Text generation device, text generation method, and computer program

Abstract

This text generation device comprises an acquisition means that acquires electronic medical expense breakdown information for a patient, a generation means that uses a trained model that has learned past electronic medical expense breakdown information that excludes detailed symptom statements and the patterns of the text of detailed symptom statements to generate template text for a detailed symptom statement for the patient from the electronic medical expense breakdown information for the patient, and an output means that outputs the generated template text.

Description

Text generation device, text generation method, and recording medium

[0001] The present disclosure relates to a text generation device, a text generation method, and a recording medium.

[0002] There are technologies to reduce the burden of preparing documents for doctors and other medical professionals.

[0003] For example, Patent Document 1 discloses a method of extracting written expressions contained in medical information, interpreting the extracted written expressions as medical facts, extracting definite diseases that have been definitively diagnosed from the medical information, and creating an explanation of the diagnostic basis for the definite diseases that have been definitively diagnosed based on the medical facts.

[0004] JP 2015-138402 A

[0005] Incidentally, medical institutions prepare a medical fee statement (receipt) when claiming medical fees. If a medical institution feels that the explanation of the treatment is insufficient based on only the name of the illness or injury, it is required to include details of the treatment in the symptom description section of the receipt. However, including details of the treatment in the symptom description section is time-consuming.

[0006] An example of an objective of the present disclosure is to provide a sentence generation device or the like that can reduce the effort required to write detailed symptom descriptions.

[0007] A text generation device in one aspect of the present disclosure includes an acquisition means for acquiring a patient's electronic prescription information, a generation means for generating a template text for the patient's detailed symptom description from the patient's electronic prescription information using a trained model that has learned past electronic prescription information excluding detailed symptom descriptions and patterns of symptom description text, and an output means for outputting the generated template text.

[0008] In one aspect of the present disclosure, a text generation method involves a computer acquiring a patient's electronic prescription information, using a trained model that has learned past electronic prescription information excluding symptom details and symptom description text patterns, generating a template text for the patient's symptom details from the patient's electronic prescription information, and outputting the generated template text.

[0009] In one aspect of the present disclosure, a recording medium stores a program that causes a computer to acquire a patient's electronic prescription information, generate a template sentence for the patient's detailed symptoms from the patient's electronic prescription information using a trained model that has learned past electronic prescription information excluding detailed symptoms and patterns of sentences in the detailed symptoms, and output the generated template sentence.

[0010] According to one example of the effect of the present disclosure, the effort required to write detailed symptom descriptions can be reduced.

[0011] FIG. 1 is a block diagram showing the configuration of a sentence generation device according to the present disclosure. FIG. 2 is a diagram showing a hardware configuration in which the sentence generation device according to the present disclosure is realized by a computer device and its peripheral devices. FIG. 3 is a diagram showing an example of a template sentence according to the present disclosure. FIG. 4 is a flowchart showing the operation of sentence generation according to the present disclosure. FIG. 5 is a block diagram showing the configuration of a sentence generation device according to the present disclosure. FIG. 6 is a diagram showing an example of a sentence detailing symptoms according to the present disclosure. FIG. 7 is a flowchart showing the operation of sentence generation according to the present disclosure. FIG. 8 is a diagram showing an example of a sentence detailing symptoms according to the present disclosure.

[0012] Hereinafter, with reference to the drawings, embodiments of a text generation device, a text generation method, a program, and a non-transitory recording medium for recording a program according to the present disclosure will be described in detail. The present embodiments do not limit the disclosed technology.

[0013] [First Embodiment] Fig. 1 is a block diagram showing the configuration of a text generation device 100 according to the present disclosure. As shown in Fig. 1, the text generation device 100 includes an acquisition unit 101, a generation unit 102, and an output unit 103. The text generation device 100 is a device that creates a template text for a "detailed symptom description" that needs to be written when the explanation of the medical treatment is deemed insufficient based only on the name of the illness or injury on a medical fee statement (receipt). A medical fee statement is a document used by a medical institution to bill a health insurance association for medical expenses. The detailed symptom description needs to describe the progression of the patient's symptoms, the details of the medical treatment, and the specific reasons why the medical treatment is necessary.

[0014] In addition, when a medical institution submits a medical fee claim, if the reviewing organization determines that the claim content, including the detailed symptom description, is inappropriate or questionable, an assessment or refund may be made. Assessment refers to the reviewing organization deducting points from the content of the items that it deems inappropriate, and then paying the reduced amount of medical fee. On the other hand, a refund refers to the reviewing organization unilaterally sending back a claim when it is difficult to determine whether the medical procedure was appropriate. If the assessment is reduced due to incomplete detailed symptom description, the patient will not receive fair compensation for the medical treatment. Furthermore, if a refund is made, the patient will have to submit the claim again, which is a hassle.

[0015] 2 is a diagram showing an example of a hardware configuration in which the sentence generation device 100 according to the present disclosure is realized by a computer device 500 including a processor. As shown in FIG. 2, the sentence generation device 100 includes a CPU (Central Processing Unit) 501, memories such as a ROM (Read Only Memory) 502 and a RAM (Random Access Memory) 503, a storage device 505 such as a hard disk for storing a program 504, a communication I / F (Interface) 508 for network connection, and an input / output interface 511 for inputting and outputting data. In the present disclosure, electronic receipt information acquired by the acquisition unit 101 is input to the sentence generation device 100 via the communication I / F 508.

[0016] The CPU 501 runs an operating system to control the entire text generation device 100 according to the present disclosure. The CPU 501 also reads programs and data into memory from a recording medium 506 attached to a drive device 507, for example. The CPU 501 also functions as the acquisition unit 101, generation unit 102, and output unit 103 according to the present disclosure, or as part of these units, and executes processing or commands in the flowchart shown in FIG. 4, which will be described later, based on the program.

[0017] The recording medium 506 is, for example, an optical disk, a flexible disk, a magneto-optical disk, an external hard disk, or a semiconductor memory. A part of the recording medium in the storage device is a non-volatile storage device, and the program is recorded therein. The program may also be downloaded from an external computer (not shown) connected to a communication network.

[0018] The input device 509 is realized by, for example, a mouse, a keyboard, built-in key buttons, etc., and is used for input operations. The input device 509 is not limited to a mouse, a keyboard, or built-in key buttons, and may be, for example, a touch panel. The output device 510 is realized by, for example, a display, and is used to check output.

[0019] As described above, the text generation device 100 shown in FIG. 1 is realized by the computer hardware shown in FIG. 2. However, the means for realizing each part of the text generation device 100 in FIG. 1 is not limited to the configuration described above. Furthermore, the text generation device 100 may be realized by a single physically coupled device, or may be realized by two or more physically separate devices connected by wire or wirelessly. For example, the input device 509 and the output device 510 may be connected to the computer device 500 via a network. Furthermore, the text generation device 100 shown in FIG. 1 can also be configured using cloud computing or the like.

[0020] The acquisition unit 101 is a means for acquiring electronic medical receipt information for a patient. Electronic medical receipt information is electronic data in which a medical fee statement is recorded in text format such as CSV (Comma Separated Values), and is information that a medical institution submits to an examination and payment institution. In the present disclosure, the electronic medical receipt information acquired by the acquisition unit 101 is electronic medical receipt information that requires detailed symptom information to be entered, but is data in which the detailed symptom information has not yet been entered. The acquisition unit 101 acquires electronic medical receipt information for a specific patient stored in a server device or the like of the medical institution, and outputs it to the generation unit 102.

[0021] The acquiring unit 101 may extract electronic medical receipt information requiring detailed symptom description from electronic medical receipt information created at predetermined intervals. For example, the acquiring unit 101 identifies conditions requiring detailed symptom description from previously submitted electronic medical receipt information, and extracts electronic medical receipt information that meets the conditions. Specifically, the required conditions are the name of a patient's illness or injury, the content of medical treatment, etc., that require detailed symptom description, which are obtained by machine learning of previously submitted electronic medical receipt information.

[0022] The generation unit 102 is a means for generating a template sentence for a patient's detailed symptoms from the patient's electronic prescription information using a trained model that has learned past electronic prescription information excluding detailed symptoms and patterns of sentences for detailed symptoms.

[0023] The generation unit 102 inputs the electronic medical receipt information of the target patient into the trained model to obtain a template sentence for the patient's detailed symptom description generated by the trained model. The template sentence is a template for describing the specific reasons why the medical procedure performed on the patient is necessary, and is text data obtained by excluding the patient's unique information from the detailed symptom description. Here, the unique information is information unique to each patient, including personal information such as the patient's name, date of birth, and electronic medical record information.

[0024] The trained model is a model that has learned from medical receipt information excluding symptom details and symptom description sentence patterns previously submitted by medical institutions. Medical receipt information excluding symptom details includes, for example, the name of an illness or injury, medical treatment, etc. A symptom description sentence pattern is, for example, the items included in the symptom description sentence for a specific medical treatment or illness or injury name. However, the symptom description in electronic medical receipt information for appraisal or return may be trained to the symptom description sentence pattern when the medical institution is paid the amount of medical treatment fee claimed.

[0025] The trained model outputs a template sentence in response to a prompt that instructs template generation, for example. The prompt may include conditions for the template sentence to be generated in addition to electronic prescription information. The conditions may be, for example, the length of the sentence, the degree of summarization, and data to be included in the symptom description. The length of the sentence is indicated, for example, by the number of characters or the number of lines. The degree of summarization indicates, for example, the extent to which the sentence is summarized relative to the length of the original sentence.

[0026] The trained model may be an interactive model that receives prompts and responds to them. The trained model may be a large-scale language model trained with a large amount of text data, or a model obtained by transfer learning of the large-scale language model. Examples of the large-scale language model that can be used include Generative Pre-Training-2 (GPT-2), GPT-3, and GPT-4. Examples of the large-scale language model that can be used include Text-to-Text Transfer Transformer (T5), Bidirectional Encoder Representations from Transformers (BERT), Robustly optimized BERT approach (RoBERTa), and Efficiently Learning an Encoder that Classifies Token Replacements Accurately (ELECTRA). The large-scale language model may be generated in a system external to the text generation device 100 and stored in the storage device 505. The process of generating text using the large-scale language model may be performed in a system external to the text generation device 100.

[0027] The output unit 103 is a means for displaying the generated template sentence on a display device such as a monitor. The output unit 103 outputs the template sentence, for example, on the screen of an application program for generating a detailed symptom description. The output unit 103 may also output patient-specific information so that it can be input. The output unit 103 may also transmit the generated template sentence to a terminal device used by a doctor, or may output the template sentence in response to a doctor's operation to create a detailed symptom description for a specific patient.

[0028] FIG. 3 is an example of a template sentence for describing symptoms in detail in the present disclosure. The example in FIG. 3 is a template sentence for when a red blood cell transfusion is performed as a medical procedure. As shown in FIG. 3, the template sentence generated by the output unit 103 has blank spaces for entering patient-specific information, allowing for entry on the screen. In the example in FIG. 3, the spaces for entering the doctor's findings and the patient's symptoms are blank spaces. The template sentence in FIG. 3 is an example, and other template sentences may be output.

[0029] The operation of the sentence generation device 100 configured as above will be described with reference to the flowchart of FIG.

[0030] 4 is a flowchart showing an outline of the operation of the sentence generation device 100 according to the present disclosure. Note that the processing according to this flowchart may be executed based on program control by the processor described above. Note that the processing according to this flowchart is started in response to an operation for creating a template sentence detailing the symptoms of a specific patient.

[0031] As shown in FIG. 4 , first, the acquisition unit 101 acquires the patient's electronic medical receipt information (step S101). Next, the generation unit 102 generates a template sentence for the patient's detailed symptom description from the patient's electronic medical receipt information using a trained model that has learned past electronic medical receipt information excluding the detailed symptom description and the symptom description sentence patterns (step S102). Finally, the output unit 103 outputs the generated template sentence (step S103). This completes the analysis operation of the sentence generation device 100.

[0032] In the sentence generation device 100, the generation unit 102 generates a template sentence for a patient's detailed symptom description from the patient's electronic medical receipt information using a trained model that has learned past electronic medical receipt information excluding detailed symptom descriptions and the sentence patterns of detailed symptom descriptions. This allows a doctor to generate detailed symptom descriptions by inputting only patient-specific information, thereby reducing the effort required to write detailed symptom descriptions.

[0033] [Second Embodiment] Next, a second embodiment of the present disclosure will be described in detail with reference to the drawings. Below, descriptions of content that overlaps with the above description will be omitted to the extent that the description of this embodiment is not unclear. As with the computer device shown in FIG. 2, the functions of each component in each embodiment of the present disclosure can be realized not only by hardware but also by a computer device or software based on program control.

[0034] 5 is a block diagram showing the configuration of a text generation device 110 according to the present disclosure. Referring to FIG. 5, the text generation device 110 will be described, focusing on the differences from the text generation device 100. In this embodiment, a case is assumed in which a text detailing symptoms is generated by inserting patient-specific information into a template text based on electronic medical record information. The text generation device 110 includes an acquisition unit 111, a generation unit 112, an output unit 113, and a reception unit 114.

[0035] The acquisition unit 111 acquires electronic medical record information of a patient in addition to the functions of the acquisition unit 101. The acquisition unit 111 acquires electronic medical record information linked to a patient name and a patient ID from a server device of a medical institution, and outputs the information to the generation unit 112.

[0036] The generation unit 112 generates a symptom description sentence by inserting the patient's unique information into the generated template sentence based on the electronic medical record information. For example, the generation unit 112 analyzes the context before and after blank spaces in the template using natural language processing and searches the patient's electronic medical record information for unique information to be inserted into the blank spaces. Specifically, when inserting the patient's symptoms into blank spaces in the template, the generation unit 112 inserts sentences corresponding to the patient's symptoms from the electronic medical record information. As another example, when inserting a doctor's findings into blank spaces in the template, the generation unit 112 inserts corresponding sentences from the section of the electronic medical record information where the doctor's findings are recorded. Note that the trained model used to generate the template sentence and the trained model used to insert unique information into the template sentence may be the same or different.

[0037] The output unit 113 outputs the detailed symptom description text. The output unit 113 may also output electronic medical record information that is the basis for the patient's unique information. The output unit 113 may, for example, display a link to the electronic medical record information that is the source of the text inserted in the blank space of the template, making it possible to output the electronic medical record information. The output unit 113 may also output information that identifies the source electronic medical record information. The information that identifies the electronic medical record information is, for example, information such as the date and time when the electronic medical record information was input or the person who input it.

[0038] The receiving unit 114 is a means for receiving an instruction to adopt or not adopt the output template text or detailed symptom text. For example, the receiving unit 114 receives a response from the doctor on whether or not to adopt the output text on the screen on which the template text or detailed symptom text is output. Furthermore, when the receiving unit 114 receives an instruction not to adopt the output text, the receiving unit 114 may further receive an instruction to generate the template text or detailed symptom text by changing the text generation conditions. More specifically, the receiving unit 114 may receive an instruction to recreate the template text, or may receive an instruction to use the template text as is and insert patient-specific information again. Furthermore, the receiving unit 114 may receive at least one of the text length, the degree of summary, and data to be included in the detailed symptom text as conditions for regenerating the text.

[0039] FIG. 6 is an example of a symptom description sentence according to the present disclosure. As shown in FIG. 6, patient-specific information is inserted into the symptom description sentence 61. In FIG. 6, the inserted portion may be highlighted with underlining or boldface so that it can be identified, or may be editable on the screen. Also, as shown in FIG. 6, the basis 62 of the electronic medical record information is shown for each portion where the patient-specific information is inserted. In the example of FIG. 6, the date on which the information was entered is also shown. Also, in the example of FIG. 6, a screen 63 is displayed that includes a button for selecting whether or not to adopt the output symptom description sentence.

[0040] The operation of the sentence generation device 11 configured as above will be described with reference to the flowchart of FIG.

[0041] 7 is a flowchart showing an outline of the operation of the sentence generation device 11 according to the present disclosure. Note that the processing according to this flowchart may be executed based on program control by the processor described above.

[0042] As shown in FIG. 7 , first, the acquisition unit 111 acquires the patient's electronic medical receipt information (step S201). Next, the generation unit 112 generates a template for the patient's symptom description from the patient's electronic medical receipt information using a trained model that has learned past electronic medical receipt information excluding the symptom description and the symptom description sentence patterns (step S202). Next, the generation unit 112 generates the symptom description sentence by inserting the patient's unique information based on the electronic medical record information into the generated template sentence (step S203). Next, the output unit 113 outputs the symptom description sentence (step S204). Next, if the reception unit 114 receives a request to adopt the output symptom description sentence (S205; YES), the flow ends. On the other hand, If the accepting unit 114 does not accept the adoption of the output symptom description sentence (S205; NO), the generating unit 112 changes the generation conditions (step S206) and repeats S202 to S204. With this, the sentence generating device 110 ends the sentence generation operation.

[0043] In the text generation device 110 of the present disclosure, the generation unit 112 generates a symptom description text by inserting patient-specific information based on electronic medical record information into the generated template text. This allows the doctor to automatically generate the symptom description, further reducing the effort required to write the symptom description.

[0044] Furthermore, in the text generation device 110 of the present disclosure, the receiving unit 114 receives an instruction to adopt or not adopt the output symptom description text. This allows the doctor to confirm the content of the symptom description text. Furthermore, if the receiving unit 114 receives an instruction not to adopt the output symptom description text, it receives an instruction to change the text generation conditions and generate a template text or symptom description text. This makes it possible to regenerate a text that reflects the doctor's intention to amend it.

[0045] Although the present invention has been described above with reference to the embodiments, the present invention is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

[0046] For example, although multiple operations are described in a sequential order in the form of a flowchart, the order of description does not limit the order in which the multiple operations are performed. Therefore, when implementing each embodiment, the order of the multiple operations can be changed within the scope that does not affect the content.

[0047] In each embodiment of the present disclosure, the generation units 102 and 112 may generate symptom description template sentences or symptom description sentences using a plurality of different trained models, and the output units 103 and 113 may output the generated template sentences or symptom description sentences, respectively. In this case, the reception unit 114 may receive information on which model the template sentences or symptom description sentences generated by should be adopted. As the plurality of trained models, for example, a large-scale language model released as open source and a language model using a proprietary engine may be used.

[0048] 8 shows an example of a sentence generated in the present disclosure. As shown in FIG. 8, the output unit 113 outputs symptom description sentences generated using a plurality of different trained models. The doctor can then select which trained model to use to generate the symptom description sentence. The receiving unit 114 receives a request to use the symptom description sentence selected on the screen.

[0049] In the second embodiment, the output unit 113 may output a template sentence, and the receiving unit 114 may receive a request for whether or not to adopt the output template sentence. If the receiving unit 114 does not accept the request for adoption, the receiving unit 114 may accept a request to regenerate the template sentence by changing the sentence generation conditions.

[0050] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes.

[0051] (Supplementary Note 1) A text generation device comprising: an acquisition means for acquiring a patient's electronic prescription information; a generation means for generating a template text for the patient's detailed symptom description from the patient's electronic prescription information using a trained model that has learned past electronic prescription information excluding detailed symptom descriptions and patterns of the symptom description text; and an output means for outputting the generated template text.

[0052] (Appendix 2) The text generation device described in Appendix 1, wherein the acquisition means further acquires electronic medical record information of the patient, the generation means generates a text detailing symptoms by inserting the patient's unique information into the template text based on the electronic medical record information, and the output means outputs the text detailing symptoms.

[0053] (Supplementary Note 3) The text generation device according to Supplementary Note 2, wherein the output means outputs electronic medical record information that is a basis for the unique information.

[0054] (Appendix 4) The text generation device described in any of Appendices 1 to 3, wherein the acquisition means extracts electronic prescription information that requires detailed symptom descriptions from electronic prescription information created at predetermined intervals, and the generation means generates the template text or the detailed symptom description text for the extracted electronic prescription information.

[0055] (Supplementary Note 5) The text generation device according to any one of Supplementary Notes 1 to 4, further comprising a receiving unit that receives an indication of whether or not the template text or the symptom description text is to be adopted.

[0056] (Supplementary Note 6) The sentence generation device according to Supplementary Note 5, wherein the generation means generates a template sentence for describing symptoms or a sentence for describing symptoms using a plurality of different trained models, the output means outputs the generated template sentence or the generated sentence for describing symptoms, and the reception means receives a request as to which trained model the sentence generated by should be adopted.

[0057] (Supplementary Note 7) In the text generation device according to Supplementary Note 5, when the accepting means accepts that the text will not be adopted, the accepting means further accepts that the template text or the symptom description text will be regenerated by changing the text generation conditions.

[0058] (Appendix 8) The receiving means further receives at least one of the length of the sentence, the degree of summary, and the data to be included in the symptom description as conditions for regenerating the template sentence or the symptom description sentence, and the generating means regenerates the template sentence or the symptom description sentence under the conditions.

[0059] (Supplementary Note 9) The text generation device according to any one of Supplementary Notes 1 to 8, wherein the trained model is a large-scale language model.

[0060] (Supplementary Note 10) The sentence generation device according to any one of Supplementary Notes 1 to 9, wherein the trained model is a model optimized by machine learning.

[0061] (Appendix 11) A text generation method in which a computer acquires a patient's electronic prescription information, generates a template text for the patient's detailed symptom description from the patient's electronic prescription information using a trained model that has learned past electronic prescription information excluding detailed symptom descriptions and patterns of the symptom description text, and outputs the generated template text.

[0062] (Appendix 12) The text generation method described in Appendix 11 further includes obtaining the patient's electronic medical record information, inserting the patient's unique information into the template text based on the electronic medical record information to generate a text detailing symptoms, and outputting the text detailing symptoms.

[0063] (Supplementary Note 13) The text generation method according to Supplementary Note 12, wherein electronic medical record information that is the basis for the unique information is output.

[0064] (Appendix 14) A text generation method described in any of Appendices 11 to 13, which extracts electronic prescription information that requires detailed symptom information from electronic prescription information created at predetermined intervals, and generates the template text or the detailed symptom information for the extracted electronic prescription information.

[0065] (Supplementary Note 15) The text generation method according to any one of Supplementary Notes 11 to 14, wherein an input as to whether or not to adopt the template text or the symptom description text is received.

[0066] (Supplementary Note 16) A sentence generation method according to Supplementary Note 15, which uses a plurality of different trained models to generate a template sentence for describing symptoms or a sentence for describing symptoms, respectively, outputs the generated template sentence or the generated sentence for describing symptoms, and accepts a request as to which trained model the sentence generated by is to be adopted.

[0067] (Supplementary Note 17) The text generation method according to Supplementary Note 16, further receiving an indication of whether or not to adopt the output template text or the detailed symptom description text.

[0068] (Supplementary Note 18) The sentence generation method according to Supplementary Note 17, wherein, when the decision not to adopt is accepted, the method further accepts a request to change the sentence generation conditions and regenerate the template sentence or the symptom description sentence.

[0069] (Supplementary Note 19) The sentence generation method according to any one of Supplementary Notes 11 to 18, wherein the trained model is a large-scale language model.

[0070] (Appendix 20) A recording medium storing a program that causes a computer to acquire electronic prescription information of a patient, generate a template sentence for the patient's detailed symptoms from the electronic prescription information of the patient using a trained model that has learned past electronic prescription information excluding detailed symptoms and the pattern of the sentence for detailed symptoms, and output the generated template sentence.

[0071] Some or all of the configurations described in Supplements 2 to 10, which are dependent on Supplement 1 described above, may also be dependent on Supplements 11 and 20 in the same dependent relationship as Supplements 2 to 10. Not limited to Supplements 1, 11, and 20, some or all of the configurations described as Supplements may be made dependent on various hardware, software, various recording devices for recording software, or systems, within the scope of each of the above-mentioned embodiments.

[0072] 100, 110 Text generation device 101, 111 Acquisition unit 102, 112 Generation unit 103, 113 Output unit 114 Reception unit 500 Computer device 501 CPU 502 ROM 503 RAM 504 Program 505 Storage device 506 Recording medium 507 Drive device 508 Communication interface 511 Input / output interface 512 Bus

Claims

1. A means of obtaining electronic medical claim information from patients, A generation means for generating a template text for a patient's symptom description from the patient's electronic medical claim information, using a trained model that has learned the patterns of symptom description texts and electronic medical claim information excluding symptom descriptions from the past. A document generation device comprising an output means for outputting the generated template document.

2. The acquisition means further acquires the patient's electronic medical record information, The generation means generates a detailed symptom description by inserting patient-specific information into the template document based on the electronic medical record information. The output means outputs a detailed description of the symptoms, as described in claim 1.

3. The text generation device according to claim 2, wherein the output means outputs electronic medical record information that serves as the basis for the unique information.

4. The acquisition means extracts electronic medical claim information for which detailed symptom descriptions are required from electronic medical claim information created at predetermined intervals. The document generation apparatus according to claim 1, wherein the generation means generates the template document or the symptom detail document, respectively, from the extracted electronic medical claim information.

5. Furthermore, the document generation device according to any one of claims 1 to 4 is further equipped with a receiving means for receiving whether or not to adopt the template document or the document detailing the symptoms.

6. The generation means generates a template document or a document detailing symptoms using multiple different trained models, The output means outputs the generated template document or the detailed symptom description document, The text generation apparatus according to claim 5, wherein the receiving means receives a request to decide which of the trained models to use to generate the text.

7. The document generation device according to claim 5, wherein, if the receiving means receives a request not to adopt the request, it further receives a request to change the document generation conditions and regenerate a template document or a document detailing symptoms.

8. The receiving means further receives at least one of the following as conditions for regenerating the template document or the symptom description document: the length of the document, the degree of conciseness, and the data to be included in the symptom description document. The document generation device according to claim 7, wherein the generation means regenerates a template document or a document detailing symptoms under the conditions described above.

9. Computers We obtain the patient's electronic medical claim information, Using a trained model that has learned the patterns of electronic medical claim information excluding symptom descriptions from the past and the text of symptom descriptions, a template text for the patient's symptom description is generated from the patient's electronic medical claim information. A document generation method that outputs the generated template document.

10. We obtain the patient's electronic medical claim information, Using a trained model that has learned the patterns of electronic medical claim information excluding symptom descriptions from the past and the text of symptom descriptions, a template text for the patient's symptom description is generated from the patient's electronic medical claim information. A computer program that causes a computer to output the generated template document.