System and method

Abstract

Provided is a system capable of specifying a treatment method for a patient corresponding to patient information. Provided is a system capable of specifying an outcome that would be achieved if the treatment method is implemented for a patient corresponding to patient information.　The system comprises a method specification means for requesting the execution of inference using a language model with a prompt as input, the prompt including patient information and a treatment method selection standard, thereby specifying one or more treatment methods for a patient corresponding to acquired patient information. The system comprises an outcome specification means for requesting the execution of inference using a language model with a prompt as input, the prompt including patient information, a treatment method for a patient corresponding to the patient information, and a correspondence relationship between the patient information and an outcome that would be achieved if the treatment method is implemented for the patient corresponding to the patient information, thereby specifying an outcome that would be achieved if the treatment method is performed for the patient corresponding to the acquired patient information.

Description

Systems and methods

[0001] The present disclosure relates to a program, device, and system for supporting the formulation of a patient's treatment plan. More specifically, the present disclosure relates to a program, device, and system that generates a proposed treatment plan using a correspondence table (matrix) or the like based on a new algorithm, namely, treatment selection criteria such as treatment inapplicability criteria and / or inapplicability criteria, in addition to existing treatment decision-making algorithms. Even more specifically, the present disclosure relates to a program, device, and system that generates a proposed treatment plan using a large-scale language model (LLM) and retrieval-augmented generation (RAG) based on patient information. Furthermore, the present disclosure relates to a program, device, and system that supports outcome prediction to support the formulation of a patient's treatment plan.

[0002] Treatment for cancer (synonymous with malignant disease) is determined by the condition of the disease (mainly the type of cancer and the stage of progression) (existing treatment decision algorithm). The stage of progression is classified according to the TNM classification based on the extent of the primary tumor, the degree of lymph node metastasis, and the presence or absence of distant metastasis, and a rough algorithm has been created for treatment, choosing from surgery, radiation therapy, and chemotherapy, focusing on whether surgery is possible.

[0003] In actual clinical practice, the attending physician performs the TNM classification in their head and selects the treatment based on an algorithm stored in their memory. The treatment selected based on the limited knowledge and experience of each attending physician is heavily influenced by the physician's specialty and may not be the optimal treatment for the patient. For example, if the attending physician is an oncologist who specializes in pharmacological treatment, even one metastasis will often be classified as distant metastasis. Even if a single metastasis actually allows for curative resection, the patient will miss the opportunity for cure if the attending physician does not have this knowledge.

[0004] Furthermore, in current clinical practice, patients must choose the treatment selected and presented by their doctor, or decide whether or not to accept the treatment, without any knowledge of the treatment in question.

[0005] Meanwhile, advances in medical care have led to an increase in treatment options, making treatment selection more complex and exceeding the knowledge and experience of attending physicians. It has become possible to further subdivide disease conditions, and not only have the number of treatments (major categories) increased, but the treatments themselves have also been subdivided (subcategorization). For example, the major categories of cancer treatment were surgery, radiation therapy, and systemic chemotherapy, but with the addition of new medical treatments such as immunotherapy, IVR, photoimmunotherapy, and viral therapy, the number of treatments has increased from three to seven. Regarding subcategorization, for example, radiation therapy includes linac, more advanced high-precision X-rays, α-knife, brachytherapy, particle therapy (heavy ion therapy, proton therapy), BNCT, and theranostics, and the number of treatments is expected to continue to increase in the future.

[0006] Because each specialized treatment has different therapeutic effects and adverse events, subdividing the condition and selecting the appropriate treatment from among the many specialized treatments can improve survival rates, including a cure, and reduce adverse events. Furthermore, achieving a cure would also have a positive effect on solving social issues and the medical economy. Conversely, the current rough treatment selection based on the TNM classification by attending physicians misses opportunities for a cure, resulting in social losses. This applies not only to cancer (malignant diseases) but also to benign diseases. For example, diabetes, which is a common disease, is often treated by physicians other than specialists, but because attending physicians are unaware of the details of the treatment plan, progression cannot be halted and patients are forced to undergo dialysis.

[0007] Patent Document 1 discloses a treatment policy decision support device that supports the decision of a treatment policy for a patient based on medical treatment information indicating the details of the medical treatment performed on each of a plurality of cancer patients, and that includes a construction means that constructs a master database based on the medical treatment information, in which a plurality of treatment results corresponding to each of the plurality of patients and each including a cancer type and a TNM classification value are registered, and when a treatment result request including a desired cancer type and TNM classification value is received, the treatment policy decision support device detects from the master database constructed by the construction means one or more treatment results each including a cancer type and a TNM classification value corresponding to the desired cancer type and TNM classification value, and returns the detected one or more treatment results to the issuer of the treatment result request.

[0008] However, understanding the many different medical conditions, memorizing the pros and cons of each treatment, and their effects and adverse events, and making accurate treatment choices is difficult and cannot be expected of all doctors. This is because, for doctors who are extremely busy with their daily medical practice, the vertical division of medical care by department makes it difficult to catch up on knowledge outside of their own department, and it is also difficult to gain treatment experience. For example, it is difficult for internists to correctly determine whether surgery is possible or to select the radiation therapy with the highest probability of cure from the many available radiation treatments.

[0009] To avoid this situation, clinical practice guidelines have been created for each disease and condition, providing guidance on optimal treatment for each more specific condition. However, the amount of information that attending physicians must refer to in these guidelines is enormous, and there are limits to how much they can read through them all, memorize the optimal treatment for each disease and condition, and apply it to treatment. For example, the clinical practice guidelines that gastroenterologists must read, for cancer alone, include the Esophageal Cancer Clinical Practice Guidelines, Gastric Cancer Clinical Practice Guidelines, Liver Cancer Clinical Practice Guidelines, Pancreatic Cancer Clinical Practice Guidelines, Biliary Tract Cancer Clinical Practice Guidelines, Colorectal Cancer Clinical Practice Guidelines, Small Intestine Cancer Clinical Practice Guidelines, Peritoneal Dissemination Clinical Practice Guidelines, and Metastatic Liver Tumor Clinical Practice Guidelines. Furthermore, clinical practice guidelines are updated almost annually, limiting their ability to keep up with the latest treatment information.

[0010] This situation is not limited to cancer, but applies to all chronic diseases, including rheumatoid arthritis, collagen diseases, diabetes, and dyslipidemia. Rheumatoid arthritis and collagen diseases are also subdivided into specific conditions, and guidelines for optimal treatment according to the condition are presented in clinical guidelines. And, just like cancer treatment, the guidelines themselves are updated daily.

[0011] The ideal treatment selection is based on the premise that medical care places a significant burden on patients both physically and mentally, and therefore the principle is to provide neither too much nor too little treatment, and from a comprehensive list of treatment candidates updated daily in real time for each disease and pathological condition, effective and feasible treatment candidates are presented for each subdivided medical condition in accordance with the priority of each disease, and the content, effectiveness (treatment results, etc.) and safety (adverse events, etc.) of each treatment are presented, allowing the patient to understand the treatment and make a choice together with their doctor that is in line with their values. However, the current situation is as described in the background, and there is a large gap between this and the ideal.

[0012] Japanese Patent Application Laid-Open No. 2018-147274

[0013] Examples of the objectives of the present invention include, but are not limited to, the following: A first objective of the present invention is to provide a system capable of identifying a treatment method for a patient corresponding to patient information. A second objective of the present invention is to provide a system capable of identifying an outcome when the treatment method is administered to a patient corresponding to the patient information.

[0014] In light of the above-mentioned circumstances, the present inventor has invented a new program, device, and system to support the formulation of patient treatment plans by utilizing large-scale language models (LLMs), also known as generative AI, and up-to-date and comprehensive treatment information databases that are updated in real time, such as PubMed and clinical trial databases, as well as an AI / IT technology called Retrieval-Augmented Generation (RAG) that connects them.

[0015] The present disclosure provides a program, a device, and a system for supporting the formulation of a patient's treatment plan. The present disclosure provides the following as more specific examples. [Embodiment 1] A program for supporting the formulation of a patient's treatment plan, the program causing one or more computers to execute the steps of: acquiring patient information; generating a proposed treatment plan based on the patient information using a large-scale language model (LLM), wherein the proposed treatment plan includes a list of one or more available treatment methods selected based on treatment selection criteria (such as non-indication criteria and / or indication criteria) input as prompts to the LLM by Retrieval-Augmented Generation (RAG) and the patient information; and outputting the proposed treatment plan, wherein the proposed treatment plan further includes information including advantages and disadvantages of the selected one or more available treatment methods, and / or information including predicted information on the cure probability and / or outcome of the selected one or more available treatment methods. [Embodiment 2] A program for supporting the formulation of a patient treatment plan, the program causing one or more computers to execute the steps of acquiring patient information, generating a proposed treatment plan based on the patient information, and outputting the proposed treatment plan. [Embodiment 3] The program according to embodiment 2, wherein the step of generating a proposed treatment plan based on the patient information includes selecting one or more available treatment methods based on the patient information and predefined treatment selection criteria. [Embodiment 4] The program according to any one of embodiments 2 and 3, wherein the step of selecting one or more available treatment methods based on the patient information and predefined treatment selection criteria is performed using a large-scale language model (LLM). [Embodiment 5] The program according to embodiment 4, wherein the treatment selection criteria are input to the large-scale language model (LLM) as prompts. [Embodiment 6] The program according to embodiment 5, wherein the treatment selection criteria are input to the large-scale language model (LLM) using Retrieval-Augmented Generation (RAG). [Embodiment 7] The program according to embodiment 6, wherein the treatment selection criteria are stored in a database.[Embodiment 8] The program according to embodiment 7, wherein the treatment selection criteria are stored in a database in the form of a correspondence table that associates patient information with available treatment methods. [Embodiment 9] The program according to any one of embodiments 2 to 8, wherein the proposed treatment plan includes information including the advantages and disadvantages of one or more selected available treatment methods. [Embodiment 10] The program according to any one of embodiments 2 to 9, wherein the proposed treatment plan includes information including prediction information on the cure probability and / or outcome of one or more selected available treatment methods. [Embodiment 11] Patient information includes age, sex, disease name (disease name, primary name, condition name such as metastasis or recurrence, complication name), left hepatic vein invasion, middle hepatic vein invasion, left hepatic vein invasion, vascular invasion, bile duct invasion, gastrointestinal tract invasion, Glisson's sheath invasion, metastasis to other organs, low liver function, S1, FLR (normal liver), FLR (damaged liver), massive hepatectomy (tri-segment resection, etc.), massive hepatectomy (right hepatectomy), tri-segment branch, left hepatic vein invasion, middle hepatic vein invasion, right hepatic vein invasion, portal vein invasion, distance to gastrointestinal tract (mm), blood vessels The program according to any one of embodiments 1 to 10, comprising at least one piece of information selected from the group consisting of anatomical names and positional relationships such as infiltration, bile duct invasion, gastrointestinal tract invasion, blood vessel proximity, bile duct proximity, gastrointestinal tract proximity, and Glisson's capsule proximity, extent of progression, histological and pathological findings, genetic information such as genetic abnormalities and their degrees (including histological findings and gene expression levels), immunological findings such as cancer microenvironment, and degree of function. [Embodiment 12] The program according to any one of embodiments 1 to 11, wherein at least a part of the patient information is generated by automatic interpretation of images.[Embodiment 13] Available treatment methods include conventional surgery, advanced surgery, TSH stage 2 surgery, ALPPS, ALPTPS, venous surgery, robotic surgery, arthroscopic surgery, radiation therapy, external radiation therapy, particle beam therapy, heavy ion beam therapy, proton beam therapy, high-precision X-ray therapy, IMRT, stereotactic irradiation, Cyberknife, internal radiation therapy, brachytherapy, brachytherapy, theranostics, BNCT boron neutron capture therapy, IVR, ablation, radiofrequency ablation, microwave ablation, cryotherapy, intra-arterial embolization therapy, ethanol injection, HYF high-intensity focused ultrasound, irreversible electroporation therapy (IRE), photoimmunotherapy, viral therapy, chemotherapy, hormone therapy, molecular targeted drug, antibody-drug conjugate (ADC), immunotherapy, immune checkpoint inhibitor, fecal transplant therapy, cancer vaccine, CAR-T therapy, combined immunotherapy, gene therapy, stem cell transplant, mRNA vaccine, CRISPR / Cas9, hyperthermia, TS The program according to any one of embodiments 1 to 12, comprising at least one treatment method selected from the group consisting of H2 stage surgery, ALPPS, venous combined surgery, conventional RFA, high-precision RFA, TACE, drug therapy, anticancer drugs, immunosuppressants, investigational drugs, Chinese herbal medicine, transplantation (organ transplant, bone marrow transplant, etc.), nerve block, CART, dialysis, shunt (Denver shunt, etc.), nerve decompression, nerve transplant, nerve repair, stem cell therapy, stem cell transplant, inhalation, infusion, transdermal drug, transdermal absorption preparation, transdermal patch, transdermal absorption preparation (TTS), eye drops, intracavitary administration, continuous administration, and new surgical procedures, psychotherapy (cognitive behavioral therapy, CBT, etc.), physical therapy (electroconvulsive therapy: ECT, transcranial magnetic stimulation: TMS, transcranial direct current stimulation: tDCS, deep brain stimulation (DBS), spinal cord stimulation, etc.), diet therapy, exercise therapy, weight loss surgery, oriental medical treatment such as acupuncture and moxibustion, and antioxidants. [Embodiment 14] The program according to embodiment 2, wherein the step of generating a proposed treatment plan based on patient information is performed using a machine learning model that takes patient information as input and outputs available treatment methods.

[0016] The present invention also addresses the following problems: [1] A system comprising at least one computer device, the system comprising an acquisition means for acquiring patient information, and a method identification means for identifying one or more treatment methods for a patient corresponding to the acquired patient information based on the acquired patient information and a selection criterion for the treatment method; [2] The system according to [1], wherein the method identification means identifies a surgical procedure, a drug regimen, an immunotherapy or cell therapy regimen, or a radiation therapy prescription; [3] The system according to [1] or [2], wherein the method identification means receives as input a prompt including the acquired patient information and a selection criterion for the treatment method, and requests execution of inference using a language model, thereby identifying one or more treatment methods for a patient corresponding to the acquired patient information; [4] The system according to any of [1] to [3], wherein the method identification means identifies one or more treatment methods based on an evaluation criterion for the patient's condition; [5] The system according to [3] or [4], wherein the method identification means receives a prompt including treatment method selection criteria acquired from a knowledge base as an input and requests execution of inference using a language model, thereby identifying one or more treatment methods for a patient corresponding to the acquired patient information; [6] The system according to any of [3] to [5], wherein the method identification means receives a prompt including the acquired patient information as an input and requests execution of inference using a language model fine-tuned using information related to treatment method selection criteria, thereby identifying one or more treatment methods for a patient corresponding to the acquired patient information; [7] The system according to any of [3] to [6], wherein the method identification means receives a prompt including information related to the patient's medical record as an input and / or information related to images of the patient's inside the body as an input and requests execution of inference using a language model;[8] The system according to any one of [3] to [7], wherein the input means inputs a target outcome, and the method identification means inputs a prompt including patient information and the target outcome, and requests execution of inference using a language model, thereby identifying one or more treatment methods based on the target outcome for the patient corresponding to the acquired patient information; [9] The system according to any one of [1] to [8], wherein the patient information includes image information related to the body, and the selection criteria can be determined from the image information related to the body, and the method identification means identifies one or more treatment methods for the patient corresponding to the acquired patient information based on the acquired patient information and the selection criteria for the treatment methods;

[10] The system according to any one of [1] to [9], wherein the patient information includes information that can determine the distance between a first organ and a second organ in the patient's body, and the selection criteria for the treatment method include information related to the distance between the first organ and the second organ as a criterion for enabling selection of at least one treatment method;

[11] The system according to any one of [1] to

[10] , wherein the patient information includes information that can ascertain the number of tumors in the patient's body, the size of the tumors, the extent of tumor spread, the anatomical location of the tumors, whether the site where the tumor is present is a site where it is most likely to occur, or whether the lesion is localized, and the selection criteria for the treatment method include information that can ascertain the number of tumors in the patient's body, the size of the tumors, the extent of tumor spread, the anatomical location of the tumors, whether the site where the tumor is present is a site where it is most likely to occur, or whether the lesion is localized, as criteria for enabling selection of at least one treatment method;

[12] A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information; and a method identification means for identifying one or more treatment methods corresponding to the acquired patient information based on a trained model that uses patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data;

[13] The system according to

[12] , comprising an update means for updating the trained model by re-learning based on new patient information for machine learning and treatment methods suitable for the new patient information for machine learning;

[14] A system comprising at least one computer device, the system comprising: an acquisition means for acquiring patient information; an input means for inputting a target outcome; and a method identification means for identifying one or more treatment methods corresponding to the acquired patient information and the target outcome based on a trained model that is machine-learned using patient information for machine learning and an outcome resulting from administering the treatment method to a patient corresponding to the patient information for machine learning as input data, and the treatment method administered to the patient corresponding to the patient information for machine learning as output data;

[15] The system described in

[14] , comprising an update means for updating the trained model by re-learning based on new patient information for machine learning, the outcome resulting from administering the treatment method to a patient corresponding to the new patient information for machine learning, and the treatment method administered to the patient corresponding to the patient information for machine learning;

[16] The system described in any of [1] to

[15] , comprising a first information identification means for identifying information including the advantages and / or disadvantages of the treatment method based on the treatment method identified by the method identification means;

[17] The system according to any one of [1] to

[16] , comprising a second information identification means for identifying a survival rate, a cure probability, a recurrence rate, or a local control rate resulting from the implementation of a treatment method based on the treatment method identified by the method identification means;

[18] The system according to any one of [1] to

[17] , comprising a third information identification means for identifying a predicted outcome of the treatment method based on the treatment method identified by the method identification means;

[19] The system according to any one of [1] to

[18] , comprising an input means for inputting a patient's target outcome or a patient's risk tolerance, and a suitability identification means for identifying information that can grasp the degree to which the treatment method identified by the method identification means is suitable for the input target outcome or risk tolerance;

[20] The system according to any one of [1] to

[11] and

[16] to

[19] , wherein the selection criteria for the treatment method are generated by inputting a prompt including text information about the treatment method and requesting the execution of inference using a language model;

[21] The system according to any one of [1] to

[20] , wherein the method specifying means specifies only treatment methods that are covered by public health insurance;

[22] The system according to any one of [1] to

[21] , wherein the method specifying means can specify treatment methods that are not covered by public health insurance;

[23] The system according to any one of [1] to

[22] , wherein the system is provided with a missing information specifying means that specifies missing information when information necessary for determining the appropriateness of a treatment method is missing from the acquired patient information;

[24] A system comprising at least one computer device, the system comprising: an acquisition means that acquires patient information and a treatment method for a patient corresponding to the patient information; and an outcome specifying means that specifies a predicted outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting the execution of inference using a language model, using as input a prompt including the acquired patient information, the treatment method for the patient corresponding to the patient information, and a correspondence between the patient information and the outcome when the treatment method is administered to the patient corresponding to the patient information;

[25] A system including at least one computer device, the system comprising: an acquisition means for acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification means for identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by receiving as input a prompt including the acquired patient information and the treatment method for the patient corresponding to the patient information and requesting execution of inference using a fine-tuned language model using information regarding the correspondence between the patient information and an outcome when the treatment method is administered to the patient corresponding to the patient information;

[26] A system including at least one computer device, the system comprising: an acquisition means for acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification means for identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information, based on a trained model that uses patient information for machine learning and the treatment method for the patient corresponding to the patient information for machine learning as input data and the outcome when the treatment method is administered to the patient corresponding to the patient information for machine learning as output data;

[27] The system according to

[26] , comprising an update means for updating a trained model by relearning based on new patient information for machine learning, a treatment method for a patient corresponding to the patient information for machine learning, and an outcome when the treatment method is administered to a patient corresponding to the new patient information for machine learning;

[28] The system according to any of

[24] to

[27] , comprising a method specifying means for specifying one or more treatment methods for a patient corresponding to the acquired patient information based on the acquired patient information and a treatment method selection criterion, wherein the outcome specifying means specifies an outcome based on the specified treatment methods;

[29] The system according to any of

[24] to

[27] , comprising a method specifying means for specifying one or more treatment methods corresponding to the acquired patient information based on a trained model machine-learned using patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data, wherein the outcome specifying means specifies an outcome based on the specified treatment methods;

[30] The system according to any of [1] to

[29] , wherein the patient information includes information on images taken inside the patient's body;

[31] The system according to any one of [1] to

[30] , wherein the patient information includes text information;

[32] A method executed in a system including at least one computer device, comprising: an acquisition step of acquiring patient information; and a method specifying step of identifying one or more treatment methods for a patient corresponding to the acquired patient information by requesting execution of inference using a language model using the acquired patient information and a prompt including selection criteria for the treatment method as input;

[33] A method executed in a system including at least one computer device, comprising: an acquisition step of acquiring patient information; and a method specifying step of identifying one or more treatment methods for a patient corresponding to the acquired patient information by requesting execution of inference using a language model fine-tuned using information regarding selection criteria for the treatment method as input;

[34] A method executed in a system including at least one computer device, the method comprising: an acquisition step of acquiring patient information; and a method identification step of identifying one or more treatment methods corresponding to the acquired patient information based on a trained model that has been machine-learned using patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data;

[35] A method executed in a system including at least one computer device, the method comprising: an acquisition step of acquiring patient information; an input step of inputting a target outcome; and a method identification step of identifying one or more treatment methods corresponding to the acquired patient information and a target outcome based on a trained model that has been machine-learned using patient information for machine learning and an outcome as a result of administering the treatment method to a patient corresponding to the patient information for machine learning as input data and a treatment method administered to a patient corresponding to the patient information for machine learning as output data;

[36] A method executed in a system including at least one computer device, the method comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome determination step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting execution of inference using a language model, using as input the acquired patient information, the treatment method for the patient corresponding to the patient information, and a prompt including a correspondence between the patient information and an outcome when the treatment method is administered to the patient corresponding to the patient information;

[37] A method executed in a system including at least one computer device, the method comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome determination step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting execution of inference using a language model using information regarding a correspondence between the patient information and an outcome when the treatment method is administered to the patient corresponding to the patient information;

[38] The problem can be solved by a method executed in a system including at least one computer device, the method comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information, based on a trained model that uses the patient information for machine learning and the treatment method for the patient corresponding to the patient information as input data and outputs an outcome when the treatment method is administered to a patient corresponding to the patient information for machine learning.

[0017] Examples of the effects of the present invention include, but are not limited to, the following: A first effect of the present invention is to provide a system that can identify a treatment method for a patient corresponding to patient information. A second effect of the present invention is to provide a system that can identify an outcome when the treatment method is administered to a patient corresponding to the patient information.

[0018] 1 shows a schematic configuration of a computer according to an embodiment; FIG. 2 shows an example of formulating a treatment policy for a patient using RAG according to an embodiment; FIG. 3 shows an overview of treatment selection and outcome prediction by a treatment selection support program according to an embodiment; FIG. 4 is a block diagram showing a configuration of a system according to an embodiment; FIG. 5 is a diagram showing a flowchart of output processing according to an embodiment; FIG. 6 is a diagram showing a flowchart of output processing according to an embodiment;

[0019] The following describes embodiments of the present invention, but the present invention is not limited to the following embodiments as long as they do not violate the spirit of the present invention. The order of each process constituting the flowcharts described below is random as long as no contradictions or inconsistencies occur in the process content. Furthermore, it is possible to omit some of the processes constituting the flowcharts or add new processes to each process constituting the flowcharts as long as they do not violate the spirit of the present invention. Furthermore, the device that executes each process constituting the flowcharts can be changed to another device as long as it does not violate the spirit of the present invention. In this case, the process content can be changed so as not to cause contradictions or inconsistencies in the process content.

[0020] In one aspect, the present disclosure relates to a program for assisting in the formulation of a patient's treatment plan. The patient's disease may be, for example, but not limited to, a malignant disease such as cancer (synonymous with malignant disease; classified into carcinoma, sarcoma, and blood cancer), a benign disease such as rheumatism, collagen disease, diabetes, or dyslipidemia, or a general disease, primarily a chronic disease. Furthermore, the program is not limited to patients already suffering from a disease, but also includes patients at high risk of developing a disease and patients who have undergone treatment. The former includes, for example, IPMN (intraductal papillary mucinous neoplasm) in pancreatic cancer, hepatitis in liver cancer, and hyperglycemic states in diabetes. The latter includes, for example, post-surgical treatment for cancer. The patient may be, for example, a human cancer patient, such as a patient with any type of cancer, including breast cancer, lung cancer, prostate cancer, colon cancer, pancreatic cancer, gastric cancer, hepatocellular carcinoma, leukemia, skin cancer, uterine cancer, sarcoma, soft tissue sarcoma, lymphoma, head and neck cancer, biliary tract cancer, brain tumor, esophageal cancer, cancer of unknown primary, prostate cancer, oral cancer, or ovarian cancer, etc. Cancer patients may have metastases and may have primary as well as recurrent cancer.

[0021] In some embodiments, the present disclosure provides a program for assisting in the formulation of a patient treatment plan, the program causing one or more computers to execute the steps of acquiring patient information, generating a treatment plan proposal based on the patient information, and outputting the treatment plan proposal. The one or more computers do not need to be physically located in the same place, and multiple computers with different roles (e.g., an input / output terminal, an LLM server, a database server, etc.) may be connected via a network and function cooperatively.

[0022] As used herein, "treatment strategy" includes the selection of a treatment method to be applied to a patient. For example, in the case of a cancer patient, available treatment methods include conventional surgery, advanced surgery, TSH stage 2 surgery, ALPPS, ALPTPS, venous surgery, robotic surgery, arthroscopic surgery, radiation therapy, external beam radiation therapy, particle beam therapy, heavy ion beam therapy, proton beam therapy, high-precision X-ray therapy, IMRT, stereotactic radiotherapy, CyberKnife, internal radiation therapy, brachytherapy, brachytherapy, theranostics, BNCT (boron neutron capture therapy), IVR, ablation, radiofrequency ablation, microwave ablation, cryotherapy, intra-arterial embolization therapy, ethanol injection, hyphenated high-intensity focused ultrasound, irreversible electroporation (IRE), photoimmunotherapy, viral therapy, chemotherapy, hormone therapy, molecular targeted drugs, antibody-drug conjugates (ADC), immunotherapy, immune checkpoint inhibitors, fecal transplantation therapy, cancer vaccines, CAR-T therapy, combined immunotherapy, gene therapy, stem cell transplantation, mRNA Vaccines, CRISPR / Cas9, hyperthermia, TSH stage 2 surgery, ALPPS, combined venous surgery, regular RFA, high-precision RFA, TACE, drug therapy, anticancer drugs, immunosuppressants, investigational drugs, herbal medicines, transplants (organ transplants, bone marrow transplants, etc.), nerve blocks, CART, dialysis, shunts (Denver shunts, etc.), nerve decompression, nerve transplants, nerve repair, stem cell therapy, stem cell transplants, inhaled drugs, infusions, transdermal drugs, transdermal absorption preparations, transdermal These include, but are not limited to, patches, transdermal therapy (TTS), eye drops, intracavitary administration, continuous infusion, and new surgical techniques, psychological therapies (e.g., cognitive behavioral therapy (CBT)), physical therapies (e.g., electroconvulsive therapy (ECT), transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), deep brain stimulation (DBS), spinal cord stimulation (spinal cord stimulation), dietary therapy, exercise therapy, weight loss surgery, traditional Chinese medicine treatments such as acupuncture and moxibustion, antioxidants, etc. When multiple treatment methods are available, they can be combined and applied simultaneously or sequentially. As used herein, "treatment strategy" can also include a plan for how to apply available treatment methods.

[0023] As used herein, "patient information" includes multiple items related to the patient's physical characteristics and medical condition. For example, in the case of a cancer patient, patient information includes, but is not limited to, age, sex, disease name (disease name, primary name, condition name such as metastasis or recurrence, complication name), name of left hepatic vein invasion, middle hepatic vein invasion, left hepatic vein invasion, vascular invasion, bile duct invasion, gastrointestinal tract invasion, Glisson's sheath invasion, metastasis to other organs, low liver function, S1, FLR (normal liver), FLR (damaged liver), major hepatectomy (tri-segmentectomy etc.), major hepatectomy (right hepatectomy), tri-segment branch, left hepatic vein invasion, middle hepatic vein invasion, right hepatic vein invasion, portal vein invasion, distance from the gastrointestinal tract (mm), anatomical names and positional relationships such as vascular invasion, bile duct invasion, gastrointestinal tract invasion, proximity to blood vessels, proximity to the bile duct, proximity to the gastrointestinal tract, and proximity to Glisson's sheath, degree of progression, histological and pathological findings, genetic information such as genetic abnormalities and their degree (including histological findings, gene expression levels, etc.), immunological findings such as the cancer microenvironment, and information on the degree of function.

[0024] The patient information may be acquired, for example, by input from a keyboard, input via a communication line such as the Internet, or input from storage such as a hard disk. The patient information may be stored in a patient information database and acquired as needed by accessing the database. At least a portion of the patient information may be, for example, information in an electronic medical record. At least a portion of the patient information may also be acquired by character recognition of a handwritten medical record. Furthermore, at least a portion of the patient information may be generated by automatic interpretation on a computer from test images such as PET, CT, MRI, and X-ray images. AI-based image recognition technology may be used for the automatic interpretation.

[0025] In some embodiments, generating a proposed treatment plan based on patient information includes, for example, selecting one or more available treatments based on patient information and predefined treatment selection criteria (inapplicable criteria and / or eligible criteria and / or semi-inapplicable criteria) for a malignant disease (cancer). The selected available treatments may be one or more, and a list of available treatments may be created and included in the proposed treatment plan. The treatments displayed in the list may be presented in order of the outcome prioritized for each disease. For example, for malignant diseases (cancer), the outcome of cure is important, and treatments may be presented in order of the probability of cure. The probability of cure may be calculated from the patient's condition information. A correspondence table between progression and treatment established for each disease may also be used. For example, for a malignant disease, a correspondence table between stage (progression) and treatment is used. For example, for a benign disease such as rheumatoid arthritis, a phase (progression) is determined based on the response (effect) to treatment obtained from the course of treatment, and treatments are presented according to that phase. In the case of asthma, for example, a correspondence table is used that uses symptoms and treatment progress as criteria for treatment selection. For benign diseases, original correspondence tables that match subdivided symptoms with treatments can also be created and used.

[0026] As used herein, the term "indication criteria" refers to criteria for excluding a particular treatment method from treatment options based on one or more items of patient information. For example, the indices for pancreatic cancer surgery include tumor contact or infiltration of the SMA, CA, or CHA, or contact or infiltration of the SMV / PV at less than 180 degrees and no obstruction, and contact or infiltration of the SMA or CA at 180 degrees or more, or contact or infiltration of the proper hepatic artery or CA. Therefore, if the patient information indicates contact of the SMA / PHA at 180 degrees or more, pancreatic cancer surgery will not be selected as an available treatment method.

[0027] In addition, for colorectal cancer patients with liver metastasis, if the patient information indicates left hepatic vein invasion, conventional surgery is not applicable, but advanced surgery (S1, etc.), TSH stage 2 surgery, and ALPPS surgery are available.

[0028] As used herein, "eligibility criteria" refers to criteria for including a particular treatment method in treatment options based on one or more items of patient information.

[0029] As used herein, the term "quasi-indication criteria" refers to criteria that, based on multiple items of patient information (e.g., two or more, three or more, four or more, or five or more items), can exclude a particular treatment method from treatment options or defer that decision until certain conditions are met.

[0030] As used herein, the term "treatment selection criteria" refers to a patient's condition or condition for implementing treatment. As used herein, "treatment selection criteria" is intended to encompass "non-labeled criteria," "labeled criteria," and "quasi-non-labeled criteria."

[0031] In some embodiments, the step of selecting one or more available treatment methods based on patient information and predefined treatment selection criteria (non-labeled criteria and / or labeled criteria and / or semi-non-labeled criteria) may be performed using a large language model (LLM).

[0032] In the context of this disclosure, examples of large-scale language models (LLMs) include GPT (Generative Pre-trained Transformer), BERT (Bidirectional Encoder Representations from Transformers), T5 (Text-to-Text Transformer), and Llama (Large Language Model Meta AI). As used herein, "large-scale language model (LLM)" refers to an advanced AI model that can train on vast amounts of text data and understand and generate human language. Such models, also known as generative AI, are widely used in the field of natural language processing (NLP) and have demonstrated high performance in a wide range of tasks, including text summarization, translation, question answering, text generation, and sentiment analysis. LLMs may run on cloud servers, for example.

[0033] LLMs are typically trained using large datasets containing billions of words collected from books, articles, websites, and other sources available on the internet. The training process uses a multi-layered neural network architecture called a Transformer. Transformers are particularly adept at effectively processing context-rich information and understanding interlinguistic dependencies. A distinctive feature of this architecture is its self-attention mechanism, which weights each word based on its context, taking into account its relevance to other words.

[0034] Furthermore, these models can be fine-tuned for specific tasks and adapted to advanced language-related tasks suited to specific applications or industries. For example, in the medical field, LLMs fine-tuned using specialized text data can be effectively used for complex tasks requiring specialized knowledge. Large-scale language models can serve as a fundamental technology for providing innovative solutions in a wide range of fields, from general-purpose natural language processing tasks to applications specialized in specific fields.

[0035] In some embodiments, the large-scale language model (LLM) used in the program of the present disclosure may be fine-tuned specifically for formulating a patient's treatment plan. Fine-tuning may involve using treatment selection criteria and algorithms, such as inappropriate criteria, inappropriate criteria, and / or semi-inappropriate criteria, extracted from the latest treatment guidelines related to individual diseases, databases of past clinical cases, and literature databases such as PubMed, or organized and reconstructed information obtained from these sources. Existing, widely used disease state / progression correspondence tables can be used, and these correspondence tables can be used as is. However, all of these correspondence tables can also be reconstructed to use treatment selection criteria, such as inappropriate criteria, inappropriate criteria, or semi-inappropriate criteria. By utilizing such information (datasets) to retrain the model, performance on the desired task of selecting available treatment methods can be significantly improved.

[0036] In another embodiment, the LLM used in the program of the present disclosure may have undergone transfer learning for the task of formulating a patient's treatment plan. Transfer learning involves using the weights of a model previously trained on a broad dataset as initial values ​​and then retraining the model to adapt to a new task. Specifically, a model already trained in general natural language processing or medical fields can be fine-tuned using data related to specific treatment plan formulation, enabling rapid and efficient adaptation to new tasks. Transfer learning is often considered a part of fine-tuning, which optimizes a model for a specific task. This technique allows a model to quickly adapt to specialized and advanced tasks while making the most of previously accumulated knowledge. As a result, the LLM used in the program of the present disclosure can function as a powerful support tool for formulating optimal treatment plans tailored to individual patient conditions.

[0037] As used herein, a "prompt" for a large-scale language model (LLM) refers to a textual input provided to the model. This prompt serves as a starting point or instruction for the model to generate a response and is an important factor in determining the content and format of the generated text. Prompts can serve a variety of purposes, such as directing the language model to focus on a specific topic or task and providing appropriate direction so that the model can generate appropriate and relevant information within a given context.

[0038] The input prompt determines the context of the generated text and specifies the scope of information the model will reference. Prompts can also instruct the model to generate text in a specific format or style. For example, prompts can be used to instruct the model to generate answers in the form of questions or to create summaries of the given text. Furthermore, prompt design is crucial for the model to generate more accurate responses. The content and structure of the prompt significantly affect what information the model prioritizes and the format of the response it generates. Therefore, appropriately designed prompts can make the LLM's output more specific and precise, thereby improving the quality of the model's responses.

[0039] The program disclosed herein can generate prompts to be input into a large-scale language model (LLM) based on patient information. The prompts may directly use the patient information entered by the user, or may modify the input information appropriately. The technique of designing prompts to be input into an LLM so that the model generates desired output is known as "prompt engineering," and this is an important technique for improving the accuracy of model responses.

[0040] In some embodiments, the prompts can include multiple task-relevant examples on which the model can make inferences. This technique, known as "few-shot learning," can effectively improve a model's performance by presenting a small number of examples. Few-shot learning is particularly useful in data-limited situations, improving the model's ability to quickly adapt to new tasks.

[0041] Furthermore, in some embodiments, prompts entered into the LLM can be generated using Retrieval-Augmented Generation (RAG). RAG is a method for generating more accurate and detailed responses by searching and extracting external data (e.g., treatment selection criteria related to patient information (e.g., inappropriate criteria, appropriate criteria, and semi-inappropriate criteria)) from the RAG database in real time and combining the search results with the original question (prompt). This process enables the LLM to incorporate the latest relevant information and propose highly accurate treatment plans. When obtaining necessary information using RAG, information highly relevant to the query, such as patient information, can be extracted from the RAG database based on, for example, cosine similarity. Using RAG enables real-time information updates to the LLM, reducing the LLM's training cost while generating responses that reflect the latest medical information. Furthermore, the fine-tuning described above can also be achieved using the RAG function. Further improvement in accuracy can be expected by storing the information obtained during fine-tuning in the RAG database. 2 shows an example of formulating a patient's treatment plan using the RAG in the program of the present disclosure. By utilizing the RAG, the latest external information can be obtained in a timely manner and questions can be posed to the LLM based on that information, thereby realizing more advanced medical support. The RAG database may be located, for example, on a cloud server, physically separated from the computer used as the user terminal.

[0042] In the context of the present disclosure, examples of prompts entered into the LLM include the following: (Prompt 1) "In the case of pancreatic cancer surgery, the criteria for exclusion are tumor contact with or invasion of the SMA, CA, or CHA, or contact with or invasion of the SMV / PV less than 180 degrees, without obstruction, contact with or invasion of the SMA or CA more than 180 degrees, or contact with or invasion of the proper hepatic artery or CA." (Prompt 2) "The following is information about the patient's condition: - Invasion of the duodenum - Invasion of the lower part of the common bile duct - Almost complete occlusion of the superior mesenteric vein (SMV) - Contact with the superior mesenteric artery (SMA) / hepatic artery (PHA) more than 180 degrees - Invasion of the horizontal limb of the duodenum. Can surgery be performed on this patient?"

[0043] An example of an LLM's response to the above input is as follows: "Surgery is not considered appropriate for this patient because the superior mesenteric vein (SMV) is nearly occluded, there is more than 180-degree tumor contact with the superior mesenteric artery (SMA) and hepatic artery (PHA), and there is infiltration into the duodenum and lower common bile duct. Due to these conditions, surgery is extremely risky and is not recommended. Non-surgical treatments such as chemotherapy and radiation therapy should be considered as alternatives."

[0044] In the above example, the user provides the non-indication criteria as a prompt. However, the prompt may be generated using a RAG based on patient information, and the LLM may learn treatment selection criteria, such as non-indication criteria. That is, in some embodiments, the treatment selection criteria (non-indication criteria and / or indication criteria) may be input to the large-scale language model (LLM) as a prompt. Also, in some embodiments, the treatment selection criteria (non-indication criteria and / or indication criteria and / or semi-indication criteria) may be input to the large-scale language model (LLM) using Retrieval-Augmented Generation (RAG). In some embodiments, the treatment selection criteria (non-indication criteria, non-indication criteria, semi-indication criteria) are stored in a database. The treatment selection criteria (non-indication criteria, non-indication criteria, or semi-indication criteria) stored in the database may be used in the RAG. This is also true when using an existing treatment decision algorithm stored in the database.

[0045] In some embodiments, the treatment selection criteria (non-indicated criteria and / or indicated criteria and / or semi-indicated criteria) are stored in a database in the form of a correspondence table (also referred to as a correspondence table or matrix) that associates patient information with available treatments. The correspondence table may be automatically updated and / or expanded by the LLM, if necessary. In some embodiments, when the treatment selection criteria (non-indicated criteria and / or indicated criteria and / or semi-indicated criteria) are stored in a database in the form of a correspondence table that associates patient information with available treatments, available treatments may be selected using a standard matching algorithm (searching for correspondence between patient information and available treatments) without using the LLM. For example, a treatment may be selected by matching patient information, such as test results, with indicated or indicated treatments in the correspondence table to create a list of applicable treatments. For example, treatments may be selected by excluding treatments that are unavailable depending on the patient's condition, etc., from a list of all currently available treatments, and then selecting the treatments that remain on the list. Thus, in some embodiments, a program according to the present disclosure may be a program that executes steps of acquiring patient (clinical) information, generating a proposed treatment plan based on the patient information, and outputting the proposed treatment plan, wherein the step of generating a proposed treatment plan based on the patient information includes selecting one or more available treatment methods by referring to a correspondence table that associates the patient information with available treatment methods, even when using an existing treatment decision algorithm stored in a database.

[0046] Outputting the Proposed Treatment Course In some embodiments, outputting the proposed treatment course may include, for example, outputting to a display directly or indirectly connected to the computer, audio output to a speaker, or outputting to a printer or storage connected to the computer.

[0047] In some embodiments, the output treatment plan proposal may include information on the advantages (e.g., effectiveness) and disadvantages (e.g., adverse events) of one or more selected available treatment methods. Information on the advantages and disadvantages of each treatment method may be stored in a database used by the RAG or in a separate, independent database. Information on the advantages and disadvantages of each treatment method may be obtained by the LLM or by a regular database query. When multiple available treatment methods are selected, they may be prioritized based on their advantages and disadvantages.

[0048] In some embodiments, the output treatment plan proposal may include information including the cure probability of one or more selected available treatment methods. Information including the outcome priority for each treatment method by disease (e.g., cure probability for cancer) may be stored in a database used by the RAG or in a separate, independent database. Information including the cure probability for each treatment method may be obtained by the LLM or by a normal database query. When multiple available treatment methods are selected, they may be prioritized (by disease) based on the cure probability.

[0049] In some embodiments, the output treatment plan proposal may include predicted outcome information such as recurrence. The predicted outcome information may be automatically calculated by the LLM by obtaining paper information from a paper database such as PubMed and comparing it with patient information according to an algorithm in the paper information.

[0050] In some embodiments, a program for assisting in the formulation of a patient treatment strategy according to the present disclosure may be a program that causes one or more computers to execute the steps of: acquiring patient information; generating a proposed treatment strategy based on the patient information using a large-scale language model (LLM), wherein the proposed treatment strategy includes a list of one or more available treatments selected based on the patient information and treatment selection criteria (e.g., non-indication criteria and / or indication criteria) input as prompts into the LLM by Retrieval-Augmented Generation (RAG); and outputting the proposed treatment strategy, wherein the proposed treatment strategy further includes information including advantages and disadvantages of the selected one or more available treatments and / or information including predicted information on the cure probability and / or outcome of the selected one or more available treatments. In some embodiments, the program according to the present disclosure may further define a treatment name and treatment selection criteria for each treatment and store them in a database in advance. In some embodiments, the program of the present disclosure may provide for a step of collecting, by the RAG, the treatment name and treatment selection criteria for each predefined treatment from a pre-created DB or from a DB that is updated on a rolling basis.

[0051] In some embodiments, the step of generating a proposed treatment plan based on patient information may be performed using a large-scale language model (LLM) or any other suitable machine learning model (e.g., neural network, random forest, etc.) that takes patient information as input and outputs possible treatments.

[0052] Figure 3 shows an overview of treatment selection and outcome prediction using the treatment selection support program disclosed herein. A correspondence table with patient information can be used for the algorithms of treatment selection and outcome / result prediction (such as recurrence prediction). In other words, the correspondence table can be used to stratify patients by treatment. Outcome prediction enables precise treatment selection. For example, the risk of recurrence (recurrence score) after breast cancer surgery can be calculated using outcome prediction, and the degree of need for adjuvant therapy can be presented based on treatment information on the additional effect of adjuvant chemotherapy obtained from a paper and the recurrence score.

[0053] There may be two types of treatment selection tables. One is the traditional table that maps progression / phase to treatment, which can be an approach based on the patient's "condition" (patient information). The other lists all treatments available for each disease and matches the selection criteria (inapplicability criteria, inapplicability criteria) with the patient's condition (patient information). Treatment options can be selected based on treatments that do not violate the inapplicability criteria or are applicable. Diseases (illnesses) can be broadly classified as malignant (cancer) and benign, but the feasibility of local therapy for malignant diseases may be determined primarily by the inapplicability criteria. Outcome predictions can be calculated using a table that maps patient information and factors / items related to outcomes / results proven in papers, etc., to outcomes / results, or using a formula.

[0054] 1) Creation of an original database (DB) (1) The original DB (1) contains, for example, a correspondence table for treatment selection or a correspondence table and calculation formula for predicting treatment outcome results. This treatment selection or outcome prediction algorithm (correspondence table, calculation formula) may be created using an external medical database using RAG to obtain the data necessary to create the algorithm, and then stored in the original DB (1) using LLM or the like.

[0055] 2) Creation of an original database (DB) (2) External medical DBs can include DBs such as PubMed for acquiring the latest paper information, DBs containing treatment information that incorporates doctor consensus, such as guidelines, and other DBs such as clinical trial information. The original DB (2) stores, for example, treatment information for each treatment (treatment content / explanation, treatment effects, adverse events, etc.), and the treatment information may be acquired from the external DB using RAG, organized using LLM, and stored in DB (2).

[0056] 3) Organizing patient information Patient information such as disease names (for example, colon cancer, liver metastasis, lung metastasis, etc.) obtained from medical information forms, image information (such as CT or MRI image interpretation findings), treatment progress information (date, treatment details, and findings), test information (blood tests, etc.), pathology information, genetic information, symptoms, findings, etc. can be organized using LLM, and the patient information can then be organized in accordance with (adjusted to) a correspondence table and converted into "condition data."

[0057] 4) Presenting treatment options using a comparison table and calculating outcomes and results. By comparing the "condition data" (patient information organized in 3 above) with the comparison table obtained from DB (1) using RAG using a checklist or LLM, treatment can be selected or outcome prediction can be performed.

[0058] 4a) In the case of cancer treatment selection: An example of an LLM prompt is, "Access DB (1), obtain the correspondence table from the disease name, compare the disease state with the comparison table, and if any of the exclusion criteria are met, remove the treatment from the treatment options. If the inclusion criteria are not met, remove it. Display the remaining treatments under these conditions in the same order as in the comparison table."

[0059] 4b) In the case of outcome prediction (risk of recurrence) When a correspondence table is used, the LLM prompt may be, for example, "Access DB (1), obtain the comparison table for the desired prediction, and calculate the predicted value based on the condition and the comparison table" (for example, stage 2 surgery for colon cancer liver metastasis, risk of recurrence between stage 1 and stage 2). In the case of a formula, the prompt may be, for example, "Extract the items and their values ​​necessary for prediction from the condition, substitute them into the formula, and calculate the predicted value" (for example, an equation for weighting the expression levels of cancer-related genes in a paper published in NEJM and used in OncotypeDX (registered trademark) to calculate the recurrence rate of breast cancer).

[0060] 5) Acquisition and display of treatment information Information about treatment options can be acquired from the original DB (2) using RAG and LLM. An example of an LLM prompt for acquisition is, "Please access DB (2) and acquire and display treatment information for each presented treatment."

[0061] 6) Original DB (3): Patient database Patient information and output results for each patient may be saved in one file, and the file may be stored in the original DB (3).

[0062] 7) Original DB (3): Utilization of patient database. This can be completely anonymized and used as real-world data to provide feedback and improve the accuracy of the software disclosed herein.

[0063] Those skilled in the art will understand that the features described in this specification as characteristics of the program according to the present disclosure can also be applied as appropriate to the devices and systems described below.

[0064] Apparatus for assisting in the formulation of a patient's treatment plan In one aspect, the present disclosure relates to an apparatus for assisting in the formulation of a patient's treatment plan.

[0065] In some embodiments, the present disclosure relates to an apparatus for assisting in formulating a treatment plan for a patient, the apparatus including a patient information acquisition unit that acquires patient information, a treatment plan proposal generation unit that generates a treatment plan proposal based on the patient information, and a treatment plan proposal output unit that outputs the treatment plan proposal.

[0066] In some embodiments, an apparatus according to the present disclosure may include a processor for performing various processes and a memory coupled to the processor. In some embodiments, an apparatus according to the present disclosure may be implemented by executing a program according to the present disclosure on a general-purpose computer. In some embodiments, an apparatus according to the present disclosure includes a processor coupled to a memory device that stores a computer program according to the present disclosure and that can execute instructions of the program.

[0067] In some embodiments, the device according to the present disclosure includes a means for inputting data, such as a keyboard, a mouse, or the like.

[0068] In some embodiments, the device according to the present disclosure includes a central processing unit (CPU) connected to a keyboard, mouse, etc. for inputting data, connected to a hard disk, flash memory, etc. as a storage unit, and connected to memory (storage means) such as ROM, RAM, etc.

[0069] Examples of means for outputting data including prediction results include a monitor, a printer, etc. Other examples of output means include means for storing data in storage means such as a hard disk, flash memory, ROM, RAM, etc.

[0070] The apparatus according to the present disclosure may include a means for storing a program for assisting in the formulation of a patient treatment plan according to the present disclosure. Examples of the means for storing the program include a hard disk, a flash memory, and the like. Such a storage means may be connected via a communication line. That is, the apparatus according to the present disclosure may be part of a system obtained by connecting via a communication line to a device including the means for storing the program.

[0071] Fig. 1 is a schematic diagram showing an exemplary embodiment of an apparatus according to the present disclosure. In Fig. 1, reference numeral 100 denotes a computer, which includes a control unit 101, a storage unit 102, a peripheral device I / F unit 103, an input unit 104, a display unit 105, and a communication unit 106, all of which are connected by a bus 110. Note that this configuration is merely an example, and various other configurations may be adopted as appropriate.

[0072] The control unit 101 is composed of a central processing unit (CPM), read-only memory (ROM), random access memory (RAM), etc. The CPU loads programs stored in the memory unit 102, ROM, recording media, etc. into a work memory area on the RAM and executes them, driving and controlling each device connected via the bus 110 and realizing the processing performed by the computer. The ROM is a non-volatile memory that stores programs, data, etc., such as the boot program and BIOS of the computer 100. The RAM is a volatile memory that temporarily stores programs, data, etc. loaded from the memory unit 102, ROM, recording media, etc., and also provides a work area used by the control unit 101 when performing various processing. The memory unit 102 is, for example, a hard disk drive (HDD) that stores the programs executed by the control unit 101 and various other data.

[0073] The peripheral device I / F (interface) unit 103 is a port for connecting the computer 100 to peripheral devices. The peripheral device I / F unit 103 is configured using USB, Bluetooth, IEEE 1394, RS-232C, etc. The connection to the peripheral devices may be wired or wireless. The input unit 104 has input devices such as a keyboard, a pointing device such as a mouse, and a numeric keypad, and issues operational instructions, operation instructions, data input, etc. to the computer 100. The display unit 105 is a logic circuit or device driver for displaying videos, images, etc. on a display device such as a liquid crystal panel. The input unit 104 and the display unit 105 can also be configured integrally as a touch display.

[0074] The communication unit 106 has a communication control device, a communication port, etc., and is a wired or wireless communication interface that mediates communication with the communication network 120. The bus 110 is a communication path that mediates the exchange of control signals, data signals, etc. between each device. The communication network 120 can further be connected to an external server 130 and a database (or net storage) 140.

[0075] In some embodiments, the device of the present disclosure may further include a prompt generation unit that generates prompts using RAG (Retrieval-Augmented Generation), an LLM input / output unit that inputs prompts to the LLM to obtain output, an LLM operation unit that operates the LLM, a treatment method information acquisition unit that acquires information about available treatment methods, and a treatment method information storage unit that stores information about available treatment methods.

[0076] SYSTEM FOR SUPPORTING DEVELOPMENT OF PATIENT TREATMENT POLICY The present disclosure, in one aspect, relates to a system for supporting the formulation of a patient treatment policy. The system according to the present disclosure may be comprised of one or more computers. The multiple computers may be connected via a communication network such as the Internet, and some of the computers may be configured on a cloud server. An exemplary configuration of the system according to the present disclosure includes a system comprising a client terminal that mainly handles input / output and a server that mainly handles calculation processing.

[0077] In some embodiments, the system of the present disclosure relates to a system for supporting the formulation of a treatment plan for a patient, the system including one or more computers, the one or more computers including a memory and a processor connected to the memory, the processor performing steps of acquiring patient information input by a user, generating a treatment plan proposal based on the patient information, and outputting the treatment plan proposal. In some embodiments, the memory or storage device of the computer included in the system of the present disclosure stores the program of the present disclosure, and when the program is executed, each of the above steps is performed.

[0078] In some embodiments, the system of the present disclosure may further perform the steps of generating prompts using Retrieval-Augmented Generation (RAG), inputting the prompts into the LLM to obtain output, running the LLM, obtaining information about available treatments, and storing information about available treatments.

[0079] In one aspect, the present disclosure relates to a non-transitory computer-readable recording medium storing the program of the present disclosure. Examples of computer-readable recording media include, but are not limited to, hard disk drives (HDDs), solid-state drives (SSDs), flash memories (such as USB memories and SD cards), optical disks (such as CDs, DVDs, and Blu-ray (registered trademark) disks), magnetic tapes, floppy disks, and cloud storage.

[0080] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, some potentially preferred methods and materials are now described. All publications mentioned herein are incorporated by reference to disclose and describe the methods and / or materials in connection with which the publications are cited. It is understood that the present disclosure supersedes the disclosure of the incorporated publication in the case of a conflict.

[0081] Where a range of values ​​is described, unless the context clearly dictates otherwise, it is understood that each intervening value, to the tenth of the unit of the lower limit, between the upper and lower limits of that range is also specifically disclosed. Each smaller range between any stated or intervening value in a stated range and any other stated or intervening value within that stated range is also encompassed within the disclosure. The upper and lower limits of these smaller ranges may independently be included or excluded, and each range including either, either, or both limits in the smaller ranges is also encompassed within the invention, but the specifically excluded limit in the stated range is reserved. When a stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the invention. The term "about" with respect to a numerical value means within 5%.

[0082] The embodiments described herein are intended to be merely exemplary, and those skilled in the art will be able to make numerous variations and modifications without departing from the spirit of the present invention. Furthermore, certain variations and modifications may produce less than optimal results, but still provide satisfactory results. All such variations and modifications are intended to be within the scope of the present invention as defined by the appended claims. Furthermore, any combination of the components disclosed herein, and any transformation of the present disclosure into a method, apparatus, system, computer program, data structure, recording medium, etc., are also valid aspects of the present disclosure. Therefore, details described regarding the method of the present disclosure may also be applied to a system, computer program, data structure, recording medium, etc.

[0083] The present disclosure will be further understood by reference to the following examples. These examples are provided solely to illustrate the claimed disclosure; the disclosure is not limited in scope by the exemplified embodiments, which are intended only as illustrations of single aspects of the disclosure. Any functionally equivalent methods are within the scope of the present disclosure. Various modifications of the present disclosure, in addition to those described herein, will be apparent to those skilled in the art from the foregoing description. Such modifications are intended to fall within the scope of the appended claims.

[0084] <Fifteenth and Sixteenth Embodiments> The matters described in the fifteenth and sixteenth embodiments can also be applied to the first to fourteenth embodiments. Fig. 4 is a block diagram showing the configuration of a system according to the embodiments. The system 10 includes at least one computer device. The system 10 may be configured, for example, from one computer device (information processing device) (standalone type), from one or more server devices and one or more terminal devices (client-server type), or from multiple terminal devices (peer-to-peer type).

[0085] The system 10 may include a user terminal 200 and a server device 100. The server device 100 corresponds to the computer 100 described above. The server device 100 may also function in a distributed manner across multiple computer devices. For example, instead of the server device 100, a distributed ledger technology such as a blockchain may be used.

[0086] The user terminal 200 and the server device 100 are communicably connected to each other via a communication network 120. The number of user terminals 200 is not particularly limited as long as it is one or more, and there may be more than one.

[0087] The system 10 may include a language model server 300. The server device 100 and the language model server 300 are connected to each other via a communication network 120 so that they can communicate with each other.

[0088] (User Terminal) The user terminal 200 is a terminal operated by a user of the system 10. Here, the user may be any person who uses the system 10. Examples of the user terminal 200 include a conventional mobile phone, a tablet terminal, a smartphone, and a personal computer.

[0089] The user terminal 200 includes a control unit, a storage unit, an input unit, a display unit, and a communication unit, all of which are connected by a bus.

[0090] The control unit is composed of a CPU. The control unit executes programs stored in the memory unit and controls the user terminal 200. The memory unit is a storage medium for saving programs and data. The memory unit also includes RAM. The RAM is the work area of ​​the control unit. The control unit performs arithmetic processing based on the programs and data read from RAM and data input via the input unit. The program (program product) may be stored on a recording medium such as a CD-ROM.

[0091] The display unit has a display screen. The control unit outputs a video signal for displaying an image on the display screen according to the result of the arithmetic processing. Note that the display screen of the display unit may be a touch panel equipped with a touch sensor. In this case, the touch panel functions as the input unit.

[0092] The communication unit can be connected to the communication network 120 wirelessly or via a wire, and can transmit and receive data to and from other computer devices via the communication network 120. Data received via the communication unit is loaded into RAM, and the control unit performs arithmetic processing.

[0093] 3 is a block diagram showing the hardware configuration of a server device according to an embodiment. The server device 100 includes at least a control unit 101, a storage unit 102, and a communication unit 106, which are connected to each other via an internal bus.

[0094] The control unit 101 is composed of a CPU, executes programs stored in the storage unit 102, and controls the server device 100. The control unit 101 also has an internal timer that measures time. The storage unit 102 includes a RAM and a main memory. The RAM is a work area for the control unit 101. The main memory is a storage area for saving programs and data. The main memory functions as a recording medium that stores the programs. The control unit 101 reads out the programs and data stored in the storage unit 102, loads them into the RAM, and performs program execution processing based on information received from the user terminal 200 or the language model server 300. The program (program product) may be stored on a recording medium such as a CD-ROM.

[0095] (Language Model) The language model server 300 is equipped with a language model. The language model is a model that learns statistical features obtained from a large amount of text data. The language model is preferably a large-scale language model that is trained with a large data set and has a huge number of parameters. The language model may be a generative AI. Generative AI refers to artificial intelligence that automatically generates new content such as text, images, audio, video, and program code based on input data. The generative AI may be trained from a large data set using a neural network model, particularly a deep learning model such as a Transformer architecture, and generate content through probabilistic generation or pattern prediction.

[0096] The language model server 300 includes an inference engine. The inference engine loads model parameters into VRAM. When a prompt is input, the inference engine converts the prompt string into tokens, executes forward calculations of the Transformer, and sequentially generates tokens based on the obtained probability distribution. The generated output is output as an answer to the source of the prompt.

[0097] Here, the language model is provided outside the server device 100, but it may also be provided inside the server device 100.

[0098] (System) The system 10 can identify one or more treatment methods for a patient corresponding to the patient information based on patient information and information related to treatment method selection criteria. The system 10 can identify one or more treatment methods for a patient corresponding to the patient information by requesting execution of inference using a language model using a prompt including the patient information and information related to the treatment method selection criteria as input. Note that the system 10 can also identify one or more treatment methods without using a language model.

[0099] The treatment method identified by the system 10 is not particularly limited. The system 10 can identify a surgical procedure as the treatment method. The system 10 can also identify a medication regimen as the treatment method. The medication regimen refers to a medication plan including the types of medications to be administered to a patient, the combination of medications, dosage, administration route, administration interval, and / or number of cycles.

[0100] The system 10 can specify an immunotherapy or cell therapy regimen. An immunotherapy regimen refers to an immunotherapy plan including the type of drug, dose, route of administration, administration interval, and / or number of cycles. A cell therapy regimen refers to a cell therapy plan including cell collection, cell preparation production, pretreatment, administration, monitoring, or toxicity management. Furthermore, the system 10 can specify a radiation therapy prescription. A radiation therapy prescription refers to a radiation therapy plan including whether the radiation is external or internal, radiation quality, total dose, single dose, dose rate, number of fractions, radiation frequency, and / or radiation technique.

[0101] Selection criteria are set for each treatment method, which are the criteria for enabling the use of that treatment method. If the patient information meets the selection criteria corresponding to one treatment method, that treatment method can be used. If the patient information does not meet the selection criteria corresponding to that treatment method, that treatment method cannot be used. By making such a determination for each treatment method, it is possible to identify an available treatment method from multiple treatment methods. This identification of a treatment method can be performed in step S25, which will be described later, without using a machine-learned trained model, or can be performed in step S9 using a language model.

[0102] Alternatively, instead of the above-described method, a treatment method or selection criteria similar to the patient information can be identified based on the similarity between a patient information vector obtained by vectorizing patient information and a treatment method vector obtained by vectorizing treatment methods (or a selection criterion vector obtained by vectorizing selection criteria, or a vector obtained by vectorizing information consisting of treatment methods and selection criteria). The similarity is determined based on Euclidean distance or cosine similarity. A treatment method with a similarity equal to or greater than a threshold can be identified as a treatment method suitable for the patient information. Identifying a treatment method suitable for the patient information based on the similarity between the patient information and the treatment method and / or selection criteria makes it possible to identify which treatment method is more suitable for the patient. This treatment method identification can be performed in step S25 (described below) without using a machine-learned trained model, or can be performed in step S9 using a language model. When a language model is used, the prompt may include information regarding the patient information, the treatment method, and / or selection criteria, and information requesting the identification of a treatment method based on the similarity therebetween.

[0103] Fifteenth Preferred Embodiment The matters described in the fifteenth preferred embodiment can also be applied to the first to fourteenth and sixteenth preferred embodiments.

[0104] The prompt is generated in the server device 100. The generation of the prompt is executed in step S7 or S12, which will be described later.

[0105] The patient information included in the prompt is not particularly limited as long as it is information about the patient, but is preferably information that influences a decision on a treatment method. The patient information may include, for example, basic attributes such as age and gender, information about the disease or condition such as the name of the disease (e.g., hepatocellular carcinoma, colorectal cancer, diabetes, etc.), the primary lesion (e.g., liver, pancreas, stomach, etc.), the presence or absence of metastasis (e.g., lung metastasis, liver metastasis), the presence or absence of recurrence, complications (e.g., diabetes mellitus, hypertension), tumor infiltration (e.g., left hepatic vein invasion, middle hepatic vein invasion, right hepatic vein invasion, vascular invasion, bile duct invasion, gastrointestinal tract invasion, Glisson's sheath invasion), information about anatomical location such as proximity to blood vessels, bile duct, and gastrointestinal tract, information about organs or bodily functions such as low liver function, FLR (e.g., normal liver / damaged liver), and whether or not a large liver resection is required (e.g., trisegmentectomy, right hepatectomy, etc.), genetic and molecular biological information such as pathological and histological findings, genetic abnormalities and their severity and gene expression levels, and immunological findings such as the cancer microenvironment.

[0106] Furthermore, the patient information included in the prompt may include information regarding the distance between organ A in the patient's body and another organ B different from organ A. Furthermore, the patient information in the image file format used in the inference process, which will be described later, may include information regarding the distance between organ A in the patient's body and another organ B different from organ A (information that allows the distance between organ A and organ B to be determined). For example, if a criterion regarding the distance between organ A and organ B (e.g., a criterion that treatment method X can be used when the distance between organ A and organ B is equal to or greater than ○ cm) is included as a criterion for enabling selection of treatment method X, it becomes possible to make an appropriate determination as to whether or not to identify treatment method X as a treatment method for the patient corresponding to the patient information.

[0107] The patient information included in the prompt may also include information regarding the number of tumors in the patient's body, tumor size, tumor extent, anatomical location of the tumors, whether the tumor site is a predisposing site, or whether the lesion is localized. Furthermore, the patient information in the image file format used in the inference process described below may also include information regarding the number of tumors in the patient's body, tumor size, tumor extent, anatomical location of the tumors, whether the tumor site is a predisposing site, or whether the lesion is localized (information that can grasp this information). For example, if the criteria for selecting treatment method Y include criteria regarding the number of tumors, tumor size, tumor extent, anatomical location of tumors, and whether the tumor site is a predisposing site (e.g., criteria such as "treatment method Y can be used if the number of tumors is ○ or less"), an appropriate determination can be made regarding whether treatment method Y should be identified as a treatment method for the patient corresponding to the patient information.

[0108] The patient information used in the inference process using the language model may be transmitted from the user terminal 200 to the server device 100, or may be registered in advance in a patient information database. The patient information database may be stored in the storage unit 102 of the server device 100, or may be provided outside the server device 100.

[0109] The patient information can be obtained from a patient information database. The patient information may be text information or image information. The patient information also includes text information obtained by analyzing image information using artificial intelligence capable of image analysis or multimodal AI.

[0110] The text information included in the prompt may be text information entered using a keyboard or the like, or may be information obtained by OCR scanning of an electronic or handwritten medical record. The text information may include, for example, findings obtained with the naked eye or findings obtained with a microscope. The patient information used in the inference process may include data in an electronic file format in addition to the text information entered as a prompt. The data in an electronic file format includes image data in an image file format and data in a PDF file format. The image data may be images captured using a tomographic imaging method such as PET, CT, or MRI, data related to X-ray images, ultrasound images, or images captured with an endoscope.

[0111] If the selection criteria can be determined from image information about the body, an electronic file such as image data about the patient's body and a prompt containing patient information can be used as input to perform inference using a language model (e.g., multimodal AI) and identify one or more treatment methods based on the selection criteria for the treatment methods.

[0112] The selection criteria for the treatment method included in the prompt are conditions for safely or effectively implementing the treatment method. Preferably, the prompt includes selection criteria for each treatment method, which are criteria for enabling the treatment method. Also, preferably, the prompt includes selection criteria for multiple treatment methods corresponding to the patient's disease or condition included in the patient information.

[0113] The selection criteria for the treatment method include non-approved criteria, applicable criteria, or semi-inapplicable criteria. The treatment selection criteria (non-approved criteria and / or applicable criteria and / or semi-inapplicable criteria) may be stored in the treatment method database in the form of a correspondence table that associates patient information with treatment methods applicable to the patient corresponding to the patient information.

[0114] The criteria for selecting a treatment method may include consensus, diagnostic criteria, or treatable traits. Consensus refers to, for example, recommendations or opinions reached by the joint discussion of multiple experts. Diagnostic criteria refer to defined items and thresholds for diagnosing and classifying a disease. Treatable traits refer to patient characteristics that are measurable, clinically significant, and modifiable by treatment.

[0115] As described above, the selection criteria for the treatment method may include criteria related to the distance between an organ in the patient's body and another organ different from this organ when the treatment method is performed. As described above, the selection criteria for the treatment method may also include criteria related to the number of tumors in the patient's body, the size of the tumors, the extent of tumor spread, the anatomical location of the tumors, whether the site where the tumor is present is a predisposing site, or whether the lesion is localized.

[0116] Based on patient information about a large number of real patients, the treatment methods administered to these patients, and the resulting outcomes, it is also possible to identify selection criteria corresponding to treatment methods through inference using a language model. The identified selection criteria can be registered in a treatment method database.

[0117] The selection criteria for the treatment method included in the prompt may be those registered in advance in a treatment method database. The treatment method database may be stored in the storage unit 102 of the server device 100 or may be provided externally to the server device 100.

[0118] The selection criteria for the treatment methods to be included in the prompt can be determined by searching for and identifying treatment methods that are highly similar to the patient information from among the treatment methods registered in the treatment method database. The treatment method database stores selection criteria for each treatment method, which are the criteria for whether that treatment method can be used.

[0119] The search is performed by identifying treatment methods that are highly similar to the patient information based on the Euclidean distance or cosine similarity between a patient information vector obtained by vectorizing at least a portion of the patient information (for example, information about the patient's disease or condition) and a treatment method vector obtained by vectorizing the treatment methods stored in the treatment method database (or information about the characteristics and details of the treatment methods). This makes it possible to include in the prompt selection criteria for one or more treatment methods that correspond to the patient's disease or condition included in the patient information.

[0120] Furthermore, in the treatment method database, tags may be assigned to each treatment method or each selection criterion (for example, the name of the treatment method or keywords in the selection criteria may be assigned as tags). Selection criteria for treatment methods corresponding to the patient information can be searched for by tracing the tags based on keywords included in the patient information. This makes it possible to include in the prompt one or more selection criteria for treatment methods corresponding to the patient's disease or condition included in the patient information. Note that tags may be assigned to each treatment method or each selection criterion using inference using a language model.

[0121] The server device 100 generates a prompt including, for example, patient information acquired from the patient information database or the user terminal 200 and selection criteria for a treatment method acquired from the treatment method database. The generated prompt may include information requesting that a treatment method suitable for the patient corresponding to the patient information be output.

[0122] The treatment method selection criteria included in the prompt may be generated using a language model. Furthermore, the selection criteria generated using the language model may be stored in the treatment method database. Text information regarding the treatment method may be transmitted from a management terminal operated by an administrator or operator of the system 10 to the server device 100, or the text information regarding the treatment method may be transmitted from the user terminal 200 to the server device 100. The server device 100 generates a prompt including text information regarding the treatment method, and, based on this prompt, can output the treatment method selection criteria as an answer using a language model. The output treatment method selection criteria are stored in the treatment method database.

[0123] Alternatively, selection criteria for a treatment method may be input into a management terminal operated by an administrator or operator of the system 10, and the input selection criteria may be stored in the treatment method database via the server device 100 (or without going through the server device 100). Selection criteria for a treatment method may be input into a user terminal 200, and the input selection criteria may be stored in the treatment method database via the server device 100 (or without going through the server device 100). In this way, the user can customize the selection criteria for a treatment method.

[0124] The text information on treatment methods is not particularly limited and can be selected as appropriate, but examples include papers and guidelines that incorporate the consensus of doctors.

[0125] The server device 100 inputs the generated prompt to the language model server 300, thereby requesting the execution of inference using a language model. In the language model server 300, the inference engine executes forward calculation of the Transformer based on the input prompt and outputs an answer. The output answer is transmitted from the language model server 300 to the server device 100, and further transmitted from the server device 100 to the user terminal 200. The server device 100 may process a part of the answer received from the language model server 300 and then transmit the processed answer to the user terminal 200.

[0126] The prompt generated by the server device 100 may include evaluation criteria for evaluating the patient's condition. The evaluation criteria are not particularly limited as long as they can evaluate the patient's condition and symptoms based on patient information, and may include, for example, criteria that can evaluate the severity of the patient's condition and symptoms. In particular, for benign diseases, it is effective to identify the severity of the patient's condition based on the evaluation criteria and specify a treatment method based on this severity. In other words, the selection criteria may be such that different treatment methods can be selected depending on the evaluation results based on the evaluation criteria, such as severity.

[0127] The evaluation criteria are not particularly limited, but examples include the frequency and severity of symptoms (e.g., number of attacks, pain score, number of diarrhea episodes), degree of impairment of daily living functions and physical functions (e.g., walking ability, ADL, exercise limitation), objective test values ​​and biomarkers (e.g., CRP, complement, BNP, CK levels), organ- and tissue-specific clinical findings (e.g., joint swelling, skin hardening, endoscopic score), the patient's own subjective health assessment (e.g., VAS, QoL score, sense of control), and the presence or absence of complications and serious events (e.g., nephritis, neuropathy, fistula / abscess).

[0128] Evaluation criteria are established for each disease. For example, when the disease is asthma (ACT, GINA), the evaluation criteria used are symptom frequency / severity / subjective assessment (more specifically, attack frequency, nighttime symptoms, number of rescue inhaler uses, limitation of daily activities, and sense of control). For example, when the disease is rheumatoid arthritis (DAS28, CDAI), the evaluation criteria used are organ-specific findings / biomarkers / subjective assessment (more specifically, number of tender joints, number of swollen joints, CRP / ESR, and patient VAS).

[0129] The evaluation criteria may be registered in advance in the treatment method database, or may be registered in advance in an evaluation criteria database separate from the treatment method database. The server device 100 generates a prompt including, for example, patient information acquired from the patient information database or the user terminal 200 and selection criteria for a treatment method acquired from the treatment method database.

[0130] The evaluation criteria to be included in the prompt can be identified by searching for evaluation criteria that are highly similar to the patient information from among the evaluation criteria registered in the evaluation criteria database. The search is performed by identifying evaluation criteria that are highly similar to the patient information based on the Euclidean distance or cosine similarity between a patient information vector, which is a vectorization of at least a portion of the patient information (e.g., information about the patient's disease or condition), and an evaluation criterion vector, which is a vectorization of information about the evaluation criteria stored in the evaluation criterion database. This makes it possible to include one or more evaluation criteria that correspond to the patient's disease or condition included in the patient information in the prompt.

[0131] In the evaluation criteria database, tags may be assigned for each disease or each evaluation criterion (for example, the name of the disease or the evaluation items of the evaluation criteria may be assigned as tags). The tags can be traced based on keywords included in the patient information to search for evaluation criteria corresponding to the patient information. This makes it possible to include one or more evaluation criteria corresponding to the patient's disease or condition included in the patient information in the prompt. Note that tags may be assigned for each disease or each evaluation criterion using inference using a language model.

[0132] Diseases are broadly classified into two types: malignant diseases (cancer / tumor) and benign diseases. The prompt may include information requesting the user to identify whether the patient's disease is malignant (cancer / tumor) or benign based on patient information. The prompt may include information requesting the user to identify whether the patient's disease is malignant (cancer / tumor) or benign and then specify a treatment method based on selection criteria. Information regarding whether the patient's disease is malignant or benign, as determined by inference using a language model, may be displayed on the display screen of the user terminal 200.

[0133] Furthermore, treatments for malignant diseases can be classified into systemic therapy and local therapy. The key to selection is whether the number of lesions can be counted, which can be determined using visual data, i.e., images (CT, MRI, endoscopic findings, ultrasound findings, etc.). If the number of lesions cannot be counted, systemic therapy is selected, and if the number of lesions can be counted, local therapy is selected. The prompt may include information requesting the user to determine whether systemic therapy or local therapy is appropriate for the malignant disease based on patient information. The prompt may also include information requesting the user to determine whether systemic therapy or local therapy is appropriate for the malignant disease based on patient information, and then specify a treatment method based on the determination of whether systemic therapy or local therapy is appropriate. The prompt may also include information requesting the user to determine whether the number of lesions is greater than or equal to a predetermined number based on patient information (or whether the number of lesions can be counted), and then specify whether systemic therapy or local therapy is appropriate. Note that information regarding whether systemic therapy or local therapy is appropriate, determined by inference using a language model, may be displayed on the display screen of the user terminal 200.

[0134] The prompt may include information requesting the user to create a finding based on patient information (including image data of the patient's interior). A finding refers to a fact or opinion that can be confirmed from the patient information. The prompt may also include information requesting the user to identify a treatment method based on the created finding. The finding created by inference using a language model may be displayed on the display screen of the user terminal 200.

[0135] In addition, image data taken inside a patient's body can be analyzed using artificial intelligence capable of analyzing images.

[0136] Furthermore, the prompt generated by the server device 100 may include not only the patient information and the treatment method selection criteria, but also the target outcome. The generated prompt may also include information requesting that a treatment method be output based on the target outcome. By including the target outcome in the prompt, one or more treatment methods based on the target outcome can be identified for the patient corresponding to the patient information.

[0137] An outcome refers to the symptoms, disease state, health state, physical function / ability, or quality of life that occur in a patient as a result of treatment. Outcomes also include outcome indicators that indicate survival rate, mortality rate, recovery rate, or readmission rate after a specified period of time has passed, such as a 5-year survival rate of x%. Alternatively, an outcome refers to a change in the condition of symptoms, disease state, health state, physical function / ability, or quality of life that occurs in a patient as a result of treatment.

[0138] The target outcome is a desired outcome, and is information about the results of treatment desired by patients, doctors, and other medical professionals. The content of the target outcome is not particularly limited, but examples include recovery of motor function, a change in the stage of cancer, and a 5-year survival rate of ○%.

[0139] The target outcome is input, for example, by the user terminal 200 and transmitted to the server device 100. The target outcome may also be input as text information.

[0140] The user can select whether or not to limit the treatment methods obtained as a response to only those covered by public health insurance by operating the user terminal 200. If the user selects to limit the treatment methods obtained as a response to only those covered by public health insurance, information indicating that the selection to limit the treatment methods to only those covered by public health insurance has been made is transmitted from the user terminal 200 to the server device 100.

[0141] In this case, the prompt generated by the server device 100 may include information requesting the user to identify only treatment methods that are covered by public health insurance. As a result, the language model server 300 identifies only treatment methods that are covered by public health insurance, and transmits only the identified treatment methods that are covered by public health insurance to the user terminal 200 via the server device 100.

[0142] The user can select whether or not to include treatment methods that are not covered by public health insurance in the treatment methods obtained as a response by operating the user terminal 200. When the user selects whether or not to include treatment methods that are not covered by public health insurance in the treatment methods obtained as a response, information regarding which option was selected is transmitted to the server device 100 by the user terminal 200.

[0143] If the user selects to include treatment methods that are not covered by public health insurance in the treatment methods obtained as an answer, the prompt generated by the server device 100 may include information requesting the user to specify a treatment method while including treatment methods that are not covered by public health insurance in the candidate treatment methods. As a result, the language model server 300 specifies treatment methods while including treatment methods that are covered by public health insurance as candidate treatment methods, and the specified treatment methods are transmitted to the user terminal 200 via the server device 100. On the other hand, if the user selects not to include treatment methods that are not covered by public health insurance in the treatment methods obtained as an answer, the prompt generated by the server device 100 may include information requesting the user to specify a treatment method while excluding treatment methods that are not covered by public health insurance from the candidate treatment methods.

[0144] The system 10 can also determine whether the patient information lacks information necessary to determine the appropriateness of a treatment method. Furthermore, if the information necessary to determine the appropriateness of a treatment method is lacking, the system 10 can identify the missing information. For example, the prompt generated by the server device 100 may include, along with the patient information, information requesting the user to determine whether the information necessary to determine the appropriateness of a treatment method is lacking. Furthermore, the prompt generated by the server device 100 may also include, along with the patient information, information requesting the user to output the missing information as a response if the information necessary to determine the appropriateness of a treatment method is lacking. This allows the system 10 to determine whether the information necessary to determine the appropriateness of a treatment method is lacking, and, if the information is lacking, to identify the missing information. The information output from the language model server 300 indicating that information is lacking and which information is lacking is transmitted to the user terminal 200 via the server device 100.

[0145] Here, the prompt generated by the server device 100 includes patient information and treatment method selection criteria. However, for example, the language model may be fine-tuned using information related to the treatment method selection criteria. In this case, the prompt generated by the server device 100 only needs to include patient information, and may not necessarily include the treatment method selection criteria. In this case, the generated prompt may also include information requesting the user to identify a treatment method suitable for the patient corresponding to the patient information.

[0146] The server device 100 can identify information regarding the advantages and / or disadvantages of performing the identified treatment method on the patient. The information regarding the advantages and / or disadvantages of performing the treatment method can be obtained from a treatment method database. The identified information, including the information regarding the advantages and / or disadvantages, is transmitted from the server device 100 to the user terminal 200 and displayed on the user terminal 200.

[0147] For the identified treatment method, the server device 100 can identify the patient survival rate, the probability of cure of the disease or condition, the recurrence rate of the disease or condition, or the local control rate of the disease or condition when the treatment method is performed on the patient. The patient survival rate, the probability of cure of the disease or condition, the recurrence rate of the disease or condition, or the local control rate of the disease or condition when the treatment method is performed on the patient may be acquired from the treatment method database or may be acquired from the language model server 300. The patient survival rate, the probability of cure of the disease or condition, the recurrence rate of the disease or condition, or the local control rate of the disease or condition when the treatment method is performed on the patient is transmitted from the server device 100 to the user terminal 200 and displayed on the user terminal 200.

[0148] The server device 100 can identify a predicted outcome when the identified treatment method is performed on a patient. The predicted outcome when the treatment method is performed on a patient can be identified using information for identifying a predicted outcome that is registered in a treatment method database. More specifically, the server device 100 generates a prompt including patient information, the treatment method, and information for identifying a predicted outcome, and can obtain a predicted outcome corresponding to the patient information when the treatment method is performed from the language model server 300.

[0149] Examples of information for identifying a predicted outcome include a correspondence table between outcome predictions and outcome factors (e.g., proven in a paper, etc.), or a calculation formula for identifying a predicted outcome. The calculation formula can identify a predicted outcome by substituting items identified from the disease state. Based on patient information about a large number of actual patients, the treatment methods administered to these patients, and the resulting outcomes, a correspondence table between outcome predictions and outcome factors as information for identifying a predicted outcome, or a calculation formula for identifying a predicted outcome, may be identified by inference using a language model.

[0150] The system 10 can identify the suitability of a treatment method identified by the method identification means, which is the degree of suitability for a target outcome or a risk tolerance. A user can input the target outcome or the risk tolerance by operating the user terminal 200. The target outcome or the risk tolerance input at the user terminal 200 is transmitted to the user terminal 200 via the server device 100 and displayed on the user terminal 200.

[0151] (Flowchart of Output Processing) Next, the output processing will be described. FIG. 5 is a flowchart of the output processing according to the embodiment. A user operates the user terminal 200 to start an application and log in (step S1). The application may be a web application or a native application. When logging in, the user inputs a user ID (identification information for identifying the user).

[0152] Next, the user operates the user terminal 200 to input patient information (step S2). The input patient information is transmitted from the user terminal 200 to the server device 100 (step S3) and received by the server device 100 (step S4).

[0153] Based on the patient information, the server device 100 obtains information about treatment methods associated with the patient information from the treatment method database (step S5). This information about treatment methods includes information about selection criteria for each treatment method. Next, based on the patient information, the server device 100 obtains information about evaluation criteria associated with the patient information from the evaluation criteria database (step S6). The information about evaluation criteria includes information about evaluation criteria such as the severity of each symptom / condition.

[0154] Next, the control unit of the server device 100 generates a prompt including the patient information received in step S4, information about the treatment selection criteria acquired in step S5, and information about the evaluation criteria acquired in step S6 (step S7). The prompt may include information requesting that the severity of the patient corresponding to the patient information be identified based on the evaluation criteria, and that a treatment method that meets the selection criteria be identified based on the patient information and the severity. The generated prompt is transmitted from the server device 100 to the language model server 300, and is received by the language model server 300 (step S8).

[0155] In the language model server 300, the inference engine uses the language model to perform inference based on the received prompt and generates an answer (step S9). The answer includes information about a treatment method suitable for the patient corresponding to the patient information received in step S4.

[0156] Next, the generated answer (information on a treatment method suitable for the patient) is transmitted from the language model server 300 to the server device 100 (step S10), and is received by the server device 100 (step S11).

[0157] Next, the server device 100 generates a prompt including the patient information received in step S4 and information about a treatment method suitable for the patient received in step S11 (step S12). The prompt may include information requesting a predicted outcome when the treatment method included in the prompt is administered to a patient corresponding to the patient information. The generated prompt is transmitted from the server device 100 to the language model server 300 and received by the language model server 300 (step S13).

[0158] In the language model server 300, the inference engine uses the language model to perform inference based on the received prompt and generates an answer (step S14). The answer includes the patient information received in step S4 and information about the predicted outcome when the treatment method appropriate for the patient received in step S11 is implemented.

[0159] Next, the generated answer (information regarding the predicted outcome) is transmitted from the language model server 300 to the server device 100 (step S15), and is received by the server device 100 (step S16). Next, the information regarding the treatment method appropriate for the patient received in step S11 and the information regarding the predicted outcome received in step S16 are transmitted from the server device 100 to the user terminal 200 (step S17), and are received by the user terminal 200 (step S18). The information received by the user terminal 200 is displayed on the display screen of the user terminal 2 (step S19). Steps S1 to S19 complete the output process.

[0160] <Sixteenth Embodiment> The matters described in the sixteenth embodiment can also be applied to the first to fifteenth embodiments. The system 10 of the sixteenth embodiment can have the same configuration as the system 10 shown in FIG. 4. However, the system 10 does not have to include the language model server 300. The system 10 includes at least one computer device. The system 10 may be configured, for example, from one computer device (information processing device) (standalone type), from one or more server devices and one or more terminal devices (client-server type), or from multiple terminal devices (peer-to-peer type).

[0161] The system 10 may include a user terminal 200 and a server device 100. The server device 100 corresponds to the computer 100 described above. The server device 100 may also function in a distributed manner across multiple computer devices. For example, instead of the server device 100, a distributed ledger technology such as a blockchain may be used.

[0162] The user terminal 200 and the server device 100 are communicably connected to each other via a communication network 120. The number of user terminals 200 is not particularly limited as long as it is one or more, and there may be more than one.

[0163] (User Terminal) The user terminal 200 is a terminal operated by a user of the system 10. Here, the user may be any person who uses the system 10. Examples of the user terminal 200 include a conventional mobile phone, a tablet terminal, a smartphone, and a personal computer.

[0164] The user terminal 200 includes a control unit, a storage unit, an input unit, a display unit, and a communication unit, which are all connected by a bus. The functions and roles of the control unit, storage unit, input unit, display unit, and communication unit of the user terminal 200 are as described in the fifteenth embodiment.

[0165] 3 is a block diagram showing the hardware configuration of a server device according to an embodiment. The server device 100 includes at least a control unit 101, a storage unit 102, and a communication unit 106, which are connected to each other via an internal bus. In the server device 100, the functions and roles of the control unit 101, the storage unit 102, and the communication unit 106 are the same as those described in the fifteenth embodiment.

[0166] (System) The system 10 can identify a treatment method corresponding to patient information desired by a user based on a trained model that has been machine-learned using patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data. The trained model is a model constructed by machine learning multiple data sets consisting of patient information and treatment methods suitable for patients corresponding to the patient information. This makes it possible to identify a treatment method suitable for the patient.

[0167] The server device 100 can also re-learn and update the trained model based on new patient information for machine learning and treatment methods suitable for the new patient information for machine learning.

[0168] Furthermore, the system 10 can identify one or more treatment methods corresponding to the patient information and the target outcome based on a trained model that is machine-learned using patient information for machine learning and outcomes resulting from administering a treatment method to a patient corresponding to the patient information for machine learning as input data, and the treatment methods administered to the patient corresponding to the patient information for machine learning as output data. The trained model is a model constructed by machine learning multiple data sets consisting of patient information, outcomes resulting from administering a treatment method to a patient corresponding to the patient information, and the treatment methods administered to the patient corresponding to the patient information. This makes it possible to identify a treatment method that is appropriate for the patient and consistent with the target outcome.

[0169] It is also possible to update the trained model by re-training based on new patient information for machine learning, the outcomes resulting from administering the treatment method to the patient corresponding to the new patient information for machine learning, and the treatment method administered to the patient corresponding to the patient information for machine learning.

[0170] The trained model is stored in the server device 100. When a user operates the user terminal 200 to input patient information, the input patient information is transmitted from the user terminal 200 to the server device 100. Based on the received patient information, the server device 100 can use the trained model to identify a treatment method corresponding to the received patient information. The identified treatment method is transmitted from the server device 100 to the user terminal 200 and displayed on the user terminal 200.

[0171] Furthermore, when a user operates the user terminal 200 to input patient information and a target outcome, the input patient information and target outcome are transmitted from the user terminal 200 to the server device 100. The target outcome is input as text information. Based on the received patient information and target outcome, the server device 100 can use a trained model to identify a treatment method corresponding to the received patient information. The identified treatment method is transmitted from the server device 100 to the user terminal 200 and displayed on the user terminal 200.

[0172] The patient information for machine learning, or the patient information of the patient for which a treatment method is to be identified, is not particularly limited as long as it is information about the patient, but it is preferably information that will influence the decision on the treatment method. Patient information may include, for example, basic attributes such as age and gender, information on the disease and condition such as disease name (e.g., hepatocellular carcinoma, colorectal cancer, diabetes, etc.), primary lesion (e.g., liver, pancreas, stomach, etc.), presence or absence of metastasis (e.g., lung metastasis, liver metastasis), presence or absence of recurrence, complications (e.g., concomitant diabetes, concomitant hypertension), tumor infiltration (e.g., left hepatic vein invasion, middle hepatic vein invasion, right hepatic vein invasion, vascular invasion, bile duct invasion, gastrointestinal tract invasion, Glisson's sheath invasion), information on anatomical location such as proximity to blood vessels, proximity to the bile duct, and proximity to the gastrointestinal tract, information on organ and bodily functions such as low liver function, FLR (e.g., normal liver / damaged liver), and whether or not extensive liver resection is possible (e.g., trisectionectomy, right hepatectomy, etc.), pathological and histological findings, genetic and molecular biological information such as genetic abnormalities and their degree and gene expression levels, and immunological findings such as the cancer microenvironment.

[0173] The treatment method suitable for the patient information is a treatment method suitable for the patient corresponding to this patient information. The treatment method suitable for the patient used in machine learning may be different from the treatment method actually administered to the patient. The treatment method suitable for the patient used in machine learning may be, for example, a treatment method derived as an appropriate treatment method by an expert based on the patient information.

[0174] The target outcome is a desired outcome, and is information about the results of treatment desired by patients, doctors, and other medical professionals. The content of the target outcome is not particularly limited, but examples include recovery of motor function and a change in the stage of cancer.

[0175] System 10 uses patient information for machine learning and a treatment method for a patient corresponding to the patient information for machine learning as input data, and can identify outcomes when the treatment method is administered to a patient corresponding to the acquired patient information based on a trained model that has been machine-learned using outcomes when the treatment method is administered to a patient corresponding to the patient information for machine learning as output data. The trained model is a model constructed by machine learning multiple data sets consisting of patient information, a treatment method for a patient corresponding to the patient information, and outcomes when the treatment method is actually administered to a patient corresponding to the patient information. This makes it possible to predict outcomes when the treatment is administered to a patient.

[0176] As described above, system 10 can identify a treatment method suitable for a patient corresponding to the patient information based on the patient information, or based on the patient information and a target outcome, and system 10 can also predict an outcome based on this identified treatment method.

[0177] The machine learning algorithm in the trained model is not particularly limited and can be any known algorithm, but it is preferable to use deep learning using a multilayer neural network. The multilayer neural network has an input layer, an output layer, and multiple intermediate layers. Weights are set for the edges connecting nodes in each layer. Weights corresponding to each input to the node are set for the edges, and the edges are multiplied by the weights corresponding to each input to the node, and the values ​​obtained by multiplying these weights are added to a bias. The value obtained by the addition is subjected to nonlinear transformation using an activation function to calculate an activation value. The calculated activation value becomes the input value passed to the node in the next layer. The number of intermediate layers can be designed as appropriate.

[0178] (Flowchart of Output Processing) Next, the output processing will be described. FIG. 6 is a flowchart of the output processing according to the embodiment. The user operates the user terminal 200 to start an application and log in (step S21). The application may be a web application or a native application. When logging in, the user inputs a user ID (identification information for identifying the user).

[0179] Next, the user operates the user terminal 200 to input patient information (step S22). The input patient information is transmitted from the user terminal 200 to the server device 100 (step S23) and received by the server device 100 (step S24). Based on the input patient information, the server device 100 identifies one or more treatment methods corresponding to the acquired patient information (step S25). In step S25, a trained model that uses the patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data can be used to identify a treatment method corresponding to the acquired patient information.

[0180] Next, based on the patient information received in step S24 and the treatment method identified in step S25, a predicted outcome when the treatment method identified in step S25 is administered to the patient corresponding to the patient information received in step S24 is identified (step S26). In step S26, the predicted outcome can be identified using a trained model that uses the patient information for machine learning and the treatment method for the patient corresponding to the patient information for machine learning as input data and the outcome when the treatment method is administered to the patient corresponding to the patient information for machine learning as output data.

[0181] The treatment method identified in step S25 and the predicted outcome identified in step S26 are transmitted from the server device 100 to the user terminal 200 (step S27). The information received by the user terminal 200 is displayed on the display screen of the user terminal 2 (step S28). The output process is completed after steps S21 to S28.

[0182] REFERENCE SIGNS LIST 10 System 100 Computer (server device) 101 Control unit 102 Storage unit 103 Peripheral device I / F unit 104 Input unit 105 Display unit 106 Communication unit 110 Bus 120 Network 130 External server 140 Database 200 User terminal 300 Language model server

Claims

1. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information; and a method identification means for identifying one or more treatment methods for a patient corresponding to the acquired patient information based on the acquired patient information and treatment method selection criteria.

2. The system of claim 1, wherein the method specifying means specifies a surgical procedure, a drug regimen, an immunotherapy or cell therapy regimen, or a radiation therapy prescription.

3. The system according to claim 1 or 2, wherein the method identification means receives as input the acquired patient information and a prompt including selection criteria for the treatment method, and requests execution of inference using a language model to identify one or more treatment methods for the patient corresponding to the acquired patient information.

4. The system according to any one of claims 1 to 3, wherein the method specifying means specifies one or more treatment methods based on evaluation criteria of the patient's condition.

5. The system of claim 3 or 4, wherein the method identification means receives a prompt including selection criteria for a treatment method obtained from the knowledge base as input, and requests execution of inference using a language model to identify one or more treatment methods for a patient corresponding to the obtained patient information.

6. A system according to any one of claims 3 to 5, wherein the method identification means receives a prompt including the acquired patient information as input and requests execution of inference using a fine-tuned language model using information related to selection criteria for the treatment method, thereby identifying one or more treatment methods for a patient corresponding to the acquired patient information.

7. A system according to any one of claims 3 to 6, wherein the method identification means receives as input a prompt including information about the patient's medical record and / or information about images taken inside the patient's body, and requests the execution of inference using a language model to identify one or more treatment methods for the patient corresponding to the acquired patient information.

8. A system as described in any one of claims 3 to 7, comprising: an input means for inputting a target outcome; and a method identification means for identifying one or more treatment methods based on the target outcome for a patient corresponding to the acquired patient information by inputting a prompt including patient information and the target outcome and requesting the execution of inference using a language model.

9. A system according to any one of claims 1 to 8, wherein the patient information includes image information relating to the body, the selection criteria can be determined from the image information relating to the body, and the method identification means identifies one or more treatment methods for the patient corresponding to the acquired patient information based on the acquired patient information and the selection criteria for the treatment methods.

10. A system according to any one of claims 1 to 9, wherein the patient information includes information that enables the distance between a first organ and a second organ within the patient's body to be ascertained, and the criteria for selecting a treatment method include information regarding the distance between the first organ and the second organ as a criterion for enabling the selection of at least one treatment method.

11. A system according to any one of claims 1 to 10, wherein the patient information includes information that can be used to ascertain the number of tumors in the patient's body, the size of the tumors, the extent of tumor spread, the anatomical location of the tumors, whether the area where the tumors are present is a site where tumors are likely to occur, or whether the lesion is localized, and the criteria for selecting a treatment method include information that can be used to ascertain the number of tumors in the patient's body, the size of the tumors, the extent of tumor spread, the anatomical location of the tumors, whether the area where the tumors are present is a site where tumors are likely to occur, or whether the lesion is localized, as criteria for enabling the selection of at least one treatment method.

12. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information; and a method identification means for identifying one or more treatment methods corresponding to the acquired patient information based on a trained model that uses patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data.

13. The system according to claim 12, further comprising an update means for updating the trained model by retraining based on new patient information for machine learning and a treatment method suitable for the new patient information for machine learning.

14. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information; an input means for inputting a target outcome; and a method identification means for identifying one or more treatment methods corresponding to the acquired patient information and target outcome based on a trained model that uses patient information for machine learning and the outcome resulting from administering the treatment method to a patient corresponding to the patient information for machine learning as input data, and the treatment method administered to the patient corresponding to the patient information for machine learning as output data.

15. The system described in claim 14, further comprising an update means for updating the trained model by retraining based on new patient information for machine learning, outcomes resulting from administering the treatment method to a patient corresponding to the new patient information for machine learning, and the treatment method administered to a patient corresponding to the patient information for machine learning.

16. A system according to any one of claims 1 to 15, comprising a first information identifying means for identifying information including the advantages and / or disadvantages of a treatment method based on the treatment method identified by the method identifying means.

17. A system according to any one of claims 1 to 16, further comprising a second information specifying means for specifying the survival rate, cure probability, recurrence rate, or local control rate resulting from the implementation of the treatment method identified by the method specifying means.

18. A system according to any one of claims 1 to 17, further comprising a third information specifying means for specifying a prediction of an outcome of a treatment method based on the treatment method specified by the method specifying means.

19. A system according to any one of claims 1 to 18, comprising: an input means for inputting a patient's target outcome or the patient's risk tolerance; and a suitability identification means for identifying information that can grasp the degree to which a treatment method identified by the method identification means is suitable for the input target outcome or risk tolerance.

20. A system according to any one of claims 1 to 11 and claims 16 to 19, wherein the treatment method selection criteria are generated by inputting a prompt containing text information about the treatment method and requesting the execution of inference using a language model.

21. A system according to any one of claims 1 to 20, wherein the method identification means identifies only treatment methods covered by public health insurance.

22. A system according to any one of claims 1 to 21, wherein the method identification means is capable of identifying treatment methods that are not covered by public health insurance.

23. A system according to any one of claims 1 to 22, comprising a missing information identifying means for identifying missing information when the acquired patient information lacks information necessary to determine the suitability of a treatment method.

24. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification means for identifying a predicted outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting the execution of inference using a language model, using as input a prompt including the acquired patient information, the treatment method for the patient corresponding to the patient information, and a correspondence between the patient information and the outcome when the treatment method is administered to the patient corresponding to the patient information.

25. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification means for identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by receiving as input a prompt including the acquired patient information and the treatment method for the patient corresponding to the patient information and requesting the execution of inference using a fine-tuned language model using information regarding the correspondence between the patient information and the outcome when the treatment method is administered to the patient corresponding to the patient information.

26. A system comprising at least one computer device, comprising: an acquisition means for acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification means for identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information, based on a trained model that uses patient information for machine learning and a treatment method for a patient corresponding to the patient information for machine learning as input data and outputs the outcome when the treatment method is administered to a patient corresponding to the patient information for machine learning.

27. The system described in claim 26, further comprising an update means for updating the trained model by re-training based on new patient information for machine learning, a treatment method for a patient corresponding to the patient information for machine learning, and an outcome when the treatment method is administered to a patient corresponding to the new patient information for machine learning.

28. A system according to any one of claims 24 to 27, comprising a method specifying means for specifying one or more treatment methods for a patient corresponding to the acquired patient information based on the acquired patient information and a selection criterion for the treatment method, and an outcome specifying means for specifying an outcome based on the specified treatment method.

29. A system according to any one of claims 24 to 27, comprising a method identification means for identifying one or more treatment methods corresponding to acquired patient information based on a trained model that uses patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data, and an outcome identification means for identifying an outcome based on the identified treatment methods.

30. A system according to any one of claims 1 to 29, wherein the patient information includes information relating to images taken inside the patient's body.

31. The system of any one of claims 1 to 30, wherein the patient information includes text information.

32. A method executed in a system having at least one computer device, comprising: an acquisition step of acquiring patient information; and a method identification step of identifying one or more treatment methods for a patient corresponding to the acquired patient information by requesting the execution of inference using a language model, using the acquired patient information and a prompt including treatment method selection criteria as input.

33. A method executed in a system having at least one computer device, comprising: an acquisition step of acquiring patient information; and a method identification step of identifying one or more treatment methods for a patient corresponding to the acquired patient information by requesting execution of inference using a fine-tuned language model with information regarding treatment method selection criteria, using a prompt including the acquired patient information as input.

34. A method executed in a system having at least one computer device, comprising: an acquisition step of acquiring patient information; and a method identification step of identifying one or more treatment methods corresponding to the acquired patient information based on a trained model that uses the patient information for machine learning as input data and treatment methods suitable for the patient information for machine learning as output data.

35. A method executed in a system having at least one computer device, comprising: an acquisition step of acquiring patient information; an input step of inputting a target outcome; and a method identification step of identifying one or more treatment methods corresponding to the acquired patient information and target outcome based on a trained model that uses patient information for machine learning and the outcome resulting from administering the treatment method to a patient corresponding to the patient information for machine learning as input data, and the treatment method administered to the patient corresponding to the patient information for machine learning as output data.

36. A method executed in a system having at least one computer device, comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting the execution of inference using a language model, using as input a prompt including the acquired patient information, the treatment method for the patient corresponding to the patient information, and a correspondence between the patient information and the outcome when the treatment method is administered to the patient corresponding to the patient information.

37. A method executed in a system having at least one computer device, the method comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information by requesting the execution of inference using a fine-tuned language model using information regarding the correspondence between the patient information and the outcome when the treatment method is administered to the patient corresponding to the patient information, using a prompt including the acquired patient information and the treatment method for the patient corresponding to the patient information as input.

38. A method executed in a system having at least one computer device, the method comprising: an acquisition step of acquiring patient information and a treatment method for a patient corresponding to the patient information; and an outcome identification step of identifying an outcome when the treatment method is administered to a patient corresponding to the acquired patient information, based on a trained model that uses the patient information for machine learning and the treatment method for the patient corresponding to the patient information as input data and the outcome when the treatment method is administered to a patient corresponding to the patient information for machine learning as output data.

Images