FIG. 4A, processor 111 generates query information to be input to the large-scale language model. An example of the query information is, "From the following medical record information, please extract information on admission and discharge, diagnosis, medical treatment details, and discharge prescriptions, which are necessary for generating a referral letter, and classify them by item. (The character string of medical record information 10 is entered after this entry, but is omitted here.)" Such query information may be generated by a user inputting an instruction via input interface 113, or may be generated automatically by processing of processor 111.

[0062] The processor 111 inputs the generated query information into a large-scale language model and extracts medical information required for generating a medical document as output information.

[0063] As described above, since the medical information of a subject contains a large amount of information and is expressed in free text format, it is cumbersome for medical professionals to search for the necessary information each time, but by extracting the medical information necessary for generating a medical document through processing by the processor 111 as in S114, more efficient generation of medical documents becomes possible. Note that, although various methods for the process of extracting the medical information of a subject are exemplified in items (D-1) and (D-2), the process can also be performed by combining these methods.

[0064] (E) Conversion and processing of the subject's medical information Next, the processor 111 executes a process of converting the extracted medical information into a regular expression to correct variations in the notation of each piece of extracted medical information (S115). As an example of this process, the processor 111 converts each piece of extracted medical information by referring to a regular expression dictionary table (not shown) prepared in advance. For example, the regular expression dictionary table pre-stores "ABPC / SBT" as a normalized expression in association with "ABPC / SBT," "ampicillin-sulbactam," and "ampicillin / sulbactam." Therefore, when "ampicillin-sulbactam" is extracted as medical information, the processor 111 refers to the regular expression dictionary table and converts "ampicillin-sulbactam" to "ABPC / SBT." The processor 111 stores the converted medical information in the subject management table in association with subject ID information.

[0065] Note that such conversion processing is not limited to the method using the regular expression dictionary table described above, and various methods such as a method using a trained model can be used. Furthermore, such conversion processing does not need to be performed on all of the extracted medical information, and may be performed on only a portion of the extracted medical information, or may not be performed at all.

[0066] (F) Output processing of the subject's medical information Next, the processor 111 outputs at least one of the medical information extracted in S114 and the medical information converted in S115 as first medical information (S116). As an example of this processing, the processor 111 acquires the medical information of S114 and S115 as the first medical information and generates a first screen including the first medical information. Then, the processor 111 outputs the generated first screen to a display or the like via the output interface 114.

[0067] Although the first screen will be described in detail later, as an example, the first screen is output adjacent to the second screen containing the generated medical document. That is, the first screen is displayed in a manner that allows the user to view it simultaneously with the second screen. Therefore, the user can check the medical document contained in the second screen while referring to the first medical information contained in the first screen, allowing the user to check the medical document more efficiently.

[0068] This completes the processing flow.

[0069] (2) Processing flow for outputting the second screen containing medical documents 4B is a diagram showing a processing flow executed by the processing device 100 according to an embodiment of the present disclosure. Specifically, FIG. 4B is a diagram showing a series of processing flows up to the output of a second screen including a medical document. The processing flow is mainly performed by the processor 111 of the processing device 100 reading and executing a program stored in the memory 112.

[0070] (A) Editing of medical information 4B, with the first screen including the first medical information displayed, the processor 111 determines whether a user's instruction to edit the first medical information has been received via the input interface 113 (S211). If the instruction has been received, the processor 111 edits the corresponding first medical information based on the instruction (S212). As an example of this process, the processor 111 receives an instruction from the user via the input interface 113 to correct the first medical information or input new medical information when the first medical information displayed on the first screen contains an error or when the first medical information does not contain medical information that should have been included. The processor 111 then stores the corrected or input medical information as second medical information in the subject management table in association with the subject ID information, and outputs the corrected or input medical information on the first screen, replacing the unedited medical information. Details of this editing process will be described later. Furthermore, if an instruction input has not been accepted in S211, processor 111 skips the editing process in S212.

[0071] (B) Process for generating template information that will serve as a template for medical documents Next, the processor 111 refers to the template management table in the memory 112 based on the type information of S112 in Fig. 4A, and acquires template information that serves as a model for the desired medical document (S213). As an example of this process, the processor 111 refers to the template management table (Fig. 3C) and identifies one template ID information associated with the type information included in the generation request and the read medical information. Furthermore, the template information may be identified using not only the medical information and type information, but also a clinical pathway determined from the medical information.

[0072] 4C is a diagram showing an example of a processing flow executed in the processing device 100 according to an embodiment of the present disclosure. Specifically, FIG. 4C is a diagram showing a detailed processing flow of the process of acquiring template information in S213 of FIG. 4B. This processing flow is mainly performed by the processor 111 of the processing device 100 reading and executing a program stored in the memory 112.

[0073] According to FIG. 4C, the processor 111 reads at least one of the medical information (first medical information) output in S116 of FIG. 4A and the medical information (second medical information) edited in S212 of FIG. 4B (S311). After the medical information is read, the processor 111 refers to the template management table (FIG. 3C) (S312). At this time, the processor 111 identifies one piece of template ID information from among the multiple pieces of template ID information based on the type information selected in S112 of FIG. 4A and the medical information. Then, the processor 111 identifies one or more pieces of part ID information constituting the template information based on the identified template ID information. After the part ID information is identified, the processor 111 reads out part information associated with the identified part ID information (S313). The processor 111 generates one piece of template information by joining together the read pieces of part information according to the arrangement information associated with the template ID information (S314).

[0074] FIG. 6 is a diagram illustrating an example of template information processed by the processing device 100 according to an embodiment of the present disclosure. Specifically, it is a diagram illustrating an example of template information generated according to the processing of FIG. 4C. FIG. 6 illustrates template information 20 generated from a plurality of pieces of part ID information and arrangement information associated with template ID information identified by type information (e.g., referral letter) and medical information (e.g., cholelithiasis (diagnosis)). In the template information 20, each piece of part information is arranged in line according to the arrangement information. Specifically, a preamble part 21 is arranged at the beginning of the template information 20. Below that, Item name part 22a called "Disease name" and diagnosis name part 22b called "<Diagnosis name>" - Item name part 23a called "Referral purpose" and Referral purpose part 23b called "Request for <Referral purpose>" - Item name part 24a called "Medical history" and medical history part 24b called "<Medical history>" "Medical Progress" item name part 25a and progress part 25b to progress part 25e "Discharge prescription" item name part 26a and "<Discharge prescription>" item name part 26b Ending sentence part 27 Each includes:

[0075] In this way, processor 111 acquires template information by generating one piece of template information by combining multiple pieces of part information in accordance with the processing flow of FIG. 4C.

[0076] (C) Query information generation process Returning to FIG. 4B again, the processor 111 generates query information based on at least one of the medical information (first medical information) output in S116 of FIG. 4A and the medical information (second medical information) edited in S212 (S214). Here, in the processing flow of FIG. 4B, a second trained model is used, for example, to generate the medical document. An example of such a second trained model is a trained large-scale language model. The query information is information indicating a request input to such a trained large-scale language model. The processor 111 refers to a query management table (not shown) stored in the memory 112 or the like, and reads out one piece of query form information identified based on the medical information and the type information in S112 of FIG. 4A. The processor 111 then generates query information by overwriting the read query form information with the medical information and the template information generated in S213.

[0077] FIG. 7 is a diagram illustrating an example of query information processed by the processing device 100 according to an embodiment of the present disclosure. Specifically, it is a diagram illustrating an example of query information 30 generated by the processing of S214 in FIG. 4B. According to FIG. 7, the query information 30 includes a standard region 31 and a standard region 33 having content common to each type of information, a medical information region 32 in which at least one of the medical information (first medical information) output in S116 of FIG. 4A and the medical information (second medical information) edited in S212 is input, and a template region 34 in which the template information generated in S213 is input. Thus, according to the query information illustrated in FIG. 7, the standard region 31 includes a request to a trained large-scale language model, "Please create a referral letter for referring a patient to an external medical institution, based on the following information, as a polite written message to convey to the medical institution," and a request to a trained large-scale language model, "Please write the sentence by filling in the "<>" part of the following template.", both in plain text format. Although the example shown in Figure 7 uses plain text format, the format of the query information may be any language or model that can be processed by the trained model, and may be text data, image data, raw diagnostic information data, tabulated diagnostic information data, or a combination thereof.

[0078] (D) Medical document generation processing Returning to FIG. 4B again, the processor 111 acquires the medical document generated based on the query information of S214 (S215). As an example of this process, the processor 111 acquires the medical document by sending the generated query information to the second trained model. Specifically, the processor 111 inputs the query information to the second trained model, thereby acquiring the medical document requested by the query information as an output.

[0079] Here, the second trained model may be any model generated by machine learning. When a trained large-scale language model (i.e., a natural language processing model trained using a large amount of text data) is used as the second trained model, various models can be used, such as general-purpose language models such as GPT-3, GPT-4, Claude, PaLM, NeMo LLM, or LLaMA, or language models related to the medical field such as GPT-MD or MED-PaLM. However, without being limited to these, models such as BERT and Word2vec, other neural networks, support vector machines, logistic regression, decision trees, random forests, or combinations of these models may also be used as the trained model.

[0080] FIG. 8 is a diagram illustrating an example of a medical document output by the processing device 100 according to an embodiment of the present disclosure. Specifically, it is a diagram illustrating an example of a medical document 40 acquired by the processing of S215 in FIG. 4B. As shown in FIG. 8, the medical document 40 is written in a manner that fills in the "<XXX>" portion of the template information with medical information contained in the query information in accordance with the request contained in the query information. Furthermore, in the medical document 40, in accordance with the request contained in the query information, not only is the medical information simply embedded in the template information, but the template information is also modified to use more natural language expressions and more general terminology in order to create more polite writing.

[0081] Specifically, in areas 41 to 43 and areas 57 to 59, the medical information included in the query information is embedded according to the designation of "<XXX>" in the template information. In areas 44 to 56, the medical information included in the query information is embedded according to the designation of "<XXX>" in the template information, and the sentences of the medical information and template information are corrected by adding terms such as particles and auxiliary verbs to make the language expression more natural, or by changing the terms to more general terms.

[0082] In this way, by utilizing a trained large-scale language model, it is possible to generate desired medical documents using more natural language expressions, rather than formal sentence expressions that simply embed template information and medical information.

[0083] When a model other than a large-scale language model or a combination thereof is used as the second trained model, it is possible to generate medical documents by converting information included in the query information into numerical features using the trained model and linking the numerical features to related medical terms and medical documents by clustering, etc. However, without being limited to this, medical documents may also be generated by processing using other trained models, such as a method of directly classifying information included in the query information into medical terms and documents to be generated using a trained model.

[0084] 4B, when processor 111 acquires the medical document, it stores the acquired medical document in memory 112 and outputs a second screen including the medical document (S216). As an example of this process, processor 111 acquires the medical document generated in S215 and generates a second screen including the medical document. Then, processor 111 outputs the generated second screen to a display or the like via output interface 114.

[0085] The second screen will be described in detail later, but the second screen is output adjacent to the first screen containing the first medical information and the second medical information. In other words, the second screen is displayed in a manner that allows the user to view it simultaneously with the first screen. Therefore, when the user checks the medical document displayed on the second screen, the user can simultaneously refer to the first medical information and the second medical information that are the source of that information, allowing the user to check the medical document more efficiently.

[0086] This completes the processing flow. Although not specifically shown in Fig. 4B, this processing may be performed periodically at a predetermined cycle, or may be performed every time an instruction input by the user is received, or may be performed at both of these timings.

[0087] 5. Example of a screen output from the processing device 100 9A to 9C are diagrams showing examples of screens output from the processing device 100 according to an embodiment of the present disclosure. Specifically, FIG. 9A shows an example of a screen 70a when a first screen including the first medical information output in S116 of FIG. 4A is output. FIG. 9B shows an example of a screen 70b when an edited screen is output in S212 of FIG. 4B. FIG. 9C shows an example of a screen 70c when a second screen including a medical document generated based on the edited medical information (second medical information) in S216 of FIG. 4B is output. Each of these screens is output via the output interface 114 by processing by the processor 111 of the processing device 100. The output destination may be a display of the processing device 100, a display device connected to the processing device 100 via a wired cable or the like, a medical staff terminal device, a subject terminal device, or the like.

[0088] FIG. 9A shows a screen 70a when the first screen including the first medical information output in S116 of FIG. 4A is output. The screen 70a includes user attribute tabs at the top for selecting user attribute information. On this screen, the processor 111 accepts a user's instruction input via the input interface 113 for any of the tabs included in the user attribute tabs (doctor tab 71a, nurse tab 71b, and other tab 71c), thereby selecting a desired user attribute. The screen 70a also includes a type candidate selection area 72 below the user attribute tab, which is selected according to the selected user attribute. The type candidate selection area 72 includes type information associated with the user attribute management table as options according to the selected user attribute. The processor 111 accepts a user's instruction input via the input interface 113 for any one of the options displayed in the type candidate selection area 72, thereby determining the type of the desired medical document.

[0089] In the example of FIG. 9A, among the user attribute tabs, the doctor tab 71a is displayed in a manner that allows it to be distinguished from the nurse tab 71b and the other tabs 71c. This indicates that "doctor" is currently selected as the user attribute. Furthermore, in the type candidate selection area 72, "letter of referral," "reply," and "discharge summary" are displayed as candidates for type information. This indicates that a list of type information associated with "doctor," the user attribute selected in the user attribute management table, is output. Furthermore, among the candidates, the checkbox for the "letter of referral" type is checked. This indicates that "letter of referral" has been determined as the type of medical document to be created.

[0090] 9A, for example, when the nurse tab 71b is selected, the nurse tab 71b is displayed in a distinguishable manner relative to the doctor tab 71a and the other tabs 71c. Furthermore, the type information associated with "nurse" in the user management table is displayed as a list in the type candidate selection area 72 as candidate type information. The same is true when the other tab 71c is selected.

[0091] Furthermore, the selectable user attribute tabs and the selectable type information candidates are not limited to those shown in Fig. 9A. Naturally, there may be more tabs depending on the number of user attributes, and there may be more candidates depending on the number of type information associated with each user attribute.

[0092] The screen 70a also includes a first screen 73 that includes at least first medical information, which is medical information acquired based on the determined type. The first screen 73 includes a format selection area 73a at the top for selecting one of multiple formats for the medical document to be generated. The format selection area 73a includes a list of formats that are pre-associated with the determined type information. The processor 111 selects the desired medical document format by accepting a user's instruction input for one of the multiple formats displayed in the format selection area 73a via the input interface 113.

[0093] In the example of FIG. 9A, the format selection area 73a displays "Treatment," "Examination," "Rehabilitation," "Follow-up Observation," "Examination Result Report," and "Progress Report" as format candidates. This is because the type information management table stores format ID information corresponding to these formats in association with the type information of "Letter of Referral." Furthermore, among the formats, the checkbox for the "Follow-up Observation" format is checked. This indicates that the "Follow-up Observation" format has been determined as the format of the medical document to be created. In other words, the medical document that is finally generated will be a medical document of a letter of referral regarding follow-up observation.

[0094] 9A, for example, when "Treatment" is selected in the format selection area 73a, the first medical information output on the first screen 73 and the medical document output on the second screen will be changed as needed according to the selected format. Also, when "Reply" is selected in the type candidate selection area 72, the format candidates displayed in the format selection area 73a will also be changed to different ones.

[0095] The first screen 73 also includes the first medical information of S116 in Fig. 4A below the format selection area 73a. That is, the first screen 73 includes each item of medical information extracted or converted in S115 and S116 in Fig. 4A and its specific information.

[0096] In the example of Fig. 9A, the first screen 73 displays the first medical information extracted or converted in S115 and S116 of Fig. 4, including the route of admission, date of hospital visit, chief complaint, date of discharge, outcome at discharge, medical history, diagnosis, date of onset of illness, details of medical treatment, and prescription at discharge. Additionally, various types of information, such as "emergency transport" (checked in the checkbox), date of hospital visit (2023 / 08 / 23), and details of the chief complaint (fever, chills), are displayed as specific extracted information corresponding to each item. This allows the user to quickly confirm the medical information required to generate the determined medical document.

[0097] Screen 70a also includes a second screen 74 that includes at least a medical document (a referral letter in the form of a follow-up observation) generated based on the medical information output on first screen 73. Note that the second screen 74 includes information on a medical document that was generated before the editing process of S212 shown in FIG. 4B was performed. Therefore, for example, the medical treatment details on first screen 73 display "2023 / 08 / 23: X-ray," while the corresponding medical treatment progress on second screen 74 displays "...Due to the results of the X-ray..."

[0098] 9A, the first screen 73 and the second screen 74 of the screen 70a are output adjacent to each other. That is, the user can simultaneously view the first screen 73 containing the medical information (first medical information) used to generate the medical document and the second screen 74 containing the medical document generated based on that information. Therefore, when the user checks the generated medical document, the user can simultaneously refer to and check the medical information that is the source of the medical document, allowing the user to check the medical document more efficiently.

[0099] Here, there are cases where the desired information cannot be correctly extracted or converted in the processes of S115 and S116 in Fig. 4A. Furthermore, there are also cases where the medical information required to generate a medical document is not stored in the subject management table in the first place. To flexibly respond to such cases, it is possible to edit the output medical information based on user input instructions, as shown in S212 in Fig. 4B.

[0100] 9B shows an example of screen 70b when the editing screen output in S212 of FIG. 4B is output as described above. Specifically, screen 70b includes editing screen 91 and third screen 92 that are displayed superimposed on first screen 73 and second screen 74. Screen 70b is output by processor 111 accepting, via input interface 113, an instruction input by the user regarding an item to be edited on first screen 73 of screen 70a.

[0101] This shows an example of a screen that is output when a user inputs instructions for the "Medical Treatment Details" item on the first screen 73 of the screen 70a. Therefore, the editing screen 91 includes input boxes (box 93a, box 93b, and box 93c) below the "Medical Treatment Details" item for the period, classification, and details to be input as the medical treatment details. The editing screen 91 also displays information that has already been output as the first medical information in each input box, but also includes an add icon 94 and a delete icon 95 for adding new medical treatment details or deleting previously input medical treatment details. Furthermore, the editing screen 91 includes a confirm icon 96 for confirming the edited content and returning to the screen 70a.

[0102] Furthermore, the medical information entered in the medical treatment details field is information extracted based on the medical record information from the medical information stored in the subject management table, and the third screen 92 includes the medical record information used in the medical treatment details field. That is, the processor 111 outputs the third screen 92 including medical information (third medical information) that served as the information source for the first medical information output on the first screen from among the multiple pieces of medical information stored in the subject management table. The third screen 92 also includes a "back" icon 98 and a "next" icon 97 for switching the medical information to be output to other medical information (other third medical information) that served as the information source for the first medical information.

[0103] In the example of FIG. 9B, the third screen 92 includes the medical record information 10 that was the source of the first medical information, "August 23, 2023: X-ray," extracted in S115 and S116 of FIG. 4A. Referring to the medical record information 10, it states that "Pneumonia was diagnosed based on the results of the CT scan and blood test." However, there is no mention of "X-ray." In other words, by referring to the medical record information 10 entered on August 23, 2023, it is determined that the information "X-ray" entered in the classification input box 93b on the editing screen 91 is incorrect. Therefore, the processor 111 accepts a user's instruction input via the input interface 113 into the input box 93b where "X-ray" was entered, and edits the information to the correct information, "Blood test."

[0104] The processor 111 also accepts user input of instructions to the "Back" icon 98 and the "Next" icon 97 on the third screen 92 via the input interface 113, and switches the medical information that serves as the source of information such as the medical record information to be referenced. This makes it possible to check for errors or omissions in the medical information entered on the editing screen 91. Then, depending on the results of the check, the processor 111 accepts user input of instructions to the add icon 94, the delete icon 95, and each input box, and edits the medical information.

[0105] In this way, when editing medical information, editing can be performed more quickly and accurately by displaying the information that serves as the source of the information in parallel.

[0106] Fig. 9C shows an example of a screen 70c when a second screen including a medical document generated based on the edited medical information (second medical information) in S216 of Fig. 4B is output. Similar to Fig. 9A, the screen 70c includes user attribute tabs (doctor tab 71a, nurse tab 71b, and other tab 71c), a type candidate selection area 72, a first screen 73, and a second screen 74. Of these, the user attribute tab and type candidate selection area 72 are the same as those in Fig. 9A, and therefore their description will be omitted.

[0107] According to Fig. 9C, the first screen 73 includes each medical information item and its specific information, similar to Fig. 9A. However, the first screen 73 in Fig. 9C includes the medical information (second medical information) after editing in S212 of Fig. 4B. Specifically, as a result of editing the medical history and medical treatment details items on the editing screen shown in Fig. 9B, the following edited medical information (second medical information) is output on the first screen 73 in Fig. 9C. Dashed box 81: Newly enter information such as "diabetes mellitus, dyslipidemia" in the medical history section - Dashed box 82: In the medical treatment details section, the information "X-ray" was changed to "blood sampling." - Dashed box 83: In the medical treatment details section, newly enter the information "2023 / 08 / 24~2023 / 08 / 28: Medicinal treatment (Furosemide tablets 40mg "NP")" - Dashed box 83: In the medical treatment details section, the administration period for Hamp Injection 1000 was changed from "2023 / 08 / 24" to "2023 / 08 / 24~2023 / 08 / 28." Dashed box 84: In the medical treatment details section, newly enter the information "2023 / 08 / 28~2023 / 08 / 30: Medications (Entresto tablets 50mg, Samsca tablets 7.5mg)"

[0108] Of the medical information included in the first screen 73, the medical information (second medical information) other than that indicated by the dashed frame 81 to dashed frame 84 is information that has not been edited, etc. Therefore, the first medical information output on the first screen 73 in Fig. 9A is output as is.

[0109] The second screen 74 includes a medical document output by the processing of S216 in Fig. 4B based on the edited medical information (second medical information) and the unedited medical information (first medical information). That is, the second screen 74 includes a medical document whose type is a referral letter and whose format is follow-up observation. Because this medical document has been generated based on the edited medical information (second medical information), the following changes have been made to the medical document (Fig. 9A) generated based on the unedited medical information: - Dashed box 85: Added information "Diabetes mellitus, dyslipidemia" to the medical history section - Dashed box 86: In the medical history section, the word "X-ray" was changed to "blood sampling." - Dashed box 87: In the medical history section, the statement "Furosemide and Hamp were administered the following day." was corrected to "Furosemide and Hamp were administered the following day and continued until August 28th." - Dashed box 88: In the medical history section, the following statement was added: "In addition, Entresto-Samsca was administered from August 28 until discharge."

[0110] In this way, the user can simultaneously view the first screen 73 containing the medical information (first medical information) used to generate the medical document and the second screen 74 containing the medical document generated based on that information. Therefore, when checking the generated medical document, the user can simultaneously refer to the medical information that serves as the source of the medical document, allowing the user to check the medical document more efficiently. Furthermore, when medical information is edited on the first screen 73, the medical document generated based on the edited medical information is output to the second screen 74. Therefore, the user can check the medical document generated based on the edited medical information in real time.

[0111] 9C, the first screen 73 includes edited medical information (second medical information) and unedited medical information (first medical information), as described above. Therefore, it is possible to output both types of medical information in a manner that distinguishes them from each other on the first screen 73. Specifically, the processor 111 displays, for example, dashed frame 81 to dashed frame 84 shown in FIG. 9C for the edited medical information (second medical information) to distinguish it from the unedited medical information (first medical information).

[0112] Similarly, in Fig. 9C, the second screen 74 includes a medical document generated based on the edited medical information (second medical information) and the unedited medical information (first medical information). Therefore, it is possible to output the medical document generated based on the second medical information and the medical document generated based on the first medical information in a manner that distinguishes them from each other on the second screen 74. Specifically, the processor 111 displays, for example, dashed frame 85 to dashed frame 88 shown in Fig. 9C for the portion generated based on the edited medical information (second medical information), thereby outputting the medical document so as to distinguish it from the portion generated based on the unedited medical information (first medical information).

[0113] It should be noted that the dashed frame used for distinction in the above is merely one example. For example, instead of or in addition to the dashed frame, various methods can be employed, such as highlighting the relevant portion, displaying the information before editing by inputting an instruction to the relevant portion via the input interface 113, or displaying an identification mark indicating that the portion has been edited.

[0114] As described above, in this embodiment, it is possible to provide a processing device, a processing program, and a processing method that are capable of generating medical documents more efficiently.

[0115] 6.Other <Second trained model> In the above embodiment, the description has focused on the case where a trained large-scale language model is used to generate medical documents. However, instead, it is also possible to generate medical documents using another second trained model or predetermined medical document generation rules. For example, the second trained model is generated using the following processing flow. The processor 111 acquires, as training data, the type information actually specified by a medical professional or the like when creating a training medical document and the medical information used. Next, the processor 111 acquires document data of a training medical document that was actually created using this training data. The processor 111 then executes a step of performing machine learning of a medical document creation pattern using a combination of the training data of the type information and medical information and the training medical document. As an example, this machine learning is performed by providing a neural network composed of a combination of neurons with these sets of information and repeatedly learning while adjusting the parameters of each neuron so that the output from the neural network is the same as the training medical document, which is the correct label. Then, through this learning, the processor 111 executes a step of acquiring a trained model (S215). The acquired trained model may be stored as a second trained model in the memory 112 of the processing device 100 or in another device connected to the processing device 100 via a wired or wireless network.

[0116] Then, the processor 111 inputs the type information and medical information included in the generation request into the second trained model generated as described above, thereby obtaining a medical document as an output.

[0117] <Medical information and query information> In the above embodiment, the case where medical information is read from the subject management table in S112 of FIG. 4A has been described. However, instead of this, it is also possible to accept input of necessary medical information by the user via, for example, the input interface 113, and use the medical information for subsequent processing. Also, the case where query information is generated based on medical information, etc. has been described in S214 of FIG. 4B. However, instead of this, it is also possible to generate query information by accepting input by the user via the input interface 113 while the user refers to medical information output to a display device via, for example, the output interface 114.

[0118] <Template information> In the above embodiment, the template management table stores type information and medical information in association with template ID information, and a single piece of template information is acquired based on the type information and medical information. However, instead of or in addition to the type information and medical information, attribute information may be stored in association with template ID information, and a single piece of template information may be generated based on the attribute information. Examples of such attribute information include the ending of sentences included in the medical document ("noun-finished" or "predicate-finished"), the level of detail (a version with many items or a version with few items), whether personal information is included, and whether measurement data is attached. In this way, acquiring template information while further considering attribute information enables more detailed template information to be set.

[0119] <Using editing processes> In the above embodiment, as shown in Figure 4A, a case has been described in which the first medical information is extracted using a first trained model when outputting a first screen containing the first medical information. Also, in the above embodiment, as shown in Figure 4B, a case has been described in which a medical document is generated based on second medical information that has been edited from the first medical information contained in the first screen by accepting an instruction input by a user. Therefore, the first medical information or the first screen containing the first medical information is optimized using at least one of the editing instruction input by the user and the edited second medical information.

[0120] Specifically, processor 111 provides the model generation device of the first trained model with the instruction input accepted in S211 of FIG. 4B or the second medical information acquired by editing in S212 as training data. Then, the processor of the model generation device performs additional training of the first trained model based on the acquired instruction input or second medical information. When processor 111 accepts a new subject selection in S111 of FIG. 4, it outputs a new first medical screen including new first medical information that differs from the first screen including the edited first medical information (S116). At this time, processor 111 extracts the first medical information using the first trained model that has undergone additional training (S114), and is therefore able to output a first screen optimized in consideration of the instruction input accepted in S211 or the content edited in S212.

[0121] The optimization process described above is not limited to the method of additional learning, but can also be performed by modifying the parameters used in the process of extracting (S114) and converting (S115) the first medical information contained in the first screen based on the input of editing instructions and the information on the second medical information after editing, or by updating the contents of the extraction table and conversion table used in the above process.

[0122] It should be noted that the above modifications can naturally be implemented in appropriate combinations.

[0123] The processes and procedures described herein can be realized not only by those explicitly described in the embodiments, but also by software, hardware, or a combination thereof. Specifically, the processes and procedures described herein can be realized by implementing logic corresponding to the processes in media such as integrated circuits, volatile memory, nonvolatile memory, magnetic disks, and optical storage. Furthermore, the processes and procedures described herein can be implemented as computer programs and executed by various computers, including processing devices and server devices.

[0124] Although processes and procedures described herein are described as being performed by a single device, software, component, or module, such processes or procedures may be performed by multiple devices, multiple software, multiple components, and / or multiple modules. Furthermore, although various information described herein is described as being stored in a single memory or storage unit, such information may be stored in multiple memories within a single device or multiple memories distributed across multiple devices. Furthermore, software and hardware elements described herein may be realized by integrating them into fewer components or by decomposing them into more components. [Explanation of symbols]

[0125] 1 Processing System 100 Processing equipment 200 Information provision device

Claims

1. means for acquiring medical record information from an information providing device; A means for extracting medical information from the acquired medical record information; means for normalizing the extracted medical information; A means for generating a medical document by inputting input information based on the normalized medical information into a trained model; A processing device having:

2. The processing device of claim 1 , wherein the medical documents include at least one of a discharge summary, a nursing summary, a referral letter, and a symptom description.

3. The processing device according to claim 1 , wherein the means for extracting medical information extracts the medical information from the acquired medical record information based on the name of a disease.

4. The processing device according to claim 1 , wherein the means for extracting medical information extracts the medical information from the acquired medical record information based on a type of the medical document to be generated.

5. The medical information extracted from the medical record information includes at least one of admission and discharge information and test data. The processing device of claim 1 .

6. The processing device according to claim 1 , further comprising means for outputting the generated medical document to a terminal device usable by a medical professional.

7. 2. The processing device according to claim 1, further comprising means for outputting a first screen including the extracted medical information and a second screen including the generated medical document in a manner that allows a user to view them simultaneously.

8. a means for receiving an input of an instruction to edit the first medical information displayed on the first screen from a user; When the instruction input is accepted, a medical document generated based on second medical information, which is medical information edited in accordance with the instruction input, is displayed on the second screen. The processing device of claim 7 .

9. The processing device according to claim 1 , wherein the processing device is an on-premise server device.

10. A program for causing a computer to function as each of the means of the processing device according to any one of claims 1 to 9.

11. A computer comprising: Acquire medical record information from the information providing device; extracting medical information from the acquired medical record information; normalizing the extracted medical information; generating a medical document by inputting input information based on the normalized medical information into a trained model; How to perform the action.

Images