Treatment support system

Abstract

We provide a treatment support system that can reduce doctors' working hours. [Solution] The treatment support system 10 is a virtual medical professional means that virtually assumes the role of a medical professional, and has multiple virtual medical professional means (agents) (virtual medical professional means group 20) that output content related to the patient's treatment plan using a generated AI model based on the input patient information. The virtual medical professional means group 20 includes an internist agent that outputs content related to the patient's internal treatment methods based on the input patient information, a surgeon agent that outputs content related to the patient's external treatment methods based on the same information as the patient information, and a treatment plan aggregation agent that takes the content related to internal treatment methods output from the internist agent and the content related to external treatment methods output from the surgeon agent as input information to comprehensively determine the patient's final treatment plan.

Description

Technical Field

[0001] The present invention relates to a treatment support system that supports the determination of a treatment policy for a patient based on the input patient information.

Background Art

[0002] In the medical field, various technologies have been developed so far for improving convenience, preventing misdiagnosis, shortening the operation time, and the like.

[0003] For example, Patent Document 1 describes a platform that mimics the role of a doctor and uses artificial intelligence, machine learning, medical history, and natural language to provide medical treatment without the presence of a doctor. Patent Document 1 describes that the platform diagnoses abnormalities through dialogue and provides prescriptions and treatment plans. Further, Patent Document 1 describes that the platform includes an external communication module that communicates with a plurality of external medical sources and edits the user's medical records into a single format, and a diagnosis module that automatically diagnoses the user's diseases and treatment options for those diseases using the user's medical records and a database of all diseases and symptoms, and a case monitoring module that tracks the user's recovery time and adjusts the recovery time when a treatment change is executed.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, under the leadership of the Ministry of Health, Labour and Welfare, a reform to doctors' working styles was implemented in April 2024 with the aim of ensuring the quality of medical care and maintaining a sustainable medical care delivery system. As a result, limits were placed on doctors' overtime and holiday work hours. Therefore, reducing working hours (overtime) has become an urgent task, especially for young doctors who lack experience and knowledge.

[0006] In response to these practical challenges, the invention described in Patent Document 1 can mimic the role of a physician and provide medical care without the presence of a physician. However, questions remain as to whether it is actually possible to safely and quickly provide medical services to patients without the presence of a physician (fully autonomous).

[0007] Rather, Patent Document 1 does not address whether or not medical services can be provided safely and quickly to patients without the presence of a physician, nor does it disclose any specific means for providing medical services safely and quickly to patients without the presence of a physician.

[0008] Therefore, the invention described in Patent Document 1 cannot solve the aforementioned practical problems. Accordingly, the present invention aims to provide a treatment support system that can help physicians provide medical services to patients safely and quickly, and reduce physicians' working hours. [Means for solving the problem]

[0009] The treatment support system of the present invention is a system that supports the determination of a treatment plan for a patient based on the patient information that is input. Furthermore, the treatment support system of the present invention is a virtual medical professional means that virtually assumes the role of a medical professional, and has a plurality of virtual medical professional means (hereinafter referred to as "agents") that output content related to the patient's treatment plan using a generated AI model based on the patient information that is input. The plurality of agents include an internist agent that outputs content related to the patient's internal treatment methods based on the patient information that is input, a surgeon agent that outputs content related to the patient's external treatment methods based on the same information as the information input to the internist agent, and a treatment plan aggregation agent that comprehensively determines the patient's final treatment plan using the content related to internal treatment methods output from the internist agent and the content related to external treatment methods output from the surgeon agent as input information.

[0010] This allows for a thorough examination of the patient's treatment plan from both internal and external treatment perspectives.

[0011] Furthermore, it is desirable to include, as an internal medicine agent, a first internal medicine agent that outputs a draft of internal treatment for the patient based on the input patient information; a pharmacist agent that modifies the draft from the perspective of pharmacotherapy and outputs it, using the draft output from the first internal medicine agent as input information; and a second internal medicine agent that outputs a final treatment plan regarding the patient's internal treatment, using the draft output from the pharmacist agent as input information. It is also desirable to include, as a surgeon agent, a first surgeon agent that determines whether or not surgery is necessary for the patient based on the input patient information; a second surgeon agent that, if the first surgeon agent determines that surgery is necessary, outputs a draft of external treatment for the patient based on the input patient information; a clinical engineer agent that adds supplementary information regarding the equipment and / or technology used in the draft from the perspective of clinical engineering and outputs it, using the draft output from the second surgeon agent as input information; and a third surgeon agent that outputs a final treatment plan regarding the patient's external treatment, using the draft output from the clinical engineer agent as input information.

[0012] Here, it is desirable that the first physician agent, based on the entered patient information, further output whether and / or what additional tests the patient requires.

[0013] Furthermore, it is desirable that the first internist agent, pharmacist agent, first surgeon agent, second surgeon agent, and clinical engineer agent refer to a database containing one or more of the following: medical guidelines, drug package inserts, past case information (medical record data), and hospital manuals, and output the results using Retrieval-Augmented Generation (RAG).

[0014] Furthermore, the treatment strategy aggregation agent comprehensively determines the patient's treatment strategy so that internal and external treatments complement each other, and it is desirable that this treatment strategy includes specific treatment strategies for internal treatments, specific treatment strategies for external treatments, whether or not additional tests are required and / or what those tests are, and the overall most recommended treatment strategy.

[0015] Furthermore, it is desirable that the treatment support system include a patient explanation agent that, based on the content of the patient's treatment plan output from the treatment plan aggregation agent, uses a generative AI model to output explanatory text from the perspective of a specialist internist or specialist surgeon, for use with the patient and / or related parties. [Effects of the Invention]

[0016] According to the present invention, it is possible to provide a treatment support system that helps physicians to provide medical services to patients safely and quickly, while also reducing physicians' working hours. [Brief explanation of the drawing]

[0017] [Figure 1] This is a schematic diagram of a treatment support system according to an embodiment of the present invention. [Figure 2] Figure 1 is a schematic diagram of the virtual medical personnel group of the treatment support system shown. [Figure 3] This is an example of a processing flow for a treatment support system according to an embodiment of the present invention. [Figure 4] This is another example of the processing flow of a treatment support system according to an embodiment of the present invention. [Modes for carrying out the invention]

[0018] The embodiments of the present invention will be described in detail below. However, the description of the constituent elements described below is an example (representative example) of the embodiments of the present invention, and the present invention is not limited to the following content as long as it does not deviate from the gist thereof. For configurations having the same function, the same reference numerals are given in the drawings, and the detailed description thereof is omitted. In addition, in this description, internal therapy and non-invasive therapy are synonymous, and external therapy and invasive therapy are synonymous.

[0019] [Therapy Support System] First, the therapy support system according to an embodiment of the present invention will be described. FIG. 1 is a schematic configuration diagram of a therapy support system according to an embodiment of the present invention. The therapy support system 10 communicates with the terminal 30 and the external device 40 mutually by wired or wireless means.

[0020] The terminal 30 is a personal computer, a tablet, a smartphone, etc., and transmits input information to the therapy support system 10 and receives output information from the therapy support system 10.

[0021] On the other hand, the external device 40 is a device having a database in which medical guidelines, drug package inserts, past case information (medical record data), and hospital manuals are registered. These information sources can be obtained from their respective providers (for example, the Ministry of Health, Labour and Welfare in the case of medical guidelines).

[0022] For these information sources, it is desirable that data such as text, images, and voices are stored in the database of the external device 40 as numerical arrays (vectors) so that they can be searched and managed at high speed based on similarity. So to speak, it is desirable that the external device 40 has a vector database.

[0023] Also, the therapy support system 10 has an input means 11, an output means 12, a learning means 13, a storage means 14, and a group of virtual medical staff means 20.

[0024] The input means 11 receives input information transmitted from the terminal 30 or external device 40. For example, the input means 11 receives patient information transmitted from the terminal 30, and information such as the above-mentioned medical guidelines transmitted from the external device 40.

[0025] Meanwhile, the output means 12 transmits output information to the terminal 30. For example, it outputs the final treatment plan for the patient, which has been formulated by the treatment support system 10, to the doctor's computer or the like.

[0026] The learning means 13 constructs and trains a generative AI model based on the various input information. The generative AI model is stored in the memory means 14.

[0027] The memory means 14 stores not only the generated AI model mentioned above, but also information about the input patient, information such as the medical guidelines mentioned above, and various other information necessary for the treatment support system 10 to formulate a treatment plan for the patient.

[0028] Here, patient information includes the patient's age (date of birth), sex, height, weight, occupation, current medical history, past medical history, family history, current medication information, smoking and allergy information, and exercise habits. For patients who have received examinations or tests at other hospitals (by other doctors), information such as the circumstances of the referral, the results of previously performed tests, and the findings of the doctor who reviewed the test results is also included. Patient information may also include information other than those exemplified above, such as information obtainable from electronic medical records.

[0029] (A group of virtual medical professionals) The virtual medical professional means group 20 is a collection of virtual medical professional means, as shown in Figure 2. A virtual medical professional means is a means that virtually assumes the role of a medical professional, and outputs content regarding the patient's treatment plan using a generated AI model based on the input patient information. Hereafter, for convenience, the virtual medical professional means will be referred to as "agents" in this explanation.

[0030] The agents virtually assume the roles of medical professionals. In this embodiment, the agents include, as physician agents, a first physician agent 211, a second physician agent 212, a first pharmacist agent 213, and a second pharmacist agent 214. Additionally, as surgeon agents, there are a first surgeon agent 221, a second surgeon agent 222, a third surgeon agent 223, and a clinical engineer agent 224. Furthermore, there are agents that virtually assume roles other than physicians and surgeons, such as a treatment plan aggregation agent 231 and a patient explanation agent 232.

[0031] In addition, the virtual medical professional means group 20 of this embodiment includes a physical therapist agent 233, an occupational therapist agent 234, a radiological technologist agent 235, a clinical laboratory technologist agent 236, and the like. Of course, since agents virtually assume the roles of medical professionals, agents that assume the roles of medical professionals not exemplified can also be included in the virtual medical professional means group 20.

[0032] Next, to elaborate on each agent in this embodiment, first, the physician agent outputs information about the patient's internal treatment based on the input patient information. On the other hand, the surgeon agent outputs information about the patient's external treatment based on the same information input to the physician agent.

[0033] The treatment plan aggregation agent 231 then uses the information regarding internal treatments output from the internist agent and the information regarding external treatments output from the surgeon agent as input information to comprehensively determine the patient's final treatment plan.

[0034] Here, the treatment strategy aggregation agent 231 comprehensively determines the patient's final treatment plan so that internal and external treatments complement each other. Furthermore, the treatment strategy aggregation agent 231 determines the final treatment plan so that it includes specific treatment plans for internal treatments, specific treatment plans for external treatments, whether or not additional tests are required and / or what those tests will be, and the overall most recommended treatment plan.

[0035] To elaborate, generally speaking, non-invasive treatment methods that minimize harm to the body (reducing the burden on the patient) are chosen for treating illnesses. However, in addition to non-invasive treatments, there are cases where surgery is preferable. In such cases, it may be necessary to switch medications taken several days to several months in advance, such as from "drug A" to "drug B," or certain medications may not be allowed after surgery. Thus, non-invasive treatments and surgery (invasive treatments) can mutually influence the treatment process.

[0036] Therefore, the aforementioned mutual complementarity means that the treatment strategy aggregation agent 231 considers how these non-invasive and invasive treatments mutually influence each other in determining the optimal final treatment strategy. Of course, the treatment strategy aggregation agent 231 may select only non-invasive treatment as the final treatment strategy, or it may select both non-invasive and invasive treatment, or it may select only invasive treatment.

[0037] Furthermore, the patient explanation agent 232, based on the content of the patient's final treatment plan output from the treatment plan aggregation agent 231, uses a generating AI model to output an explanatory text from the perspective of a specialist internist or specialist surgeon, for explanation to the patient and / or the patient's related parties.

[0038] Furthermore, to elaborate on the physician agent of this embodiment, first, the first physician agent 211 outputs a draft of the patient's internal medicine treatment plan based on the input patient information. In addition, the first physician agent 211 can also output whether and / or the content of any additional tests necessary for the patient, based on the input patient information.

[0039] Furthermore, the first pharmacist agent 213 takes the draft output from the first internist agent 211 as input information, modifies the draft from the perspective of pharmacotherapy, and outputs it.

[0040] Furthermore, the second internist agent 212 takes the draft output from the first pharmacist agent 213 as input information and outputs the final treatment plan regarding the patient's internal treatment.

[0041] In addition, the second pharmacist agent 214 uses the draft output from the first pharmacist agent 213 as input information, checks the draft from a pharmacotherapy perspective, and corrects and outputs it if necessary. Checking from a pharmacotherapy perspective involves mutual checking of drug usage, dosage, drug interactions, etc.

[0042] On the other hand, to elaborate on the surgical agent of this embodiment, first, the first surgical agent 221 determines whether or not surgery is necessary for the patient based on the input patient information.

[0043] Furthermore, if the first surgical agent 221 determines that surgery is necessary, the second surgical agent 222 outputs a draft of the patient's external treatment plan based on the input patient information.

[0044] Furthermore, the clinical engineering agent 224 takes the draft output from the second surgeon agent 222 as input information and outputs supplementary information from a clinical engineering perspective regarding the equipment and / or technology used in the draft.

[0045] Furthermore, the third surgical agent 223 takes the draft output from the clinical engineering agent 224 as input information and outputs the final treatment plan regarding external treatment methods.

[0046] In addition, physical therapist agent 233, occupational therapist agent 234, radiological technologist agent 235, and clinical laboratory technologist agent 236 add supplementary information or make revisions to the draft treatment plan from the perspective of their respective roles (positions). For example, if physical therapist agent 233 determines that exercise therapy or physical therapy is necessary in response to the draft internal treatment plan for the patient formulated by the physician agent, it will propose specific menus (such as range of motion exercises, muscle strengthening, and aerobic exercise for exercise therapy, or thermotherapy, electrical stimulation therapy, and hydrotherapy for physical therapy).

[0047] [Processing flow of the treatment support system] Next, with reference to Figure 3, the processing flow of the treatment support system 10 will be explained. In this explanation, the first internist agent 211 and the second internist agent 212 will be referred to as "cardiologists," and the first surgeon agent 221, the second surgeon agent 222, and the third surgeon agent 223 will be referred to as "cardiac surgeons."

[0048] First, the treatment support system 10 receives patient information entered from terminal 30 via input means 11. The received patient information is then notified to the first internist (cardiologist) agent 211 (step S11).

[0049] Then, the first internist (cardiologist) agent 211 outputs a draft of the patient's internal treatment plan based on the input patient information, as well as whether and / or what additional tests the patient may need.

[0050] Here, the first internist (cardiologist) agent 211 outputs a preliminary plan for the patient's internal treatment using a generative AI model. Specifically, such a preliminary plan can be output by inputting the following prompts into the generative AI model. As a cardiologist, please create a draft of a non-invasive treatment plan (pharmacotherapy) and any additional necessary tests (including invasive tests) based on the patient's basic information, diagnosis, and test results. ...(omitted)..."

[0051] Next, the first pharmacist agent 213 takes the draft output from the first internist (cardiology) agent 211 as input information and modifies the draft from the perspective of pharmacotherapy before outputting it (step S12).

[0052] At this time, by inputting a prompt into the generating AI model such as, "As a drug therapy specialist, please evaluate the drug therapy portion of the treatment plan proposed by the cardiologist and propose any necessary revisions," it is possible to generate such revision proposals.

[0053] Subsequently, the second internist (cardiologist) agent 212 uses the draft output from the first pharmacist agent 213 as input information to output the final treatment plan regarding the patient's internal treatment (step S13).

[0054] At this time, by inputting a prompt into the generating AI model such as, "As a cardiologist, integrate evaluations and revisions from other specialists (pharmacology specialists) and create a final treatment plan (other than surgical treatment)," it is possible to output a final treatment plan regarding internal treatment for such patients.

[0055] Meanwhile, the patient information received by the input means 11 is also notified to the first surgeon (cardiac surgeon) agent 221 (step S21). Then, the first surgeon (cardiac surgeon) agent 221 determines, based on the entered patient information, whether or not surgery is necessary for the patient.

[0056] At this time, by inputting a prompt into the generating AI model such as, "As a cardiac surgeon, please determine the necessity of surgical treatment based on the patient's basic information, diagnosis, test results, and the treatment plan proposed by the cardiologist," the model can be made to determine whether or not surgery is necessary for the patient.

[0057] If the first surgeon (cardiac surgeon) agent 221 determines that surgery is necessary, the second surgeon (cardiac surgeon) agent 222 outputs a draft of the patient's external treatment plan based on the input patient information (step S22). If the first surgeon (cardiac surgeon) agent 221 determines that surgery is unnecessary, the treatment support system 10 will not suggest any invasive treatment methods.

[0058] At this time, by inputting a prompt into the generating AI model such as, "As a cardiac surgeon, please consider a surgical treatment plan based on the patient's basic information, diagnosis results, test results, and treatment plan proposed by the cardiologist," it is possible to output a draft of the patient's external treatment plan.

[0059] Next, the clinical engineering technician agent 224 takes the draft output from the second surgeon (cardiac surgeon) agent 222 as input information and outputs supplementary information from a clinical engineering perspective regarding the equipment and / or technology used in the draft (step S23).

[0060] At this time, by inputting a prompt into the generating AI model such as, "As a clinical engineer, please provide advice on the medical devices and techniques to be used based on the surgical plan considered by the cardiac surgeon," it is possible to add such supplementary information.

[0061] Subsequently, the third surgeon (cardiac surgeon) agent 223 uses the draft output from the clinical engineering technician agent 224 as input information to output the final treatment plan regarding external treatment (step S24).

[0062] At this time, by inputting a prompt into the generating AI model such as, "As a cardiac surgeon, please determine the final surgical treatment plan after receiving advice from a clinical engineer," it is possible to output a final treatment plan regarding external treatments for such patients.

[0063] Finally, the treatment plan aggregation agent 231 takes the final treatment plan regarding the patient's internal treatment output from the second internist (cardiologist) agent 212 and the final treatment plan regarding the patient's external treatment output from the third surgeon (cardiac surgeon) agent 223 as input information to comprehensively determine the patient's final treatment plan (step S31).

[0064] At this time, by inputting a prompt into the generating AI model such as, "As a cardiologist overseeing the patient's treatment plan, please integrate the proposed non-invasive and invasive treatment plans to create a final treatment proposal," it is possible to output such a patient's final treatment plan. Furthermore, by inputting the content that should be output in the prompt (the content you want to be output), it is possible to output the estimated disease name, specific treatment plans for internal treatments (names of necessary drugs, dosages, administration methods, precautions, etc.), specific treatment plans for external treatments (content, purpose, risks, recommendation level, etc.), whether or not additional tests are needed and / or their content, and the overall most recommended treatment plan.

[0065] The final treatment plan formulated by the treatment plan aggregation agent 231 is output to the terminal 30 by the output means 12, and the physician confirms the contents of the final treatment plan from the terminal 30 (step S32).

[0066] Then, if there are any problems with the final treatment plan (and if modifications are needed), the necessary modifications are provided as feedback (Step S33). Known techniques such as Human-in-the-Loop can be used for the feedback.

[0067] On the other hand, if there are no problems with the content of the final treatment plan, the patient explanation agent 232 uses a generating AI model to output an explanatory text from the perspective of a cardiologist or cardiac surgeon, based on the content of the patient's final treatment plan output from the treatment plan aggregation agent 231, for explanation to the patient and / or related parties (step S34). Therefore, doctors (especially young doctors with little experience) can explain to the patient while referring to this explanatory text.

[0068] At this time, by inputting a prompt such as, "As a cardiologist, please create an explanatory document to clearly explain the final treatment plan to the patient and their family," the generation AI model can be made to output an explanatory document for explaining to the patient.

[0069] Through the procedures described above, the treatment support system 10 can formulate a final treatment plan tailored to the individual patient's condition and present it to the physician. At this time, because the treatment support system 10 employs a workflow-type configuration (see Figure 3), the output of one agent can be reviewed and examined by the next agent. In other words, because the treatment support system 10 involves multiple agents with individual roles reviewing and examining the treatment plan, it can be said that it replicates a conference in hospital operations within the system.

[0070] Therefore, physicians can refer to the highly accurate final treatment plan presented by the treatment support system 10 and actually formulate a treatment plan for the patient. As a result, the time and effort required for physicians to devise a treatment plan from scratch is eliminated, thus reducing physicians' working hours.

[0071] In this embodiment, the first internist agent 211, the first pharmacist agent 213, the first surgeon agent 221, the second surgeon agent 222, and the clinical engineer agent 224 refer to information sources such as medical guidelines obtained from the external device 40 and output using search and extended generation.

[0072] By using search-enhanced generation with these agents, it becomes possible to output more accurate and reliable answers based on the latest information and unique data (e.g., a unique hospital manual). On the other hand, the second internist agent 212 and the third surgeon agent 223 output final treatment plans regarding internal and external treatments for the patient, respectively, based on treatment plans formulated by the other agents, and therefore do not use search-enhanced generation in this embodiment. Of course, it is possible to arbitrarily change the design to determine which agents use search-enhanced generation.

[0073] Furthermore, the workflow-type configuration can also be redesigned as needed. For example, Figure 4 shows an alternative processing flow for the treatment support system 10. To explain only the differences from the processing flow shown in Figure 3, in the processing flow shown in Figure 4, the process is the same up to the point where the first internist (cardiologist) agent 211 outputs a draft of the patient's internal treatment plan, but then the first pharmacist agent 213 takes the draft output from the first internist (cardiologist) agent 211 as input information, modifies the draft from the perspective of pharmacotherapy, and outputs it (step S121). Furthermore, the second pharmacist agent 214 takes the draft output from the first pharmacist agent 213 as input information, checks the draft from the perspective of pharmacotherapy, modifies it if necessary, and outputs it (step S122).

[0074] This allows for mutual checking of the usage, dosage, and drug interactions of contraindicated medications, enabling the development of safer final treatment plans for patients' internal treatments.

[0075] Although not described in detail in this explanation, each agent, such as physical therapist agent 233, can be incorporated into the workflow-type configuration as appropriate. Furthermore, while this explanation uses the terms "cardiologist" for the internist agent and "cardiac surgeon" for the surgeon agent, it is certainly not limited to these roles. For example, the internist agent could be assigned roles in gastroenterology, respiratory medicine, neurology, nephrology, etc., and the surgeon agent could be assigned roles in gastrointestinal surgery, neurosurgery, orthopedics, urology, dermatology, etc. [Industrial applicability]

[0076] This invention is a treatment support system that helps physicians provide medical services to patients safely and quickly, while also reducing physicians' working hours. Because it can be used by physicians working in medical institutions nationwide, it is industrially useful. [Explanation of Symbols]

[0077] 10 Treatment support systems 11 Input means 12 Output means 13 Learning Methods 14 Memory means 20 Virtual Medical Personnel Tools 211 First Physician Agent 212 Second Internist Agent 213 First Pharmacist Agent 214 The Second Pharmacist Agent 221 The First Surgeon Agent 222 The Second Surgeon Agent 223 The Third Surgeon Agent 224 Clinical Engineering Technician Agent 231 Treatment Plan Aggregation Agent 232 Patient Information Agent 233 Physical Therapist Agent 234 Occupational Therapist Agent 235 Medical Radiologic Technologist Agent 236 Clinical Laboratory Technologist Agent 30 devices 40 External device

Claims

1. A treatment support system that assists in determining the treatment plan for a patient based on the patient information entered, This is a virtual medical professional means that virtually assumes the role of a medical professional, and has multiple virtual medical professional means (hereinafter referred to as "agents") that output content related to the patient's treatment plan using a generating AI model based on input patient information, including at least the patient's current medical history and the results of tests performed, and instructions including the content to be output. Multiple of the aforementioned agents, An internal medicine physician agent that outputs information about the patient's internal treatment based on the patient's input information, A surgeon agent outputs information regarding the patient's external treatment methods based on the same information as that entered into the physician agent, The system includes a treatment strategy aggregation agent that comprehensively determines the final treatment plan for the patient, using as input information the content regarding internal treatment methods output from the internal physician agent and the content regarding external treatment methods output from the surgical agent, As the aforementioned internal medicine physician agent, A first physician agent outputs a draft of internal medicine treatment for the patient based on the input patient information, A pharmacist agent takes the draft output from the first internist agent as input information and modifies the draft from the perspective of pharmacotherapy and outputs it, The system includes a second internist agent that takes the draft output from the pharmacist agent as input information and outputs a final treatment plan regarding the patient's internal treatment method, As the aforementioned surgical agent, A first surgical agent that determines whether or not surgical intervention is necessary for the patient based on the patient's information entered, If the first surgical agent determines that surgery is necessary, a second surgical agent outputs a draft of an external treatment plan for the patient based on the input patient information. A clinical engineering technician agent takes the draft output from the second surgical agent as input information and outputs supplementary information from a clinical engineering perspective regarding the equipment and / or technology used in the draft, A third surgeon agent, which takes the draft output from the clinical engineering technician agent as input information, outputs a final treatment plan regarding the patient's external treatment, and includes the following: Treatment support system.

2. A treatment support system that assists in determining a treatment plan for a patient based on the patient information entered, This is a virtual medical professional means that virtually assumes the role of a medical professional, and has multiple virtual medical professional means (hereinafter referred to as "agents") that output content related to the patient's treatment plan using a generating AI model based on input patient information, including at least the patient's current medical history and the results of tests performed, and instructions including the content to be output. Multiple of the aforementioned agents, An internal medicine physician agent that outputs information about the patient's internal treatment based on the patient's input information, A surgeon agent outputs information regarding the patient's external treatment methods based on the same information as that entered into the physician agent, The system includes a treatment strategy aggregation agent that comprehensively determines the final treatment plan for the patient, using as input information the content regarding internal treatment methods output from the internal physician agent and the content regarding external treatment methods output from the surgical agent, The aforementioned physician agent includes two or more agents who play the role of a physician, and one or more agents who play the role of a medical professional other than a physician. The aforementioned surgeon agent includes two or more agents that play the role of surgeons and one or more agents that play the role of medical professionals other than surgeons. Treatment support system.

3. The first physician agent, based on the input patient information, further outputs whether or not additional tests are necessary for the patient and / or what those tests are. The treatment support system according to claim 1.

4. The first internist agent, the pharmacist agent, the first surgeon agent, the second surgeon agent, and the clinical engineer agent refer to a database containing one or more of the following: medical guidelines, drug package inserts, past case information, and hospital manuals, and output using search and extended generation. The treatment support system according to claim 1.

5. The aforementioned treatment strategy aggregation agent comprehensively determines the treatment strategy for the patient so that internal and external treatments complement each other. Furthermore, the treatment plan includes specific treatment guidelines for internal treatments, specific treatment guidelines for external treatments, whether or not additional tests are required and / or what those tests entail, and the overall most recommended treatment plan. A treatment support system according to any one of claims 1 to 4.

6. Based on the content of the patient's treatment plan output from the treatment plan aggregation agent, a patient explanation agent is included that uses a generating AI model to output an explanatory text for explaining the treatment plan to the patient and / or related parties from the perspective of a specialist internist or specialist surgeon. A treatment support system according to any one of claims 1 to 4.

Images