Information processing device, information processing method, and program

Abstract

The present invention provides an information processing device, information processing method, and program that can generate natural language sentences that can comprehensively and individually describe viewpoints required by the system while reducing the burden on practitioners. [Solution] The information processing device 10 is a device that generates natural language sentences based on application-related information of a treatment recipient to assist in the preparation of medical expense application documents based on the insurance system, and is characterized by comprising an information acquisition unit 12 that acquires application-related information input by a practitioner, a natural language processing unit 14 that generates natural language sentences to describe the reasons for continuing treatment according to the treatment recipient based on the application-related information, and an output unit 16 that outputs the natural language sentences generated by the natural language processing unit 14.

Description

[Technical Field]

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

[0002] When practitioners who perform medical-like procedures, such as judo therapists and acupuncturists, use health insurance to provide treatment, there is a system in place whereby a portion of the treatment costs are paid by the insurer if the patient (the person receiving treatment) applies for medical expenses. Under this system, practitioners must fill out the necessary information on a "Medical Expenses Reimbursement Application Form," and in the case of long-term or frequent treatment, they must also prepare and submit attached documents such as a "Letter of Reason for Continuing Long-Term Treatment" to the insurer. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7501883 [Patent Document 2] Patent No. 7669603 Summary of the Invention [Problem to be solved by the invention]

[0004] Traditionally, when applying for medical expenses, documents such as a statement of reasons for continuing long-term treatment prepared by a judo therapist or other such person have been required by the system to include elements such as the name of the injury or illness, specific details of the symptoms, the reason why treatment is necessary, any hindrances to daily life, malfunctions, etc., and the therapist must fill out these comprehensively and consistently. In recent years, insurers have been making stricter examinations, and applications that contain incomplete information or are notoriously "returned" have been increasingly common, increasing the administrative burden on the therapist.

[0005] However, many practitioners are unfamiliar with writing documents, making it difficult for them to properly record the need for long-term treatment for each patient. For this reason, there are cases where practitioners reuse documents they have prepared in the past (so-called "recycling"), which increases the risk of documents being returned or not being paid.

[0006] On the other hand, in recent years, advances in natural language processing technologies such as generative AI (large-scale language models) have made it possible to automatically generate natural-sounding sentences based on standard inputs and keywords. Utilizing these technologies is expected to support practitioners in inputting information and creating documents that meet written requirements.

[0007] The object of the present invention has been made in consideration of the above-mentioned problems, and is to provide an information processing device, an information processing method, and a program that are capable of generating natural language sentences that can describe the perspectives required by the system in a comprehensive and individualized manner while reducing the burden on practitioners. [Means for solving the problem]

[0008] The information processing device according to the present invention is an information processing device that generates a natural language sentence based on application-related information of a treatment recipient in order to support the creation of a medical care expense application document based on an insurance system, and includes an information acquisition unit that acquires the application-related information input by a practitioner, and an information acquisition unit that acquires the application-related information of the treatment recipient based on the application-related information. Individual for each a natural language processing unit that generates the natural language sentence; a determination unit that determines whether the natural language sentence generated by the natural language processing unit includes predetermined aspects including the location of injury, symptoms, and reasons for the need to continue treatment based on an administrative notice required for applying for medical expenses; and an input assistance unit that, when the determination unit determines that there are missing items among the predetermined aspects, presents input items corresponding to the missing items to prompt the practitioner to enter additional information; and an output unit that outputs the natural language sentence generated by the natural language processing unit.

[0012] The information processing method according to the present invention is an information processing method for generating a natural language sentence based on application-related information of a treatment recipient in order to support the preparation of a medical care expense application document based on an insurance system, the method comprising: a step of acquiring the application-related information input by a practitioner; Individual for each generating the natural language sentence; a step of determining whether the natural language sentence contains predetermined aspects including the location of injury, symptoms, and reasons for the need to continue treatment based on an administrative notice required for applying for medical expenses; and a step of, when it is determined that there are missing items among the predetermined aspects, presenting input items corresponding to the missing items to prompt the practitioner to enter additional information; and outputting the natural language sentence generated by the generating step.

[0013] The program according to the present invention is a program for generating natural language sentences based on application-related information of a person to be treated in order to support the preparation of medical care expense application documents based on an insurance system, and the program includes a process for acquiring the application-related information input by a practitioner, and a process for generating natural language sentences based on the application-related information of the person to be treated. Individual for each A process for generating the natural language sentence; A process for determining whether the natural language sentence contains predetermined aspects including the location of injury, symptoms, and reasons for the need to continue treatment based on an administrative notice required for applying for medical expenses; and a process for prompting the practitioner to enter additional information by presenting input items corresponding to the missing items when it is determined that there are missing items among the predetermined aspects. and a process of outputting the natural language sentence obtained by the generation process. [Effects of the Invention]

[0014] According to the present invention, it is possible to automatically generate natural language sentences that cover all the perspectives required by the system and are individualized for each treatment recipient based on application-related information such as symptoms and names of injuries and illnesses entered by the practitioner. This improves the quality of medical care expense application documents such as the statement of reasons for continuing long-term treatment, and reduces the risk of refunds or non-payment by the insurer.

[0015] Furthermore, by equipping the system with a function to automatically determine whether the generated text appropriately incorporates institutional perspectives, it is possible to check for omissions and excesses in the information provided before the application stage and prompt the practitioner to enter additional information. This prevents practitioners from making mistakes or omissions in their entries, contributing to the efficiency and standardization of medical care expense application procedures.

[0016] Furthermore, in the text generation process, rather than simply reusing past standard phrases and templates, application-related information is analyzed and the sentence structure and expression are dynamically changed for each treatment recipient, thereby avoiding problems such as the subjectivity of the description and reuse, and enabling the creation of documents that conform to the purpose of the system. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a diagram illustrating an information processing apparatus according to an embodiment of the present invention. [Figure 2]10 is a flowchart illustrating a procedure for generating a sentence using an information processing device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0018] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following, similar elements in all drawings will be designated by the same reference numerals, and duplicate explanations will be omitted. Furthermore, in the description below, previously described reference numerals will be used as necessary.

[0019] 1 is a diagram showing an information processing device 10 according to an embodiment of the present invention. The information processing device 10 has a function of generating natural language sentences based on application-related information of a treatment recipient to support the preparation of medical care expense application documents under the insurance system. The information processing device 10 includes an information acquisition unit 12, a natural language processing unit 14, an output unit 16, a determination unit 18, and a storage unit 20.

[0020] The information processing device 10 is connected to a practitioner 4 (a judo therapist or an acupuncturist) via a network 2. The storage unit 20 may be configured to include a prompt database (prompt DB) that stores instruction sentence templates for effectively generating sentences for explaining why treatment should be continued, and the natural language processing unit 14 may be configured to automatically extract or compose an appropriate prompt from the prompt DB based on information input by the practitioner.

[0021] The information acquisition unit 12 has a function to acquire application-related information input by the practitioner 4. Specifically, the information includes the name of the injury or illness, the injured area, the start date of treatment, symptoms, problematic behavior, obstacles to daily life, the frequency of treatment, and the reason for continuing treatment. These are input by the practitioner 4 via an input form equipped with a GUI, and the information acquisition unit 12 records them in a structured format. Here, "injured area," "symptoms," and "reason for continuing treatment" are treated as keyword information that is particularly important for natural language sentence generation among the application-related information.

[0022] The information acquisition unit 12 can also be equipped with a complementary input function based on the input history of the practitioner 4 and past similar cases, thereby speeding up the input work and improving its accuracy. Furthermore, it is desirable to have a configuration that includes check logic to detect contradictions and inconsistencies in the input content and can monitor the consistency of the entries in real time.

[0023] The input form is equipped with multiple UI components such as pull-down menus, radio buttons, check boxes, and text boxes to improve input efficiency and standardize input content. Furthermore, it may have interactive support functions that automatically complete some items by referencing past application history and display warnings in real time for content where input errors are suspected.

[0024] Furthermore, the information acquisition unit 12 may be configured to have a function to link with the medical history database and medical records of the treatment facility, so that re-entry from past records can be omitted. It may also be configured to have a voice recognition function that converts voice input in natural language into text data and registers it.

[0025] The natural language processing unit 14 analyzes application-related information and dynamically generates natural language sentences suited to the treatment recipient. Rather than simply pasting templates, it automatically adjusts the sentence structure to capture the semantic structure of the information.

[0026] The natural language processing unit 14 changes the order of sentence structure and expressions depending on a combination of multiple elements such as the injured area, the type and degree of symptoms, the details of the malfunction, the obstacles to daily life, and the reason for continuing the treatment, so as to create a description that more accurately conveys the need for continuing the treatment. Furthermore, the natural language processing unit 14 may have a processing function of constructing an appropriate prompt (instruction sentence) via a prompt generation unit based on the application-related information including the "injured area," "symptoms," and "reason for continuing the treatment," and using this to form an input query to the generation AI.

[0027] The storage unit 20 may also be configured to include a prompt DB that manages prompt sentences used when the natural language processing unit 14 cooperates with the externally generated AI. The prompt DB associates appropriate instruction sentences with classifications of injury / illness name, injury location, symptoms, and reasons for continuing treatment, and the natural language processing unit 14 dynamically selects and edits the corresponding prompts based on application-related information, thereby making it possible to improve the quality and system conformity of the output natural language sentences.

[0028] The natural language processing unit 14 may be configured to link with an external large-scale language model (e.g., a cloud-based generation AI) via an API, etc., send the prompts and application information required for generation, and then format and edit the returned output sentence to create the final natural language sentence. This configuration enables flexible expression generation using advanced natural language understanding capabilities, contributing to improving the quality of descriptions and ensuring compliance with institutional requirements.

[0029] It is also desirable to adjust the system to generate natural-sounding sentences that are both individual and comprehensive, while referring to the input tendencies of the practitioner 4 and past document history. Utilizing an external generative AI can also improve the naturalness of the expressions and conformity to regulations. Furthermore, if the practitioner 4 makes manual corrections after the generated sentences are output, the corrections can be recorded in the memory unit 20 and used to adjust the personalized model for future sentence generation.

[0030] The output unit 16 displays the natural language sentence generated by the natural language processing unit 14 on a screen so that the practitioner 4 can check and correct it. The display screen preferably has a difference display function so that the original sentence and the corrections made by the practitioner 4 can be compared, which makes it easier to visualize the history and intention of the corrections.

[0031] The system also has a function to store the text in the storage unit 20 after the practitioner 4 has performed the approval operation. The storage format is expected to be plain text, PDF, or a format that conforms to the data format of the hospital system. Furthermore, it is desirable that the stored natural language text be managed in association with metadata such as the target practitioner, treatment date, and case classification, with the aim of future reference and reuse. In addition, if the stored documents are designed to be anonymized and used as a shared resource, it will also contribute to the sharing of knowledge among colleagues.

[0032] Additionally, the output unit 16 may have a function to highlight keywords and phrases that correspond to the regulatory requirements in the generated natural language text. For example, by displaying phrases such as the name of the injury or illness, the injured area, the start date of treatment, symptoms, problem behavior, obstacles to daily life, frequency of treatment, and reasons for continuing treatment in different colors, the final confirmation work by the practitioner can be visually supported.

[0033] The determination unit 18 determines whether the generated natural language sentence satisfies the viewpoints (description items) required by the insurance system. For example, it has a function to check whether the sentence contains the name of the injury or illness, the injured part, the start date of treatment, symptoms, troublesome behavior, obstacles to daily life, frequency of treatment, reason for continuing treatment, etc., based on the "Notice of Points to Note Regarding Applications for Medical Expenses for Judo Therapy Treatment" issued by the Ministry of Health, Labor and Welfare.

[0034] In addition, the judgment unit 18 may be configured to work in cooperation with a proofreading engine that automatically evaluates whether the generated natural language sentence contains grammatical errors or unnatural structures in Japanese, and to present candidate correction sentences as necessary.

[0035] If there is any missing information in the generated natural language sentence, the determination unit 18 identifies the relevant element and presents the corresponding input item to the practitioner 4. Possible presentation methods include redisplaying the input form, a pop-up warning, or an alert with a specific explanation of the missing element, and a design that naturally guides the practitioner 4 to enter additional information is desirable.

[0036] The new application-related information obtained in this way is immediately fed back to the natural language processing unit 14, and the text is regenerated. The regenerated document is structured to improve consistency of content and compliance with the system by taking into account the differences from the previous document and adjusting supplementary expressions and sentence structure.

[0037] The storage unit 20 can store and reuse application information, generated statements, and configuration check results as intermediate data, and it is expected that this will be used as reference information for similar cases in the future. This will enable practitioners 4 to expect continuous improvements in work efficiency, and will also help the system to standardize application documents and improve transparency.

[0038] Furthermore, the storage unit 20 may function as a patient information database (patient information DB) that can individually record and search the symptoms, treatment history, treatment frequency, past application details, generated sentence history, etc. for each treatment recipient (patient). This makes it possible to optimize document generation by referencing past data and adjust expressions according to treatment trends when continuing treatment or reapplying.

[0039] Furthermore, the storage unit 20 is used for statistical analysis of application trends and document quality at each treatment facility, and can also be used as a database for practitioner training and business improvement. From the perspective of ensuring security, the storage unit 20 preferably has a data management function based on access rights, preventing the risk of personal information and sensitive medical information being viewed illegally.

[0040] In the information processing device 10, each function other than the storage unit 20 can be realized by either a hardware configuration or a software configuration. For example, when these functions are realized by software, registration information for operating the information acquisition unit 12, natural language processing unit 14, output unit 16, and determination unit 18 is created on a server device actually configured with a CPU or MPU, RAM, ROM, etc., and the created registration information is downloaded via the Internet to application software stored in a recording medium such as RAM, ROM, or hard disk, and the application software is then operated.

[0041] Furthermore, the program according to the embodiment of the present invention is written as a series of instructions for causing a computer to execute each of the above-mentioned processing functions, and the storage unit 20 storing the program may include a form stored in a physical medium such as a hard disk drive, an optical disk, a flash memory, a semiconductor memory, or a server system that can distribute the program via a network. Based on operation input by the practitioner 4, the program sequentially executes each process, and a natural language sentence is automatically generated and output.

[0042] This program may also be provided in the form of a recording medium such as a DVD or USB memory, or in the form of a downloadable program via a communication path such as the Internet. This program is configured to be started by a GUI operation by the practitioner 4 (for example, pressing a start application button) and to be terminated by approval of the output statement or a save operation. The program may also be provided in the form of a web application, a standalone application, or a cloud-based SaaS service.

[0043] Next, the operation of the information processing device 10 configured as described above will be described. Fig. 2 is a flowchart showing the processing procedure for generating a natural language sentence to be attached to a medical care expense application form using the information processing device 10 according to an embodiment of the present invention. This flowchart shows in chronological order the processing executed by the information processing device 10 in response to an operation by the practitioner 4 to create the application document.

[0044] As shown in Fig. 2, the information processing method according to the embodiment of the present invention is composed of a series of processing steps initiated by the operation of the practitioner 4. Specifically, it includes a step (S2) of acquiring application-related information, a step (S4) of generating a natural language sentence based on the information, and a step (S10) of outputting the generated natural language sentence. These steps are executed automatically or semi-automatically by a computer in sequence, and are realized as a series of processing flows that assist the practitioner 4 in carrying out application procedures.

[0045] In addition, the information processing method according to an embodiment of the present invention is configured as a series of processing steps, including an acquisition step of acquiring application-related information, a generation step of generating a natural language sentence based on the information, an output step of outputting the generated natural language sentence, and, if necessary, a step of judging the natural language sentence and a step of prompting additional input based on missing information.

[0046] The processing method of the present invention is realized by a computer sequentially executing this series of steps. Furthermore, in each step, the processing results are sequentially passed from the previous step to the subsequent step according to the input data of the application-related information. For example, in the acquisition step, information about the treatment recipient entered through the GUI is recorded as structured data, in the generation step, a sentence structure is determined based on that data, and in the output step, the corresponding document is presented on a screen or as a PDF.

[0047] Next, the procedure for generating natural language sentences will be specifically explained based on the flowchart in Figure 2. First, the practitioner 4 starts creating medical care expense application documents for the patient to be treated. A dedicated creation screen is launched on an in-hospital terminal or the like, and the practitioner 4 starts this processing flow by operating a button such as "AI input."

[0048] Next, the information acquisition unit 12 starts up and collects application-related information entered by the practitioner 4, such as the name of the injury or illness, the location of the injury, the start date of treatment, symptoms, problematic behavior, obstacles to daily life, frequency of treatment, and reasons for continuing treatment. This information is entered using a specified GUI input form and processed in an internally structured form (S2). Here, "location of injury," "symptoms," and "reason for continuing treatment" are treated as keyword information that is particularly important for generating natural language sentences among the application-related information.

[0049] After step S2, the natural language processing unit 14 generates a natural language sentence based on the acquired application-related information (S4). In the generation process, instead of using a registered example sentence template as is, the sentence structure is dynamically changed depending on the content of the application-related information, so that a sentence appropriate for each treatment recipient is generated.

[0050] After step S4, the determination unit 18 analyzes the generated natural language sentence and determines whether or not the information required by the law is included (S6). This determination process is performed automatically against a checklist based on an administrative notice or the like.

[0051] If it is determined in step S6 that there are insufficient entries, an entry field corresponding to the relevant item is presented to the practitioner 4, and additional entry is requested (S8). It is also possible to configure the GUI screen to display an explanation of the missing items and example sentences.

[0052] This determination process by the information processing device 10 is not limited to a configuration executed on a local terminal installed in the treatment facility, but may be realized as a service deployed on a cloud server. This allows for the provision of centrally updated determination logic and templates to treatment facilities nationwide, making it possible to quickly respond to system revisions, etc.

[0053] In step S8, when the practitioner 4 completes the additional input, the natural language processing unit 14 operates again to regenerate an updated natural language sentence that reflects the input content. This regenerated sentence is configured in a form that resolves the deficiencies of the previous sentence.

[0054] If it is determined in step S6 that the entered information is sufficient, the output unit 16 displays the final natural language sentence on the screen (S10). The practitioner 4 can check the sentence and manually correct it as necessary. If there are no problems with the content, the generated sentence is recorded in the memory unit 20 by operating a save button or the like, and is output as part of the medical care expense application documents.

[0055] As described above, the information processing device 10 according to the present invention can efficiently create natural language sentences with improved compliance with the system while reducing the burden on the practitioner 4. This can eliminate the problems of text creation depending on the individual and incomplete / ambiguous descriptions, which have been problems in the past, and can raise the overall quality of medical care expense application documents.

[0056] Furthermore, the system automatically checks whether the generated documents meet the system requirements, and provides real-time input support as needed, thereby guiding practitioners 4 to ensure that they complete all entries in accordance with the screening criteria, which has the effect of significantly reducing the incidence of returns and non-payment.

[0057] In addition, this invention integrates a series of business processes, such as input support, document generation, component checking, and history storage, into a system, which not only standardizes and streamlines business operations but also reduces variations in the quality of documentation between practitioners.

[0058] Furthermore, by utilizing the accumulation of treatment history and sentence generation history, it will be possible to continuously improve the model and provide feedback on example descriptions, and in the future it is expected that a knowledge base that aggregates the know-how of each treatment center will be built. This could contribute to improving the transparency of system operation and accelerating the review process. [Explanation of symbols]

[0059] 2 Network, 4 Practitioner, 10 Information Processing Device, 12 Information Acquisition Unit, 14 Natural Language Processing Unit, 16 Output Unit, 18 Judgment Unit, 20 Memory Unit.

Claims

1. An information processing device that generates natural language sentences based on application-related information of a person to be treated in order to support the preparation of medical care expense application documents under an insurance system, an information acquisition unit that acquires the application-related information input by the practitioner; a natural language processing unit that generates the natural language sentence that is individual for each treatment recipient based on the application-related information; a determination unit that determines whether the natural language sentence generated by the natural language processing unit includes predetermined aspects including the location of injury, symptoms, and reasons for needing continuation of treatment based on an administrative notice required for applying for medical expenses; an input assistance unit that, when the determination unit determines that there is a missing item among the predetermined aspects, presents input items corresponding to the missing item to prompt the practitioner to make an additional input; an output unit that outputs the natural language sentence generated by the natural language processing unit; An information processing device comprising:

2. An information processing method for generating natural language sentences based on application-related information of a person to be treated in order to support the preparation of medical care expense application documents under an insurance system, comprising: acquiring the application-related information input by a practitioner; generating the natural language sentence that is individual to each treatment recipient based on the application-related information; a step of determining whether the natural language sentence includes predetermined aspects including the location of injury, symptoms, and reasons for needing continuation of treatment based on an administrative notice required for applying for medical expenses; a step of prompting the practitioner to input additional information by presenting input items corresponding to the missing items when it is determined that there are missing items among the predetermined aspects; a step of outputting the natural language sentence generated by the generating step; An information processing method comprising:

3. A program for generating natural language sentences based on application-related information of a person to be treated in order to support the preparation of medical care expense application documents under an insurance system, On the computer, A process of acquiring the application-related information input by a practitioner; A process of generating the natural language sentence that is individual for each treatment recipient based on the application-related information; A process of determining whether the natural language sentence contains predetermined aspects including the location of injury, symptoms, and reasons for needing continued treatment based on an administrative notice required for applying for medical expenses; a process of, when it is determined that there is a missing item among the predetermined aspects, presenting an input item corresponding to the missing item to prompt the practitioner to make an additional input; a process of outputting the natural language sentence obtained by the generation process; A program characterized by executing the following.

Images