Treatment planning, treatment evaluation, and treatment instruction creation support system

Abstract

We provide a treatment planning, treatment evaluation, and treatment instruction creation support system that uses additional learning from case data for which diagnosis and treatment have been completed in a large-scale language model. [Solution] The system includes a case data recording means for recording case data containing case feature data and treatment content data for patients whose diagnosis has been confirmed and whose treatment has been completed, a case learning means for adding pairs of case feature data and treatment content data for each case to a large-scale language model (LLM) as supervised learning data, and a treatment content data suggestion means for suggesting treatment content data from the case feature data of patients for whom treatment is to be initiated.

Description

[Technical Field]

[0001] The present invention relates to a system for supporting the creation of treatment plans, treatment evaluations, and treatment instructions using additional learning in a large-scale language model using case data for which diagnosis and treatment have been completed. [Background technology]

[0002] In the medical field, determining an accurate diagnosis of a disease based on the symptoms reported by the patient, the physical findings (symptoms and findings) obtained by the doctor during the examination, and the results of tests such as blood tests and radiological tests, and prescribing effective treatment are fundamental issues in medical care.

[0003] Machine learning has made remarkable progress in recent years, with large-scale language models (LLMs) becoming increasingly widespread. Neural network parameters, numbering in the tens of billions to trillions, are trained on massive amounts of data, amounting to terabytes, to handle tasks such as translation, speech and image recognition, and text summarization. Generative AI, which generates images, music, and documents based on queries, questions, or prompts, is also being put to practical use. It has become known that performance improves by scaling models (scaling laws), and the scale of models is rapidly increasing. By additionally training large amounts of clinical data into large-scale language models and quantifying the relationships between items using attention mechanisms, it is now possible to make diagnoses based on actual clinical symptoms, findings, and test data (Patent Document 2).

[0004] As the range of applications of large-scale language models expands, knowledge data from various fields is required. Furthermore, with social conditions and technological advances in each field, there is a need to constantly incorporate the latest knowledge data. However, training large amounts of data requires large-scale computational resources, enormous amounts of electricity, and costs. For this reason, large-scale language models cannot be reconstructed frequently. When large-scale language models are queried, they can generate answers that are not based on facts, even if they do not have the necessary knowledge stored internally. This phenomenon, known as hallucination, has limited their widespread use in fields such as medicine, where errors can lead to serious accidents.

[0005] To utilize large-scale language models in the field, it is necessary to acquire additional information specific to the field, and there are currently two approaches being taken. One is called additional learning (fine tuning) or transfer learning, in which additional information is used to train part of the output layer of an existing large-scale language model (fine tuning) or only the final layer (transfer learning), to create a specialized large-scale language model. This is highly useful as it results in a domain-specific large-scale language model.

[0006] Another approach is the RAG Retrieval-Augmented Generation (RAG) approach. The large-scale language model itself is not modified. Additional information is stored in a separate database (the RAG database). The information necessary to solve a question is searched for in the RAG database, and the obtained information is added to the question to obtain an answer from the large-scale language model. While there is a large amount of potentially additional data, the additional information required to answer a given question is limited. Therefore, only this limited additional information is searched for in the RAG database, and the extracted additional information (context) is added to the question in the question input box of the large-scale language model. This approach does not involve any learning changes to the large-scale language model itself; it simply adds context information to the question input box, resulting in low costs and technical hurdles. Furthermore, it offers excellent responsiveness, allowing for the immediate reflection of urgent information, such as information on emergency drug side effects (Non-Patent Document 1, Non-Patent Document 2).

[0007] In the present invention, we propose a treatment planning, treatment evaluation, and treatment instruction creation support system that uses additional learning, which is the first approach mentioned above, and which becomes possible once a certain amount of case data has been accumulated.

[0008] Prior art documents relevant to this application include the following: [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Patent No. 6792750 "Diagnosis Support System" [Patent Document 2] Patent No. 7592232 "Diagnosis and prescription support system using large-scale language models" [Patent Document 3] Patent application no. 2024-226197 "Large-scale language model using search expansion generation" [Patent Document 4] Patent No. 7441391 "Electronic medical record system using large-scale language models" [Patent Document 5] Patent No. 6902693 "Document Creation System"

[0010] [Non-Patent Document 1] https: / / ja.wikipedia.org / wiki / %E5%A4%A7%E8%A6%8F%E6%A8%A1%E8%A8%80%E8%AA%9E%E3%83%A2%E3%83%87%E3%83%AB [Non-patent document 2] https: / / www.idnet.co.jp / column / page_308.html Summary of the Invention [Problem to be solved by the invention]

[0011] In clinical settings, not only are patients' records and test orders required, but a large amount of documentation is also required, including treatment plans for doctors, nurses, rehabilitation specialists, and registered dietitians, as well as recording and evaluating the results of implementing those plans. The time and effort required to create documentation related to plans, evaluations, and treatment instructions accounts for at least 30% of staff time, causing increased fatigue. Planning and evaluation patterns differ for each medical staff member, including doctors, and uniform planning and evaluation does not easily reflect the planning and evaluation patterns of each staff member, so in the field, it is time-consuming to customize plans and evaluations individually.

[0012] The present invention has been made to solve these conventional problems, and its purpose is to extract (1) the patient attributes, symptoms, physical findings, and test data that characterize the case data (case characteristic data), and (2) the actually created treatment plan, evaluation, and treatment instruction content (treatment content data) from the medical record that records the entire case of a patient whose treatment has been completed, into a case data recording means, and to perform supervised additional learning on a large-scale language model, thereby enabling the smooth creation of appropriate treatment plans, evaluations, and treatment instruction content for individual new patients that are tailored to the treatment patterns of medical staff. [Means for solving the problem]

[0013] As a means for achieving the above-mentioned objective, the treatment plan, treatment evaluation, and treatment instruction creation support system described in claim 1 is characterized by comprising: (1) a case data recording means for recording case data of patients whose diagnosis has been confirmed and whose treatment has been completed; (2) the case data consists of case feature data including at least one of case attributes, symptoms, physical findings, and test data that characterize the case, and treatment content data including at least one of treatment plan, evaluation, treatment instruction content, diagnosis, treatment results, and attending physician created based on the judgment of the therapist; (3) a case learning means for additionally learning pairs of the case feature data and the treatment content data into a large-scale language model (LLM) as supervised learning data for each case; and (4) a treatment content data proposing means for proposing the treatment content data from the case feature data of the target patient for whom treatment is to be started.

[0014] The treatment plan, treatment evaluation, and treatment instruction creation support system described in claim 2 is characterized in that, in the treatment plan, treatment evaluation, and treatment instruction creation support system described in claim 1, it is equipped with a case data extraction means for extracting some or all of the case data from the electronic medical record.

[0015] The treatment plan, treatment evaluation, and treatment instruction creation support system described in claim 3 is characterized in that, in the treatment plan, treatment evaluation, and treatment instruction creation support system described in claim 1, it is provided with an electronic medical record input means for converting the proposed treatment plan, treatment evaluation, and treatment instruction content into an electronic medical record format and inputting it. [Effects of the Invention]

[0016] The treatment plan, treatment evaluation and treatment instruction creation support system according to claim 1 includes a case data recording means, which records case data of patients whose diagnosis has been confirmed and whose treatment has been completed. The case data consists of case characteristic data, which includes at least one of case attributes, symptoms, physical findings, and test data that characterize the case, and treatment content data, which includes at least one of the treatment plan, evaluation, treatment instructions, diagnosis, treatment results, and attending physician created based on the therapist's judgment. Since the case learning means is provided, pairs of the case feature data and the treatment content data for each case are additionally learned as supervised learning data in a large-scale language model (LLM). The medical treatment content data proposing means is provided, and proposes medical treatment content data from the case feature data of the target patient for whom treatment is to be started.

[0017] The treatment plan, treatment evaluation and treatment instruction creation support system according to claim 2 is provided with a case data extraction means, which extracts part or all of the case data from the electronic medical record.

[0018] The treatment plan, treatment evaluation and treatment instruction creation support system according to claim 3 includes an electronic medical record input means, so that the proposed treatment plan, treatment evaluation and treatment instruction contents are converted into the format of the electronic medical record and input. [Brief explanation of the drawings]

[0019] [Figure 1] 1 illustrates an embodiment of a network configuration according to the present invention. [Figure 2]This is an example of a user interface for a large-scale language model (LLM). [Figure 3] 1 shows an example of a case data group recorded in a case database. [Figure 4] 1 shows a schematic configuration of case data. [Figure 5] This is an example of an inpatient treatment plan. [Figure 6] This is an example of a nursing plan. [Figure 7] This is an example of a risk assessment document. [Figure 8] This shows the process of extracting case data from electronic medical records using a case data extraction means, recording it in a case database, and then performing supervised learning on a large-scale language model using case feature data as questions and treatment details as answers. [Figure 9] The system inputs the case characteristic data of the case for which a treatment suggestion is desired into the question box of the large-scale language model, and receives treatment suggestion data from the answer box. The proposed treatment suggestion data is converted into an electronic medical record format and entered into the electronic medical record. DETAILED DESCRIPTION OF THE INVENTION

[0020] FIG. 1 shows a typical system configuration of the present invention. Large-scale language models consist of huge amounts of data, numerous CPUs (Central Processing Units), GPUs (Graphics Processing Units), and high-speed networks connecting them, so they are built on huge servers such as cloud data centers and provided via the web. In hospitals and other facilities, many PC terminals are connected via a LAN (Local Area Network) connected to the web. There are also servers within the hospital that operate electronic medical records. In recent years, there has been an increasing number of cases where people use mobile devices such as smartphones and tablets to access cloud services such as large-scale language models and electronic medical records while moving around the hospital or from outside the hospital.

[0021] Servers, terminals, and mobile devices all consist of memory for recording programs and data, recording media such as hard disks that permanently record said programs and data as needed, a CPU that reads and processes said programs and data, a GPU that performs high-speed parallel processing as needed, and a communication module. As cloud services become more stable and cheaper, there are increasing cases of people migrating some or all of their electronic medical records to the cloud. Conversely, there is also a movement underway to move some or all of the large-scale language models to servers or terminals installed within hospitals with increased processing capabilities and storage capacity (edge ​​computing). Furthermore, the development of small-scale language models, which limit the number of parameters in the language model, is also progressing. Note that, although they are small-scale, they are still sufficiently large compared to the time before the advent of large-scale language models, and any embodiment, including this form, is included in the present invention.

[0022] Figure 2 shows an example of a user interface for a large-scale language model (LLM). LLMs are currently being rapidly developed, and many models have been developed, including ChatG PT (a registered trademark of OpenAI), Bard, LaMDA (a registered trademark of Google), and LLaMA (a registered trademark of Meta). Any LLM is included in the scope of this invention. Naturally, the user interface will differ, but typically, as shown in Figure 2, it consists of a box (prompt input box) for inputting prompts to instruct and inquire of the LLM, a box (answer display box) for displaying the answers to those prompts, and a box (usage history box) for displaying the history of prompts and answers as a usage log.

[0023] Recently, in addition to using LLMs alone as described above, there has been an increase in cases where LLMs themselves have APIs (Application Programming Interfaces) and LLM functions are used from external software. In this case, prompts, answers, and history are input and output to and from external software via API, so the display format is controlled by the external software regardless of the image in Figure 2.

[0024] In large-scale language models, to represent a certain vocabulary, a one-hot vector is used, which is a long vector consisting of zeros with the same number of dimensions as the number of types of vocabulary used, with a single 1 placed in the position corresponding to the vocabulary in question. All the vocabulary in a large amount of literature is replaced with vectors of this format, and deep learning is used to determine the associations (attention) between each vocabulary vector.In response to a query (prompt), vocabulary that is likely to appear next to the query and the answer sentences that have already been generated is generated and added one by one to create an answer sentence.

[0025] To utilize a large-scale language model in a specific field, additional information on that specific field is required. This is called additional learning (fine tuning) or transfer learning, and involves using additional information to train part of the output layer of an existing large-scale language model (fine tuning) or only the final layer (transfer learning), resulting in a specialized large-scale language model. Large-scale language models are already trained using large amounts of data on common phenomena, so the goal can be achieved by adding a relatively small amount of training only as needed for a specific situation. The amount of training data used for additional information varies depending on the target site. The amount of training data added needs to be increased until a practical level of accuracy is achieved. Furthermore, because sensitive information such as personal information and descriptions of medical conditions contained in the additional information is used in learning, there is a risk that it may be referenced outside the organization. It is useful to explicitly declare that the LLM should not be referenced outside the organization, or to use a paid version of a large-scale language model that guarantees that the learning content will not be referenced outside the organization, or even to use a large-scale language model that has been independently built within the medical institution.

[0026] FIG. 3 shows an example of the configuration of case data recorded in the case database. The case data includes case characteristic data that characterize the case, such as case attributes, symptoms, physical findings, and test results, as well as treatment content data such as the diagnosis obtained from the case characteristic data, the treatment plan that was planned and used, the contents of evaluation documents, treatment procedures such as medication, surgical treatment, and rehabilitation, the final treatment results, and the name of the treating physician. The case feature data is independent and not influenced by the intentions of the therapist, whereas the treatment content data reflects the judgment and intentions of the therapist. "Case attributes" are used to identify cases at the medical institution, such as patient ID, age, and gender. Depending on the disease, age and gender may be important. "Symptoms" is a list of the patient's subjective complaints. "Physical findings" are objective physical findings noticed by doctors and medical staff, such as body temperature and swelling of joints, tenderness, etc. This information is mainly obtained by doctors and nurses in the examination room.

[0027] "Test Results" shows the results of tests performed using testing machines, such as blood test results such as elevated CRP or positive rheumatoid reactions such as RAHA, X-ray results such as bone erosion, etc. For images such as X-rays, MRIs, and ultrasounds, if LLM is unable to recognize the image properly, you can enter a description of the image. "Diagnosis" is the name of the disease confirmed by a doctor, such as rheumatoid arthritis. "Plans and evaluation documents" are the group of contents written in medical treatment plans prepared by doctors, nursing care plans prepared by nurses, rehabilitation plans prepared by rehabilitation staff, and evaluation documents that record the results of treatment. "Treatment" refers to treatments given for the disease, such as administered medication, surgery, rehabilitation, etc. "Treatment outcome" refers to the extent to which the patient's condition improved as a result of the treatment. The "physician in charge" is the so-called primary physician, who is responsible for the treatment of the patient. If necessary, the person in charge of nursing or rehabilitation may also be added as an item.

[0028] Figure 4 summarizes Figure 3. While case characteristic data is a group of independent items that are not influenced by the therapist's intentions, treatment content data is a group of items that reflect the therapist's intentions and environment. In terms of multiple regression analysis, case characteristic data is the explanatory variable and treatment content data is the target variable. An object of the present invention is to predict and propose the content of each item constituting treatment content data using data of each item of case feature data. In multivariate analysis such as multiple regression analysis, all data for each item of case characteristic data must be available without any missing data. However, in the present invention, the purpose can be achieved even with incomplete data by using feature vectors and the ATTENTION mechanism. In principle, case data should be based on actual cases, but in the case of rare diseases with low incidence, it is useful to deliberately record typical data from fictitious cases of the rare disease, as this can suggest the possibility of the rare disease during the diagnostic process. The notation format can be plain text, but JSON, XML, or Markdown formats that clearly show the logical structure are preferable.

[0029] Figure 5 is an example of an inpatient treatment plan. Figure 6 is an example of a nursing care plan. Figure 7 is an example of a risk assessment document. There are over several hundred different types of plans, evaluation documents, and treatment instruction documents created in the medical field. Various surcharges are set out in medical fees, but as a calculation requirement for these surcharges, it is mandatory to produce documentation proving that the relevant act was performed. This has resulted in a huge amount of documentation work.

[0030] Figure 8 shows the flow of extracting case data from electronic medical records using a case data extraction means, recording it in a case data recording means, and then performing supervised additional learning on a large-scale language model using case feature data as questions and treatment details as answers. In this case, although it is possible to manually input each item of data, it is more preferable to extract it from the electronic medical record using a technique such as that shown in Patent Document 5. The case data of the extracted treatment-completed cases is recorded in a case database (case data recording means). Here, the case data is extracted from the electronic medical records, but if the medical institution uses paper medical records, the case data may be manually input / created in the case data recording means.

[0031] In Figure 9, case feature data for a case for which a treatment suggestion is desired is entered into the prompt input box of the large-scale language model, and treatment suggestion data is received from the response display box. The proposed treatment content data is converted into electronic medical record format and entered into the electronic medical record. As mentioned above, it is not necessary for all data items in the case characteristic data to be complete without any missing data. Even if there are missing items, questions can be asked at any stage during the treatment, which is useful in clinical practice. Here, various types of treatment plans, evaluation documents, and treatment instruction documents may be proposed all at once, or proposals may be received for each individual document by utilizing the usage history frame (treatment content data proposal means). Although Figure 9 explains the process in accordance with the image in Figure 2, in reality, it is also possible to directly input case characteristic data and output treatment content data suggestions using the API. Furthermore, the document text of the proposed treatment plan, evaluation document, and treatment instruction document can be converted into the format of each document in the electronic medical record, and document creation or treatment prescription orders can be issued in the electronic medical record, which will further increase its usefulness.

[0032] The items constituting the case characteristic data may be any combination of items that represent the characteristics of the case. For example, items such as the surgical method performed and the attending physician are usually included in the treatment content data, but depending on the purpose, it may be more reasonable to include them in the case characteristic data and query them, such as to reflect the attending physician's preferred treatment content or to reflect changes in treatment content for each surgical procedure.

[0033] Furthermore, since there is a risk of information leakage if the content entered in the question box is used to train a large-scale language model, it is useful to explicitly declare in the prompt that learning is prohibited, or to use a paid version of a large-scale language model that is guaranteed not to be used for learning, or even to use a large-scale language model that has been independently constructed within the medical institution.

[0034] Although the embodiments have been described above, the specific configuration of the present invention is not limited to the above-described embodiments, and the present invention also includes design changes and the like within the scope of the gist of the invention. For example, although the explanation has been given mainly on medical facilities such as hospitals, a similar support system can also be operated in nursing care facilities regarding case characteristic data and nursing care plans for individuals requiring care.

Claims

1. (1) A case data recording means is provided for recording case data of patients whose diagnoses have been confirmed and whose treatment has been completed; (2) The case data includes case characteristic data including at least one of case attributes, symptoms, physical findings, and test data that characterize the case, and treatment content data including at least one of a treatment plan created by the therapist's judgment, evaluation, treatment instructions, diagnosis, treatment results, and attending physician, (3) A case learning means is provided for additionally learning a pair of the case feature data and the treatment content data for each case into a large-scale language model (LLM) as supervised learning data, (4) A treatment plan, treatment evaluation, and treatment instruction creation support system, comprising a treatment content data suggestion means for suggesting the treatment content data from the case characteristic data of a patient for whom treatment is to be started.

2. 2. The treatment planning, treatment evaluation and treatment instruction creation support system according to claim 1, further comprising a case data extraction means for extracting a part or all of the case data from the electronic medical record.

3. 2. The treatment plan, treatment evaluation, and treatment instruction creation support system according to claim 1, further comprising an electronic medical record input means for converting the proposed treatment plan, treatment evaluation, and treatment instruction contents into an electronic medical record format and inputting the converted contents.

Images