Medical information processing device, medical information processing method, and program

Abstract

To enable the appropriate acquisition of patient information that cannot be confirmed through the patient's chief complaint or a doctor's examination. [Solution] The medical information processing device of the embodiment is a medical information processing device that receives questions from a patient and answers on behalf of the patient using a digital twin that reproduces the patient's condition, wherein the digital twin is an AI model that has been trained by adding the patient's information to at least a large-scale language model, and comprises a reception unit, an information acquisition unit, an update unit, an answer unit, and a display control unit. The reception unit receives the question. The information acquisition unit acquires disease condition information that represents at least the patient's condition. The update unit updates the digital twin based on the disease condition information. The answer unit outputs answer information that represents the answer to the question using the digital twin. The display control unit generates display data for presenting the answer information and displays it on a display device.

Description

Technical Field

[0001] The embodiments disclosed in this specification and the drawings relate to a medical information processing apparatus, a medical information processing method, and a program.

Background Art

[0002] Conventionally, in medical institutions such as hospitals, SOAP (Subjective: subjective, Objective: objective, Assessment: evaluation, Plan: plan) is used to address the diseases that patients suffer from. That is, a doctor belonging to a hospital conducts an evaluation (analysis, consideration) based on subjective chief complaint information such as information heard from the patient through face-to-face examination and information described by the patient in a questionnaire, and objective information such as numerical values of the patient's test results and examination results, to make a definitive diagnosis of identifying the disease that the patient suffers from, and to formulate a treatment plan for treating the disease. )

[0003] However, it is conceivable that a judgment error may occur when a doctor evaluates (analyzes, considers) the disease that a patient suffers from, or that there may be confirmation items that require reconfirmation with the patient to identify (definitively diagnose) the disease. The causes of this may include, for example, that the patient has forgotten the content of the chief complaint to be reported, or the doctor has forgotten the items to be confirmed, etc., due to some reason, the necessary content is insufficient. Furthermore, the patient reports as the chief complaint the content of poor physical condition (information such as when it started, what kind of symptoms appeared, and what the frequency was, etc.) to the doctor, but this chief complaint is based on the patient's personal subjectivity (physical condition, mood, pain), and it is considered that factors such as the lack of logical connection with the date, time, actions, etc. also have an impact.

[0004] Therefore, with the conventional SOAP method, the quality of medical care provided to a patient is influenced by the quality of information gathered by the physician from the patient. This is because not only the physician's analysis, consideration (speculation), and identification of the disease, but also the medical care provided by other healthcare professionals such as radiologists, nurses, and rehabilitation specialists, is based on instructions from the physician and the patient information conveyed by the physician. Furthermore, it is time-consuming and laborious for both the physician to reconfirm missing information with the patient and for other healthcare professionals to confirm necessary information with the physician.

[0005] In recent years, there has been research into creating personalized large-scale language models (LLMs) by adding patients' personal information to them, and using these models in the medical field. Furthermore, there is also research into combining this with Retrieval Augmented Generation (RAG) technology, which uses externally acquired information to improve the accuracy and reliability of generative AI (Artificial Intelligence) models. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Patent No. 7441391 [Overview of the Initiative] [Problems that the invention aims to solve]

[0007] The problem that the embodiments disclosed in this specification and drawings aim to solve is to support the diagnosis and medical procedures performed by healthcare professionals by enabling them to suitably obtain patient information that cannot be confirmed through the patient's chief complaint or a physician's examination. However, the problem that the embodiments disclosed in this specification and drawings aim to solve is not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described later can also be positioned as other problems. [Means for solving the problem]

[0008] The medical information processing device of the embodiment is a medical information processing device that receives questions from medical professionals to patients and answers on behalf of the patient using a digital twin that reproduces the patient's condition, wherein the digital twin is an AI model that has been trained by adding the patient's information to at least a large-scale language model, and comprises a reception unit, an information acquisition unit, an update unit, an answer unit, and a display control unit. The reception unit receives the question. The information acquisition unit acquires at least disease condition information representing the patient's condition. The update unit updates the digital twin based on the disease condition information. The answer unit uses the digital twin to output answer information representing the answer to the received question. The display control unit generates display data for presenting the answer information and displays it on a display device. [Brief explanation of the drawing]

[0009] [Figure 1] A diagram showing the configuration of a medical information processing system equipped with a medical information processing device according to an embodiment, and an example of the operating environment for the medical information processing device. [Figure 2] A diagram showing an example of the functional configuration of a medical information processing device according to the embodiment. [Figure 3] A diagram showing an example of a display screen that presents a question and its answer to a medical information processing device according to an embodiment. [Figure 4] A flowchart showing an example of the process flow for updating a patient's digital twin in a medical information processing device according to an embodiment. [Figure 5] A flowchart showing an example of the processing flow when answering a question in a medical information processing device according to the embodiment. [Figure 6] A diagram showing an example of a question and answer to a medical information processing device according to an embodiment. [Modes for carrying out the invention]

[0010] The medical information processing device, medical information processing method, and program of the embodiment will be described below with reference to the drawings.

[0011] A medical information processing device is a device that answers questions from a patient on behalf of the patient, for example, from a physician or other healthcare professional (such as a radiologist, nurse, or rehabilitation specialist who performs medical procedures on the patient based on instructions from the physician and patient information provided by the physician; hereinafter simply referred to as "healthcare professional"). In the following description, when there is no distinction between physicians and healthcare professionals, the term "healthcare professional" will be used. In the following description, a medical information processing system including a medical information processing device that provides answers to questions from physicians and healthcare professionals will be described.

[0012] [Usage Environment for Medical Information Processing Systems] Figure 1 shows an example of the configuration of a medical information processing system equipped with a medical information processing device according to an embodiment, and an example of the operating environment for the medical information processing device. The medical information processing system 1 is configured to include, for example, a medical information processing device 100 installed in a medical institution such as a hospital 10, a patient information collection device 200 installed in the home 20 of an actual patient (hereinafter referred to as "real patient") Pr who is the target of diagnosis, and a server device 30 which includes a patient information storage device 32 that stores various information about the real patient Pr. In the configuration of the medical information processing system 1 shown in Figure 1, an example of the patient information collection device 200 is shown, which includes three data collection devices 210, a life log collection device 220, and a personal data setting device 230.

[0013] The real patients (Pr) targeted for diagnosis include, for example, patients with chronic diseases, patients with lifestyle-related diseases, patients undergoing follow-up after some kind of medical procedure, and patients being observed in preparation for surgery. In addition to patients who are suffering from or have suffered from some kind of disease, the group also includes individuals who have been flagged during a health checkup or healthy individuals who wish to manage their own health, as well as multiple individuals whose health is managed by doctors or other medical professionals.

[0014] In the medical information processing system 1, the medical information processing device 100, the patient information collection device 200, and the server device 30 communicate via a network NW. The network NW is a wireless communication network that includes, for example, the internet, WAN (Wide Area Network), LAN (Local Area Network), provider equipment, and wireless base stations. Figure 1 shows a configuration in which the patient information collection device 200 collects information from one real patient Pr, but there may be multiple real patients Pr in the same home 20. Furthermore, Figure 1 shows a state in which one patient information collection device 200 is connected to the medical information processing device 100 and the server device 30 respectively via the network NW, but there may be multiple patient information collection devices 200, that is, each patient information collection device 200 installed in the home 20 of multiple real patients Pr, and each of them may be connected to the medical information processing device 100 and the server device 30 respectively via the network NW.

[0015] The patient information collection device 200 collects records of the real patient Pr's living conditions (hereinafter referred to as "lifelog"), which are measured and acquired by the data collection device 210. The lifelog includes information and data such as the real patient Pr's weight, heart rate, respiratory rate, blood pressure, diet, travel history, activity status, and activities performed under the guidance of a doctor or medical professional. The patient information collection device 200 transmits the lifelog collected by the data collection device 210 to the server device 30 (which may also be a medical information processing device 100) via the network NW using the lifelog collection device 220. The patient information collection device 200 also transmits the personal data set by the real patient Pr to the personal data setting device 230 to the server device 30 (which may also be a medical information processing device 100) via the network NW.

[0016] Life logs are examples of "medical condition information," "patient information," "daily life records," and "secondary patient information." Personal data are examples of "medical condition information," "patient information," "secondary patient information," and "third patient information."

[0017] Each data acquisition device 210 communicates with the life log collection device 220 and outputs (transmits) the acquired life log of the real patient Pr to the life log collection device 220. The communication between the life log collection device 220 and the data acquisition devices 210 may be, for example, wired communication using a communication cable, or short-range wireless communication such as Bluetooth®. Each data acquisition device 210 outputs (transmits) the acquired life log of the real patient Pr to the life log collection device 220 at predetermined time intervals.

[0018] Figure 1 shows an example of data acquisition equipment 210, which includes a wristwatch-type wearable device 212 owned by the real patient Pr, a vital sign measurement device 214 that measures vital data such as the real patient Pr's body temperature, blood pressure, and electrocardiogram (heart rate), and an imaging device 216 that takes images of the real patient Pr from a predetermined position within the home 20. The wearable device 212, when worn by the real patient Pr, measures the amount of physical activity such as walking and exercise in the real patient Pr's daily life and outputs (transmits) it as a life log to the life log acquisition device 220. The vital sign measurement device 214 measures biometric data including vital data such as the real patient Pr's body temperature, blood pressure, and electrocardiogram (heart rate), and outputs (transmits) it as a life log to the life log acquisition device 220. The vital data may also be measured by the wearable device 212 worn by the real patient Pr. The vital sign measurement device 214 may be, for example, a mobile device that combines the functions of a mobile phone using an existing mobile communication network with the functions of a personal digital assistant (PDA), such as a smartphone or tablet device. The imaging device 216 images the real patient Pr at home 20 and outputs (transmits) the data of the captured image as a life log to the life log collection device 220. The imaging device 216 may perform predetermined image processing on the captured image to analyze, for example, the body temperature of the real patient Pr, and output (transmit) information representing the results of the analysis as a life log to the life log collection device 220. The imaging device 216 may perform predetermined image recognition processing on the captured image to grasp, for example, the eating, drinking, and smoking status of the real patient Pr, and output (transmit) information representing the results of the grasp as a life log to the life log collection device 220. The data collection device 210 shown in Figure 1 is merely an example, and the data collection device 210 includes various devices for acquiring the life log of the real patient Pr. For example, the data acquisition device 210 may be a pedometer (registered trademark) carried by the real patient Pr, or a blood pressure monitor that the real patient Pr uses to spontaneously measure their blood pressure.For example, the data collection device 210 may be a device for checking whether there are various disorders in the real patient Pr, such as a salinity sensor capable of measuring salinity concentration and various sensors capable of measuring urine, feces, saliva, blood, etc. In the life log collected by each data collection device 210 and output to the life log collection device 220, information on the time when the life log was measured (acquired) is associated.

[0019] For example, the data collection device 210 stores the life log measured during the outing of the real patient Pr in a memory (storage unit) not shown, and when the real patient Pr returns home 20 and communication with the life log collection device 220 is established, the life log stored in the memory not shown is output (transmitted) to the life log collection device 220. The memory not shown is realized by, for example, semiconductor memory elements such as ROM (Read Only Memory), RAM (Random Access Memory), and flash memory.

[0020] The data collection device 210 is an example of a "collection device".

[0021] The life log collection device 220 receives the life log output (transmitted) by each data collection device 210 and transmits the received life log to the server device 30 (which may be the medical information processing device 100) via the network NW. The life log collection device 220 is provided with, for example, a storage device (storage unit) not shown, collects the life log output (transmitted) by each data collection device 210, and transmits the collected life log to the server device 30 (which may be the medical information processing device 100) via the network NW at predetermined time intervals. The storage device not shown is realized by, for example, semiconductor memory elements such as ROM, RAM, and flash memory, and a hard disk drive (HDD).

[0022] The life log collection device 220 is an example of a "collection device".

[0023] Personal data is set in the personal data setting device 230 by the real patient Pr. The personal data setting device 230 is a terminal device (for example, a tablet device) that exchanges personal data with the server device 30 (which may be a medical information processing device 100) via a network NW. Personal data is data (information) about the real patient Pr that the doctor (Dr) has not interviewed, such as personal information, past medical history, medications being taken, and allergy information. Figure 1 shows a configuration in which the personal data setting device 230 directly exchanges data and information (personal data) with the server device 30 (which may be a medical information processing device 100) via a network NW, but this is just one example, and the personal data setting device 230 may also output (transmit) the personal data set by the real patient Pr to the life log collection device 220, and the life log collection device 220 may transmit the personal data to the server device 30 (which may be a medical information processing device 100).

[0024] The personal data setting device 230 is an example of a "data collection device".

[0025] The server device 30 is a server device on a network NW, such as one integrated into a medical institution's network or a cloud computing system. The server device 30 stores life logs and personal data transmitted by the patient information collection device 200 and transmitted via the network NW in the patient information storage device 32. More specifically, the server device 30 stores the history of life logs transmitted by the life log collection device 220 in the life log storage device 322, and the history of personal data transmitted by the personal data setting device 230 in the personal data storage device 324. The patient information storage device 32, that is, the life log storage device 322 and the personal data storage device 324, are implemented using semiconductor memory elements such as ROM, RAM, and flash memory, or hard disk drives (HDDs). In response to a request from the medical information processing device 100, the server device 30 transmits the life logs of real patients stored in the life log storage device 322 and the personal data of real patients stored in the personal data storage device 324 to the medical information processing device 100 via the network NW. Figure 1 shows a case where one server device 30 is connected to a network NW, but this is merely an example, and the server device 30 may include multiple server devices connected to the network NW, or it may include other server devices connected to other networks to which the server device 30, the medical information processing device 100, and the patient information collection device 200 can be connected.

[0026] Figure 1 shows a configuration in which the server device 30 stores lifelogs and personal data transmitted by the patient information collection device 200 in the patient information storage device 32 (lifelog storage device 322 or personal data storage device 324). However, the server device 30 may be configured to implement some or all of the functions of the medical information processing device 100, which will be described later. In this case, the main unit of the medical information processing device 100 and the server device 30, which implements some of the functions, communicate with each other via a network NW. Here, if the server device 30 is configured to implement all the functions of the medical information processing device 100, then at least a display device and an input interface must be installed in the hospital 10, and the server device 30, which is the main unit of the medical information processing device 100, will be configured to communicate with the display device and input interface via a network NW.

[0027] The server device 30, and the life log storage device 322 and personal data storage device 324 included in the patient information storage device 32 provided by the server device 30, may be examples of "collection devices".

[0028] The medical information processing device 100 provides answers to questions posed by a physician (including medical professionals) to a real patient Pr, acting on behalf of the real patient Pr. The medical information processing device 100 is implemented by a computer device such as a personal computer (PC) installed in, for example, a consultation room in a hospital 10. Figure 1 shows an example where the medical information processing device 100 is implemented by a personal computer located in hospital 10. When the medical information processing device 100 is implemented by a personal computer, the medical information processing device 100 is connected to an input interface for the physician to operate the medical information processing device 100 and input information (questions), and a display device for presenting information (answers) to the physician. The input interface and display device may be connected to the medical information processing device 100 by wireless communication.

[0029] The input interface can be implemented by, for example, a mouse, keyboard, touch panel, or microphone. If the input interface is a touch panel, it may be formed integrally with a display device connected to the medical information processing device 100. In this specification, the input interface is not limited to those equipped with physical operating components such as the mouse or keyboard described above. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the medical information processing device 100 and outputs this electrical signal to the medical information processing device 100 is also included as an example of an input interface. The display device can be, for example, a liquid crystal display (LCD), a CRT (Cathode Ray Tube) display, or an organic EL (Electroluminescence) display.

[0030] The medical information processing device 100 may be implemented as a computer device such as a terminal device (e.g., a tablet device) carried by a medical professional Dp belonging to the hospital 10. In this case, the medical professional Dp operates the medical information processing device 100 and inputs information (questions) using an input interface built into the terminal device, and the information (answers) is presented to the medical professional Dp by a display device (e.g., an LCD (liquid crystal display)) built into the terminal device. When the medical information processing device 100 is implemented as a terminal device, the input interface may be a microphone built into the terminal device, and the information (answers) may be presented to the medical professional Dp by a speaker built into the terminal device instead of, or in addition to, the display device.

[0031] The medical information processing device 100 may be implemented by a server device (which may be server device 30) on a network NW, an unillustrated network, or incorporated into a cloud computing system. In this case, the examination room only needs to have at least an input interface and a display device installed, and the server device, which is the main unit of the medical information processing device 100, communicates with the input interface and display device via the network NW or an unillustrated network. Furthermore, only some of the functions of the medical information processing device 100, as described later, may be implemented by the server device, in which case the server device, which is the main unit of the medical information processing device 100, and the server device that implements some of the functions communicate with each other via the network NW or an unillustrated network. The unillustrated network, like the network NW, includes, for example, the internet, WAN, LAN, provider equipment, wireless base stations, etc.

[0032] The medical information processing device 100 is not limited to being implemented as an independent device such as a computer or terminal device, but may also be implemented as a function of a medical system composed of other different devices and equipment. Examples of medical systems include electronic medical record systems that manage diagnostic results and test results from previous patient diagnoses and examinations as medical records (electronic medical records), and database systems such as medical image management systems (PACS: Picture Archiving and Communication Systems) that manage medical image data of patients. Furthermore, the medical information processing device 100 and its functions may also be implemented as a function of a medical information integration system that integrates information managed by each medical system and presents it to a physician, or as a measurement application (medical information integration application) to realize that function. In these cases, at least the input interface and display device are shared with devices and equipment such as medical systems that realize the functions of the medical information processing device 100.

[0033] The medical information processing device 100 answers questions about the real patient Pr by using an AI (Artificial Intelligence) model of the real patient Pr, which has been trained by adding the real patient Pr's personal information to a Large Language Model (LLM). The AI ​​model of the real patient Pr is realized, for example, as a digital twin (Patient Twin; hereinafter referred to as "Patient Digital Twin") that reproduces the state of the real patient Pr as a twin in digital space. The AI ​​model of the real patient Pr may also be realized, for example, as a persona that reproduces the characteristics of the real patient Pr. In the following description, the medical information processing device 100 will answer questions about the real patient Pr based on the state of the Patient Digital Twin. The Patient Digital Twin may be generated in the medical information processing device 100 or in the server device 30.

[0034] Figure 1 shows an example where a patient digital twin is generated based on actual information and data about the disease that a real patient Pr is suffering from, including, for example, medical record data stored in an electronic medical record system 12 that manages the chief complaint of the real patient Pr and the medical record (electronic medical record) of the real patient Pr examined by a physician, and examination data stored in an examination management system 14 that manages the results of examinations performed on the real patient Pr. The patient digital twin may also be generated based on medical image data stored in a medical image management system (PACS) that manages the medical images of the real patient Pr. Figure 1 also shows an example where, for example, a physician Dr. who is the attending physician of the real patient Pr, or other medical professionals Dp such as a radiologist R, a nurse N, and a rehabilitation specialist S, each ask questions to the medical information processing device 100, and a digital patient Pd, which schematically represents the patient digital twin in the medical information processing device 100, answers each question. In other words, this illustrates a situation where hospital 10 is always equivalent to having a digital patient Pd, and doctors Dr and healthcare professionals Dp (radiologist R, nurse N, rehabilitation specialist S, etc.) can ask questions to the real patient Pr (actually the digital patient Pd) at any time regarding, for example, the content of the chief complaint that the patient had forgotten, or confirmations that the doctor Dr was unable to confirm initially, and receive answers.

[0035] When medical record data stored in the electronic medical record system 12 or examination data stored in the examination management system 14 are updated, the medical information processing device 100 acquires the updated medical record data and examination data, and updates the patient digital twin based on the acquired medical record data and examination data. Furthermore, when life logs collected by the data collection device 210 or personal data set in the personal data setting device 230 are updated, the medical information processing device 100 acquires the updated life logs and personal data from the server device 30 via the network NW, and updates the patient digital twin based on the acquired life logs and personal data. The updated patient digital twin here is a digital twin that includes not only information and data about the disease that the real patient Pr being diagnosed is suffering from, but also (reproduces) the real patient Pr's living conditions. As a result, the medical information processing device 100 can answer subsequent questions to the real patient Pr based on the state of the updated patient digital twin (a digital twin that reproduces the real patient Pr's living conditions as well). Furthermore, while there are concerns that the chief complaints of real patients may not be logically related to dates, times, and actions, the updated patient digital twin is updated based on a life log that collects records of the real patient's living conditions, thus resulting in a digital twin that does not contain temporal inconsistencies or inaccurate information. In other words, the updated patient digital twin eliminates the possibility of hallucination in the AI ​​model, for example, in a large-scale language model (LLM), and therefore provides more accurate answers.

[0036] If the current patient digital twin cannot answer a question posed by a physician (Dr) or healthcare professional (Dp) to the real patient (Pr), the medical information processing device 100 acquires the current (i.e., latest) medical record data, test data, life logs collected by the data collection device 210 up to that point, and personal data set in the personal data setting device 230. Based on the acquired medical record data, test data, life logs, and personal data, the medical information processing device 100 updates the patient digital twin. Then, based on the updated state of the patient digital twin, the medical information processing device 100 answers the question posed by the physician (Dr) or healthcare professional (Dp) to the real patient (Pr). At this time, the medical information processing device 100 adds (stores) information related to the question posed by the physician (Dr) or healthcare professional (Dp) to the real patient (Pr), such as the answer, to the electronic medical record system 12. In this case, the medical information processing device 100 adds (stores) the additional questions and answers associated with the medical record data already stored in the electronic medical record system 12 (hereinafter referred to as "QA information") to the electronic medical record system 12, distinguishing between the medical record data already stored in the electronic medical record system 12 and the additional questions and answers associated with the questions (hereinafter referred to as "QA information").

[0037] The medical information processing device 100 notifies the real patient Pr of the content of instructions given by a physician (Dr) or medical professional (Dp) to treat the disease. The content of the instructions may include, for example, information regarding post-operative follow-up or post-discharge follow-up for the real patient Pr. Figure 1 schematically shows an example of how the content of instructions given by a physician (Dr) or medical professional (Dp) is transmitted to a terminal device T (e.g., a mobile device) owned by the real patient Pr. The medical information processing device 100 may also notify the real patient Pr of the content of instructions given by a physician (Dr) or medical professional (Dp) by transmitting it to a personal data setting device 230. In this case, the medical information processing device 100 adds (stores) information representing the content of instructions given by a physician (Dr) or medical professional (Dp) to the electronic medical record system 12, for example. In this case, the medical information processing device 100 distinguishes between the medical record data (which may include QA information) already stored in the electronic medical record system 12 and the information representing the additional guidance content (hereinafter referred to as "guidance information"), and adds (stores) it in the electronic medical record system 12.

[0038] [Functional Configuration of Medical Information Processing Devices] Figure 2 shows an example of the functional configuration of the medical information processing device 100 according to the embodiment. Figure 2 also shows the components within the hospital 10, home 20, and server device 30 related to the function of the medical information processing device 100 that uses a patient digital twin to answer questions about the real patient Pr.

[0039] The medical information processing device 100 includes, for example, a processing circuit 110 and an information storage unit 120.

[0040] The information storage unit 120 stores various information and data used when the medical information processing device 100 answers questions from doctors (Dr) and medical professionals (Dp) to real patients (Pr). The information storage unit 120 stores the latest patient digital twin of the real patient (Pr), as well as corresponding medical record data, test data, life logs, and personal data. The information storage unit 120 can be implemented using, for example, semiconductor memory elements such as ROM, RAM, or flash memory, or a hard disk drive (HDD).

[0041] The processing circuit 110 performs processes such as an information acquisition function 112, a question reception function 113, a digital twin processing function 114, a display control function 115, and a notification control function 116. The information acquisition function 112 performs processes such as a patient information acquisition function 1122 and an examination information acquisition function 1124. The digital twin processing function 114 performs processes such as a difference information extraction function 1142, a missing information extraction function 1144, a digital twin update function 1146, and a digital twin response function 1148.

[0042] The functional configuration of the processing circuit 110 shown in Figure 2 is the configuration assuming that the latest patient digital twin of the real patient Pr is stored in the information storage unit 120, that is, that a patient digital twin that reproduces the state of the real patient Pr has already been generated. If a patient digital twin of the real patient Pr has not been generated, it will be newly generated in the medical information processing device 100, the server device 30, or other information processing device. When a patient digital twin of the real patient Pr is newly generated in the medical information processing device 100, for example, a patient digital twin generation function will be added to the functional configuration of the processing circuit 110 shown in Figure 2. However, the generation of the patient digital twin in this patient digital twin generation function can be carried out using existing technologies, for example, by adding the personal information of the real patient Pr to a large-scale language model (LLM) to generate an AI model (this may be combined with RAG (Retrieval Augmented Generation) technology). Therefore, a detailed explanation of the functional configuration that realizes the function of generating a patient digital twin in the medical information processing device 100 and the processing circuit 110, as well as its operation and processing, will be omitted.

[0043] The processing circuit 110 implements the following functions by having a hardware processor execute a program (software) stored in a memory (storage unit) not shown: information acquisition function 112 (including patient information acquisition function 1122 and examination information acquisition function 1124), question acceptance function 113, digital twin processing function 114 (including difference information extraction function 1142, missing information extraction function 1144, digital twin update function 1146 and digital twin response function 1148), display control function 115, and notification control function 116. The memory not shown is implemented by semiconductor memory elements such as ROM, RAM, and flash memory, a hard disk drive (HDD), or an optical disc.

[0044] A hardware processor refers to circuits such as CPUs (Central Processing Units), GPUs (Graphics Processing Units), LSIs (Large Scale Integrations), SOCs (System on Chips), Application Specific Integrated Circuits (ASICs), and programmable logic devices (e.g., Simple Programmable Logic Devices (SPLDs), Complex Programmable Logic Devices (CPLDs), and Field Programmable Gate Arrays (FPGAs)). Instead of storing programs in memory (not shown), the hardware processor may be configured to directly embed programs within its circuitry. In this case, the hardware processor implements its functions by reading and executing the programs embedded within the circuitry. A hardware processor is not limited to being configured as a single circuit; it may also be composed of multiple independent circuits combined to implement various functions. Multiple components may be integrated into a single hardware processor to implement various functions. Multiple components may also be incorporated into a single dedicated LSI to implement various functions. Here, the program (software) may be stored in advance in a storage device that constitutes a storage device such as a semiconductor memory element like ROM, RAM, or flash memory, or a hard disk drive (HDD) (a storage device equipped with a non-transient storage medium), or it may be stored in a removable storage medium (non-transient storage medium) such as a DVD or CD-ROM, and installed in a storage device (not shown) provided in the medical information processing device 100 when the storage medium is mounted in a drive device provided in the medical information processing device 100. The program (software) may also be downloaded in advance from another computer device via a network (not shown) and installed in the storage device provided in the medical information processing device 100.A program (software) installed in the storage device of the medical information processing device 100 may be transferred to and executed by the processing circuit 110 of the medical information processing device 100.

[0045] The information acquisition function 112 acquires information and data about real patients (medical record data, life logs, test data, personal data), and outputs the acquired information and data to the digital twin processing function 114.

[0046] The patient information acquisition function 1122 acquires information and data (which are also medical record data) about the chief complaint and the medical condition of a real patient entered by a physician (Dr), as well as medical record data stored in the electronic medical record system 12. The chief complaint of a real patient (Pr) includes, for example, information and data describing the nature of the physical ailment that led the real patient (Pr) to seek medical attention (information and data such as when the symptoms started, what kind of symptoms they were, and how frequently they occurred). Information and data about the medical condition examined by the physician (Dr) includes, for example, information and data describing the appearance of the real patient (Pr) as observed by the physician (Dr), and information and data about the results of the examination conducted by the physician (Dr). Figure 2 schematically shows, for example, the state in which medical record data is acquired from both the physician (Dr) and the electronic medical record system 12. The patient information acquisition function 1122 acquires life logs and personal data from the patient information storage device 32 (more specifically, the life log storage device 322 and the personal data storage device 324). Figure 2 schematically shows, for example, the state in which life logs and personal data are acquired from the life log storage device 322 and personal data storage device 324, respectively, which are included in the patient information storage device 32. The patient information acquisition function 1122 outputs the acquired medical record data, life logs, and personal data (hereinafter, when medical record data, life logs, and personal data are not distinguished, they will be referred to as "patient information") to the digital twin processing function 114. The patient information acquisition function 1122 may also store the acquired patient information in the information storage unit 120 and notify the digital twin processing function 114 of this, so that the digital twin processing function 114 acquires the patient information from the information storage unit 120.

[0047] The examination information acquisition function 1124 acquires examination information related to examinations performed on a real patient Pr. The examination information includes, for example, information on procedures performed by a physician (Dr) on a real patient Pr in the examination room, and examination data stored in the examination management system 14 (for example, numerical values ​​for each examination item performed on a real patient Pr). The examination information may also include, for example, medical image data of a real patient Pr taken by a medical imaging diagnostic device, such as CT images taken by a computed tomography (CT) device, MR images taken by a magnetic resonance imaging (MRI) device, and ultrasound images (echo images) taken by an ultrasound diagnostic device, which are stored in a medical image management system (PACS), as well as information representing findings diagnosed by a physician (Dr) or radiologist based on the medical images. Figure 2 schematically shows, for example, the state in which examination data is acquired from the examination management system 14. The examination information acquisition function 1124 outputs the acquired examination information to the digital twin processing function 114. The inspection information acquisition function 1124 may store the acquired inspection information in the information storage unit 120 and notify the digital twin processing function 114 of this, so that the digital twin processing function 114 can acquire the inspection information from the information storage unit 120.

[0048] The information acquisition function 112, the patient information acquisition function 1122, and the test information acquisition function 1124 are examples of the "information acquisition unit." Medical record data are examples of "medical condition information," "patient information," and "first patient information." Test data are examples of "medical condition information," "patient information," and "test information."

[0049] The information acquisition function 112 may output patient information acquired by the patient information acquisition function 1122 and test information acquired by the test information acquisition function 1124 to the display control function 115. The information acquisition function 112 may store patient information and test information in the information storage unit 120 and notify the display control function 115 of this, so that the display control function 115 can acquire patient information and test information from the information storage unit 120. In this case, when the medical information processing device 100 answers questions from doctors (Dr) or medical professionals (Dp), the display control function 115 will be able to present patient information and test information to the questioner.

[0050] The question reception function 113 receives questions from physicians (Dr) and medical professionals (Dp) regarding real patients (Pr). Figure 2 schematically shows, for example, the reception of questions from radiologists (R) regarding real patients (Pr). The method of receiving questions in the question reception function 113 may be any method. For example, physicians (Dr) and medical professionals (Dp) may input the question content into the medical information processing device 100 by operating the input interface. For example, if the voice spoken by physicians (Dr) and medical professionals (Dp) can be picked up by a microphone installed in the examination room and connected to the medical information processing device 100, or by a microphone built into the terminal device, the question reception function 113 may accept the voice information spoken by physicians (Dr) and medical professionals (Dp) as information about the question content. In this case, the question reception function 113 receives the question content based on the voice spoken by physicians (Dr) and medical professionals (Dp) using, for example, the large-scale language model (LLM) provided by the patient digital twin or the natural language processing function of another large-scale language model (LLM: not shown). The question reception function 113 outputs information and data representing the content of the received question (hereinafter referred to as "question information") to the digital twin processing function 114, the information acquisition function 112, and the display control function 115, respectively. The question reception function 113 may store the received question information in the information storage unit 120 and notify the digital twin processing function 114, the information acquisition function 112, and the display control function 115 of this, so that the digital twin processing function 114, the information acquisition function 112, and the display control function 115 each acquire the question information from the information storage unit 120. In the medical information processing device 100, the question information is added (stored) in the electronic medical record system 12 as the question content in the QA information by the information acquisition function 112.

[0051] As described above, the medical information processing device 100 notifies the real patient Pr of instructions from, for example, a doctor (Dr) or a healthcare professional (Dp). This instruction may also be received by the question receiving function 113. In other words, the question receiving function 113 may receive not only questions from doctors (Dr) or healthcare professionals (Dp) to the real patient Pr, but also instructions for the real patient Pr. In this case, the question receiving function 113 may, for example, use the natural language processing function of the large-scale language model (LLM) described above to determine whether the voice spoken by the doctor (Dr) or healthcare professional (Dp) is a question or instruction for the real patient Pr. When the question receiving function 113 receives instructions for the real patient Pr, it outputs information and data representing the content of this instruction (instruction information) to the notification control function 116 and the information acquisition function 112, respectively. The question receiving function 113 may store the received guidance information in the information storage unit 120 and notify the notification control function 116 and the information acquisition function 112 of this, so that the notification control function 116 and the information acquisition function 112 each acquire the guidance information from the information storage unit 120. As a result, in the medical information processing device 100, the guidance information is notified to the real patient Pr by the notification control function 116, and the guidance information is added (stored) in the electronic medical record system 12 by the information acquisition function 112.

[0052] The question reception function 113 is an example of a "reception section".

[0053] The digital twin processing function 114 performs processing to answer questions directed to the real patient Pr using the patient digital twin. The digital twin processing function 114 answers the question directed to the real patient Pr, as represented by the question information output by the question receiving function 113, using the current patient digital twin if, for example, it can be answered using the patient digital twin currently stored in the information storage unit 120. If the current patient digital twin cannot be used to answer the question directed to the real patient Pr, the digital twin processing function 114 updates the current patient digital twin based on the patient information and examination information output by the information acquisition function 112, and then answers the question using the updated patient digital twin. More specifically, if the current patient digital twin cannot be used to answer the question directed to the real patient Pr, the digital twin processing function 114 first updates the current patient digital twin based on the medical record data and examination data acquired by the information acquisition function 112, and then answers the question using the updated patient digital twin. If the updated patient digital twin is unable to answer questions about the real patient, the digital twin processing function 114 further updates the patient digital twin based on the life log and personal data acquired by the information acquisition function 112, and then uses the updated patient digital twin to answer the questions. The digital twin processing function 114 outputs information and data representing the answers to the questions about the real patient (hereinafter referred to as "answer information") to the display control function 115, presenting the answers to the questioner. In this way, the medical information processing device 100 presents answers to questions about the real patient's chief complaint and the condition of the real patient (such as illness) that could not be confirmed by the doctor's examination, thereby supporting the diagnosis and medical procedures performed by doctors and healthcare professionals on the real patient.

[0054] The difference information extraction function 1142 extracts the differences between the information and data acquired by the information acquisition function 112, such as medical record data, test data, life logs, and personal data, and the corresponding information and data stored in the information storage unit 120. The difference information extraction function 1142 adds (stores) the extracted difference information (difference information) to the information acquisition function 112 and also notifies the digital twin update function 1146.

[0055] The difference information extraction function 1142 is an example of a "difference extraction unit." Information and data such as medical record data, test data, life logs, and personal data acquired by the information acquisition function 112 are examples of "new medical condition information." Information and data such as medical record data, test data, life logs, and personal data stored in the information storage unit 120 are examples of "previous medical condition information."

[0056] The missing information extraction function 1144 determines whether the current patient digital twin can answer the question content to the real patient Pr represented by the question information output by the question reception function 113. If the missing information extraction function 1144 determines that it can answer the question, it outputs the question content to the digital twin answer function 1148. On the other hand, if it determines that it cannot answer the question, the missing information extraction function 1144 extracts the missing information or data (missing information) that is necessary to answer the question. More specifically, the missing information extraction function 1144 extracts the missing information by comparing the information or data corresponding to the question content with information or data such as medical record data, test data, life log, and personal data stored in the information storage unit 120. The missing information extraction function 1144 then instructs the information acquisition function 112 to acquire the latest medical record data, test data, life log, and personal data. As a result, the information acquisition function 112 (more specifically, the patient information acquisition function 1122 and the test information acquisition function 1124) acquires the latest patient information and test information. The missing information extraction function 1144 notifies the digital twin update function 1146 that there is missing information or data (missing information) to answer the current question.

[0057] The missing information extraction function 1144 is an example of a "missing information extraction unit".

[0058] The digital twin update function 1146 updates the patient digital twin stored in the information acquisition function 112 based on the medical record data, test data, life log, and personal data stored in the information acquisition function 112 when it is notified that the difference information extraction function 1142 has added (stored) difference information to the information acquisition function 112, or when it is notified that the information acquisition function 112 has stored the latest patient information or test information in the information acquisition function 112. In other words, the digital twin update function 1146 updates the patient digital twin stored in the information acquisition function 112 to a patient digital twin that includes newer medical record data, test data, life log, and personal data. The digital twin update function 1146 stores the updated patient digital twin in the information acquisition function 112. The digital twin update function 1146 notifies the digital twin response function 1148 that the patient digital twin has been updated.

[0059] The digital twin update function 1146 is an example of an "update unit".

[0060] The digital twin response function 1148 uses the patient digital twin stored in the information acquisition function 112 to answer the questions directed to the real patient Pr represented by the question information output by the question reception function 113. The digital twin response function 1148 outputs information and data (answer information) representing the answers to the questions to the display control function 115. Furthermore, the digital twin response function 1148 outputs the answer information to the information acquisition function 112. The digital twin response function 1148 may also store the answer information in the information storage unit 120 and notify the information acquisition function 112 of this, so that the information acquisition function 112 can acquire the answer information from the information storage unit 120. In the medical information processing device 100, the answer information is added (stored) in the electronic medical record system 12 as an answer in the QA information, associated with the corresponding question content, by the information acquisition function 112.

[0061] The digital twin response function 1148 is an example of a "response section".

[0062] The display control function 115 generates a display image to present to the physician Dr. or medical professional Dp who asked the question to the real patient Pr, using information and data (question information) representing the content of the question received by the question receiving function 113 and information and data (answer information) representing the answer to the question output by the digital twin processing function 114 (more specifically, the digital twin answer function 1148). The display control function 115 presents the content of the question and the answer to it to the physician Dr. or medical professional Dp by displaying the generated display image on a display device. Figure 2 schematically shows, for example, the state in which a question is answered by displaying the display image of the digital patient Pd on a display device 101 (for example, an LCD (liquid crystal display)) connected to the medical information processing device 100.

[0063] The method of presenting the question content and its answer in the display control function 115 to a physician (Dr) or healthcare professional (Dp) is not limited to the method of displaying the digital patient (Pd) display image shown in Figure 2. For example, the display control function 115 may be configured so that a chatbot automatically answers the question.

[0064] Here, an example of the display by the display control function 115 in this case will be described. Figure 3 is a diagram showing an example of a display screen that presents a question and its answer to the medical information processing device 100 according to the embodiment. Figure 3 shows an example of a display screen IM1 when the display image generated by the display control function 115 is displayed on a display device such as an LCD (liquid crystal display) incorporated into a tablet terminal TB carried by a doctor Dr (=XXyy). More specifically, the display screen IM1 shown in Figure 3 shows an example where, when the doctor Dr has entered the question "On average, what time do you go to bed every day, and how many hours do you sleep?" in natural language into the question input area A1, the digital patient Pd (=AAbb), which schematically represents the patient digital twin, answers in natural language, "On average, I go to bed at 24:00 and get 6.5 hours of sleep." Furthermore, the display screen IM1 shown in Figure 3 indicates that the questions and answers in this instance are responses from the patient's digital twin, which was updated using life logs. In other words, this information was not previously stored as medical record data in the electronic medical record system 12, and therefore, this QA information has been added to the electronic medical record system 12. Moreover, the display screen IM1 shown in Figure 3 shows an example of displaying update information notifying that data regarding "medications prescribed at other hospitals" and "weight changes according to the weight loss guidance" obtained from the life logs acquired when updating the patient's digital twin have been added to the electronic medical record system 12. More specifically, the display screen IM1 shown in Figure 3 shows update information indicating that the information "visited HH Hospital 3 days ago (MM / DD) for treatment and was prescribed eye drops" has been added to the electronic medical record system 12, and that data regarding "weight changes according to the weight loss guidance" has been added to the electronic medical record system 12 as "a graph showing the trend of weight changes due to the real patient Pr's efforts to lose weight according to the guidance." Furthermore, the display screen IM1 shown in Figure 3 shows an example of buttons that a physician (Dr.) can operate (for example, by tapping or long-pressing) when checking information and data (including updated information) from the electronic medical record system 12 and the laboratory management system 14.This allows physicians (or healthcare professionals) to obtain information and data regarding the questions they want to ask when making a diagnosis, even when the actual patient is not present, meaning they cannot ask questions directly.

[0065] As described above, the medical information processing device 100 may, for example, receive questions from a doctor or medical professional to a real patient via voice. In this case, the display control function 115 generates voice data (which is also answer information) for answering the questions aloud, in place of or in addition to the display image described above. The display control function 115 then presents the answer to the question to the doctor or medical professional by having the generated voice data spoken through a speaker.

[0066] The display control function 115 is an example of a "display control unit." The displayed image is an example of "display data."

[0067] Returning to Figure 1, the notification control function 116 notifies the real patient Pr of information and data (instruction information) representing the instruction content received by the question acceptance function 113, for example. The notification control function 116 notifies the real patient Pr of the instruction information by sending, for example, an email to the real patient Pr's terminal device T. Figure 2 schematically shows the state in which instruction information is notified to the real patient Pr by sending, for example, an email to the real patient Pr's terminal device T.

[0068] The method of notifying guidance information in the notification control function 116 is not limited to the method of transmitting guidance information to the terminal device T shown in Figure 2. For example, the notification control function 116 may generate a display image to present the guidance content represented by the guidance information to the real patient Pr, and transmit the generated display image to the personal data setting device 230 via the network NW. In this case, the display image transmitted via the network NW is displayed on a display device (e.g., an LCD (liquid crystal display)) built into the personal data setting device 230, thereby presenting (notifying) the notification content to the real patient Pr.

[0069] [Processing by medical information processing equipment] Next, we will describe an example of the processing flow in the medical information processing device 100.

[0070] As described above, when medical record data stored in the electronic medical record system 12, examination data stored in the examination management system 14, life logs collected by the data collection device 210, or personal data set in the personal data setting device 230 are updated, the medical information processing device 100 updates the patient digital twin based on the updated information and data. First, an example of the processing flow of the medical information processing device 100 when updating the patient digital twin based on updated information and data will be described. Figure 4 is a flowchart showing an example of the processing flow for updating the patient digital twin in the medical information processing device 100 according to this embodiment.

[0071] The processing circuit 110 of the medical information processing device 100 checks whether medical record data or examination data has been updated (step S100). If, in step S100, it is confirmed that either or both of the medical record data and / or examination data have been updated, the processing circuit 110 acquires either or both of the updated medical record data and / or examination data using the information acquisition function 112 (step S102). More specifically, the updated medical record data is acquired by the patient information acquisition function 1122 within the processing circuit 110, and the updated examination data is acquired by the examination information acquisition function 1124 within the processing circuit 110. The information acquisition function 112 then outputs either or both of the acquired medical record data and / or examination data to the digital twin processing function 114, and the processing circuit 110 proceeds to step S120.

[0072] On the other hand, if it is confirmed in step S100 that the medical record data and examination data have not been updated, the processing circuit 110 checks whether the life log and personal data have been updated (step S110). If it is confirmed in step S110 that neither the life log nor the personal data have been updated, the processing circuit 110 returns to step S100 and repeats the processing in step S100 (which may include the processing in step S110) to continue checking whether one or more pieces of information or data from the medical record data, examination data, life log, and personal data have been updated.

[0073] On the other hand, if the processing circuit 110 confirms that the life log or personal data has been updated in step S110, the processing circuit 110 acquires either or both of the updated life log and personal data using the information acquisition function 112 (step S112). More specifically, the patient information acquisition function 1122 acquires either or both of the updated life log and personal data from the server device 30 (life log storage device 322 and personal data storage device 324 included in the patient information storage device 32) via the network NW. The information acquisition function 112 then outputs either or both of the acquired life log and personal data to the digital twin processing function 114, and the processing circuit 110 proceeds to step S120.

[0074] In the processing of step S100 or step S110, if it is confirmed that one or more pieces of information or data from medical record data, examination data, life log, and personal data have been updated, the processing circuit 110 uses the digital twin processing function 114 (more specifically, the difference information extraction function 1142) to extract the differences (difference information) between the acquired information or data and the corresponding information or data stored in the information storage unit 120 (step S120). The difference information extraction function 1142 adds (stores) the difference information to the information acquisition function 112 and notifies the digital twin update function 1146.

[0075] When the difference information extraction function 1142 notifies the information acquisition function 112 that difference information has been added (stored), the processing circuit 110 uses the digital twin processing function 114 (more specifically, the digital twin update function 1146) to update the patient digital twin stored in the information acquisition function 112 based on one or more pieces of information or data, including the updated medical record data, examination data, life log, and personal data, stored in the information acquisition function 112 (step S130). The digital twin update function 1146 stores the updated patient digital twin in the information acquisition function 112 and notifies the digital twin response function 1148. The processing circuit 110 then completes the process of this flowchart, updating the patient digital twin based on the updated information and data.

[0076] In this way, the processing circuit 110 within the medical information processing device 100 updates the patient digital twin based on the updated information and data. As a result, the digital twin processing function 114 (more specifically, the digital twin update function 1146) within the processing circuit 110 of the medical information processing device 100 can use the newer state of the patient digital twin stored in the information acquisition function 112 to answer questions about the real patient Pr received by the question reception function 113.

[0077] Next, we will describe an example of the processing flow of the medical information processing device 100 when a physician (Dr) or medical professional (Dp) asks a question about a real patient (Pr). Figure 5 is a flowchart showing an example of the processing flow when the medical information processing device 100 according to this embodiment provides an answer to a question.

[0078] The processing circuit 110 of the medical information processing device 100 checks whether or not a question has been received in the question reception function 113 (step S200). If it is confirmed in step S200 that no question has been received, the processing circuit 110 repeats the confirmation process in step S200.

[0079] On the other hand, in step S200, if it is confirmed that a question has been received, the question reception function 113 outputs question information representing the content of the received question to at least the digital twin processing function 114. Then, the processing circuit 110 uses the digital twin processing function 114 (more specifically, the missing information extraction function 1144) to determine whether it is possible to answer the question content for the real patient Pr represented by the question information output by the question reception function 113 using the current patient digital twin stored in the information storage unit 120 (step S210). In step S210, if it is determined that it is possible to answer the question content using the current patient digital twin, the missing information extraction function 1144 outputs the content of the question to the digital twin answer function 1148 and proceeds to step S280.

[0080] On the other hand, if in step S210 it is determined that the current patient digital twin cannot be used to answer the questions, the missing information extraction function 1144 extracts the missing information or data (missing information) necessary to answer the questions (step S220). The missing information extraction function 1144 then instructs the information acquisition function 112 to acquire the latest medical record data and test data, and notifies the digital twin update function 1146 that there is missing information or data (missing information) necessary to answer the questions.

[0081] The processing circuit 110 acquires the latest medical record data and examination data respectively using the information acquisition function 112 (step S230). More specifically, the latest medical record data is acquired by the patient information acquisition function 1122 within the processing circuit 110, and the latest examination data is acquired by the examination information acquisition function 1124 within the processing circuit 110. The information acquisition function 112 then outputs the acquired latest medical record data and examination data respectively to the digital twin processing function 114.

[0082] The processing circuit 110 extracts the differences (difference information) between the latest acquired medical record data and examination data and the medical record data and examination data stored in the information storage unit 120, respectively, using the digital twin processing function 114 (more specifically, the difference information extraction function 1142) (step S240). The difference information extraction function 1142 adds (stores) the difference information to the information acquisition function 112 and notifies the digital twin update function 1146.

[0083] When the difference information extraction function 1142 notifies the information acquisition function 112 that difference information has been added (stored), the processing circuit 110 updates the patient digital twin stored in the information acquisition function 112 based on the latest medical record data and examination data stored in the information acquisition function 112, respectively, using the digital twin processing function 114 (more specifically, the digital twin update function 1146) (step S250). The digital twin update function 1146 stores the updated patient digital twin in the information acquisition function 112 and notifies the missing information extraction function 1144.

[0084] The processing circuit 110 uses the digital twin processing function 114 (more specifically, the missing information extraction function 1144) to determine whether it can answer the question content for the real patient Pr represented by the question information output by the question acceptance function 113, using the updated patient digital twin stored in the information storage unit 120 (step S260). In step S260, if it is determined that the question content can be answered using the updated patient digital twin, the missing information extraction function 1144 outputs the current question content to the digital twin answer function 1148, and the processing proceeds to step S280.

[0085] On the other hand, in step S260, if it is determined that the updated patient digital twin cannot be used to answer the questions, the missing information extraction function 1144 instructs the information acquisition function 112 to acquire the latest life log and personal data, respectively. The processing circuit 110 then acquires the latest life log and personal data, respectively, using the information acquisition function 112 (step S270). More specifically, the patient information acquisition function 1122 acquires the latest life log and personal data, respectively, from the server device 30 (life log storage device 322 and personal data storage device 324 included in the patient information storage device 32) via the network NW. The information acquisition function 112 then outputs the acquired latest life log and personal data, respectively, to the digital twin processing function 114, and the processing circuit 110 returns to step S240.

[0086] In the processing of step S240, the processing circuit 110 uses the digital twin processing function 114 (more specifically, the difference information extraction function 1142) to extract the differences (difference information) between the latest acquired life log and personal data and the life log and personal data stored in the information storage unit 120. The difference information extraction function 1142 then adds (stores) the difference information to the information acquisition function 112 and notifies the digital twin update function 1146.

[0087] When the difference information extraction function 1142 notifies the information acquisition function 112 that difference information has been added (stored), the processing circuit 110, in the processing of step S250, updates the patient digital twin stored in the information acquisition function 112 again based on the latest life log and personal data stored in the information acquisition function 112, respectively, using the digital twin processing function 114 (more specifically, the digital twin update function 1146). The digital twin update function 1146 stores the updated patient digital twin in the information acquisition function 112 and notifies the missing information extraction function 1144.

[0088] In step S260, the processing circuit 110 uses the digital twin processing function 114 (more specifically, the missing information extraction function 1144) to determine whether it can answer the question to the real patient Pr represented by the question information output by the question acceptance function 113, using the updated patient digital twin stored in the information storage unit 120. If it is determined in step S260 that it can answer the question using the updated patient digital twin, the missing information extraction function 1144 outputs the question to the digital twin answer function 1148, and the process proceeds to step S280.

[0089] In step S210 or step S220, if the processing circuit 110 determines that it can answer the question using the current or updated (including further updates) patient digital twin, the processing circuit 110 uses the digital twin processing function 114 (more specifically, the digital twin answering function 1148) to answer the question for the real patient Pr represented by the question information received by the question receiving function 113 (step S280). The processing circuit 110 then terminates the processing of this flowchart for answering the received question for the real patient Pr.

[0090] In this way, the processing circuit 110 within the medical information processing device 100 updates the patient digital twin as needed and answers questions received from the real patient Pr. As a result, the digital twin processing function 114 (more specifically, the digital twin update function 1146) within the processing circuit 110 of the medical information processing device 100 can use the latest state of the patient digital twin to answer questions received from the real patient Pr by the question receiving function 113.

[0091] The flowchart described above illustrates an example where the processing circuit 110 updates the patient digital twin in two stages. However, the patient digital twin may also be updated in a single stage. That is, if the processing circuit 110 determines in step S210 that it cannot answer the question using the current patient digital twin, it may update the patient digital twin by acquiring the latest patient information and examination information (medical record data, examination data, life log, and personal data). In this case, the processing should be equivalent to the processing in the flowchart described above. Therefore, a detailed explanation of the processing of the processing circuit 110 in the medical information processing device 100 in this case is omitted.

[0092] Incidentally, even if the patient digital twin is updated, there may be cases where it cannot answer questions from doctors (Dr) or healthcare professionals (Dp). In this case, the processing circuit 110 may output the question content to, for example, the notification control function 116, and the notification control function 116 will notify the real patient (Pr). As a result, if a response is obtained from the real patient (Pr), doctors (Dr) or healthcare professionals (Dp) can obtain information and data regarding the content they want to confirm when making a diagnosis (question content), even when the real patient (Pr) is not present at the hospital.

[0093] In this way, the medical information processing device 100 uses the patient digital twin to answer questions from healthcare professionals (doctors or healthcare professionals) to the real patient Pr when diagnosing the illness the real patient Pr is suffering from. The medical information processing device 100 then acquires patient information (which may include examination information) recorded from the real patient Pr's living conditions and updates the patient digital twin to a more up-to-date state. As a result, the medical information processing device 100 can obtain information and data related to the content of questions that doctors or healthcare professionals want to confirm when diagnosing the illness the real patient Pr is suffering from, even when the real patient Pr is not present at the hospital, that is, when they cannot directly ask the real patient Pr questions. This allows doctors and healthcare professionals using the medical information processing device 100 to more effectively diagnose and examine the illness the real patient Pr is suffering from, and to provide more appropriate treatment and medical care to the real patient Pr.

[0094] By the way, the above explanation of the medical information processing device 100 assumed that after a real patient Pr has returned home, there may be matters that need to be reconfirmed with the real patient Pr, or additional matters that need to be confirmed. However, it is also possible to ask questions to the medical information processing device 100, for example, before the real patient Pr arrives at the hospital.

[0095] Here, an example of this case will be explained. Figure 6 is a diagram showing an example of a question and answer to the medical information processing device 100 according to the embodiment. In Figure 6, the display screen IM2, when the display image generated by the display control function 115 is displayed on the display device, shows the question and answer arranged in chronological order. In other words, the display screen IM2 shown in Figure 6 shows more information about the exchange between the question and answer than the range of the display image that can actually be displayed on the display device. The exchange between the question and answer in the display screen IM2 shown in Figure 6 is performed by voice input to a microphone and speech output from a speaker, for example. However, for the sake of ease of explanation, Figure 6 shows an example in which the voice is converted to text and displayed on the display screen IM2, for example, by a speech recognition function in a large-scale language model (LLM) provided by the patient digital twin. In the actual display on the display device, a portion of the display image that can be displayed on the display device is shown by moving sequentially from the top to the bottom of the page. In other words, the question and answer shown on the display screen IM2 are displayed by sequential scrolling. Furthermore, the display screen IM2 shown in Figure 6 schematically shows the emotions of the doctor Dr (=XXyy) asking the question and the real patient Pr (=AAbb) who came to the outpatient clinic.

[0096] The display screen IM2 shown in Figure 6 illustrates an example where, because the next consultation is with a real patient Pr (=AAbb), the medical information processing device 100 is checking the current status of the "weight loss guidance given to Pr three months ago" and the "visit history since the last visit" before the real patient Pr arrives. More specifically, the doctor first asks in natural language, "What is the status of the weight loss guidance given at the last visit?", and the digital patient Pd replies in natural language, "I have successfully lost weight as instructed last time." The display screen IM2 shows an example where data related to this, such as a "graph showing the progress since the last guidance," and a notification indicating that the data from that graph has been added to the electronic medical record system 12, is displayed. Next, the doctor asks in natural language, "Have you visited any other hospitals since your last visit?" and the digital patient Pd replies in natural language, "I visited HH Hospital 3 days ago (MM / DD) and was prescribed eye drops." This is an example of a situation where a notification is issued indicating that "information regarding the treatment and prescription at HH Hospital" has been added (added: stored) in the electronic medical record system 12.

[0097] Furthermore, below the display screen IM2 shown in Figure 6, an example of direct interaction between a physician (Dr) and a real patient (Pr) during an outpatient visit, based on information and data obtained in advance by the medical information processing device 100, is shown. More specifically, an example is shown where the physician (Dr) acknowledges that the real patient (Pr) is "successfully losing weight" based on the information that the patient has "successfully lost weight," and provides information on "precautions when using the medication prescribed at the previous visit and the eye drops prescribed at HH Hospital" based on the information regarding "treatment and prescriptions at HH Hospital three days ago."

[0098] Thus, with the medical information processing device 100, a doctor can obtain answers in advance using the patient's digital twin by asking questions before the real patient Pr arrives at the clinic. This allows the doctor to make diagnoses more efficiently after the real patient Pr arrives at the clinic. As a result, the real patient Pr can receive treatment with greater satisfaction and peace of mind because the doctor is aware of their current situation.

[0099] The above description of the medical information processing device 100 described the case where a real patient Pr can come to hospital 10. However, it is also possible that a real patient Pr may be transported to hospital 10 in an emergency, and that real patient Pr may not be able to present their chief complaint. Even in this case, if hospital 10 has the patient information and examination information of this real patient Pr (one or more of the following: medical record data, examination data, life log, and personal data), the medical information processing device 100 can answer questions from the doctor Dr (including healthcare professionals Dp) on behalf of the real patient Pr, and is therefore considered useful.

[0100] As described above, the medical information processing device of the embodiment uses a patient digital twin to answer questions from healthcare professionals to the patient that they want to confirm when diagnosing the illness the patient is suffering from. The medical information processing device of the embodiment then acquires patient information (which may include examination information) that records the patient's living conditions and updates the patient digital twin to a more up-to-date state. As a result, the medical information processing device of the embodiment allows healthcare professionals to obtain information and data regarding the content they want to confirm (question content) when diagnosing the illness the patient is suffering from, even when the patient is not present (and therefore cannot be asked questions directly). This enables healthcare professionals using the medical information processing device of the embodiment to more effectively diagnose and examine the illness the patient is suffering from, and to provide more appropriate treatment and medical care to the patient.

[0101] Furthermore, the medical information processing device of this embodiment provides answers to questions from medical professionals other than physicians to patients, on behalf of the patient. In other words, medical professionals other than physicians can obtain answers to questions from patients without going through a physician. Therefore, the burden on not only physicians and other medical professionals using the medical information processing device of this embodiment can be reduced, but also on the patients themselves.

[0102] In the embodiments described above, the medical information processing device was explained using an example where it is implemented on a single computer device or a server device on a network (not shown). However, this is merely an example, and the medical information processing device, or the functions it implements, may be implemented by a configuration combining multiple server devices or computer devices. In this case, the functional configuration, operation, and processing of the medical information processing device should be equivalent to those of the medical information processing device in the embodiments described above. Therefore, a detailed explanation of the functional configuration, operation, and processing of the medical information processing device or its functions in this case will be omitted.

[0103] The embodiments described above can be expressed as follows. An AI model that receives questions from medical professionals to patients and learns by adding the patient's information to at least a large-scale language model, and a medical information processing device equipped with processing circuitry that answers on behalf of the patient using a digital twin that reproduces the patient's condition, The aforementioned processing circuit is We have received the above question, At a minimum, obtain disease information representing the patient's medical condition, Based on the aforementioned medical condition information, the digital twin is updated. Using the aforementioned digital twin, answer information representing the answer to the received question is output. To generate display data for presenting the aforementioned response information and display it on a display device, Medical information processing device.

[0104] According to at least one embodiment described above, a medical information processing device (100) receives questions from a medical professional (Dr or Dp) to a patient (real patient Pr) and answers on behalf of the patient using a digital twin (patient digital twin) that reproduces the patient's condition, wherein the digital twin is an AI model trained by adding the patient's information to at least a large-scale language model (LLM), and comprises a receiving unit (113) that receives the questions, and an information acquisition unit (112) that acquires at least disease condition information (patient information and test information) representing the patient's condition, and the disease condition information Based on this, a medical information processing device, a medical information processing method, and a program can be realized that can support diagnoses and medical procedures by medical professionals by having an update unit (1146) that updates the digital twin, an answer unit (1148) that uses the digital twin to output answer information representing the answer to the received question, and a display control unit (115) that generates display data for presenting the answer information and displays it on a display device.

[0105] While several embodiments have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols]

[0106] 1...Medical information processing system, 10...Hospital (medical institution), 100...Medical information processing device, 110...Processing circuit, 112...Information acquisition function, 1122...Patient information acquisition function, 1124...Examination information acquisition function, 113...Question reception function, 114...Digital twin processing function, 1142...Difference information extraction function, 1144...Missing information extraction function, 1146...Digital twin update function, 1148...Digital twin response function, 115...Display controller Function, 116...Notification control function, 120...Information storage unit, 20...Home, 200...Patient information collection device, 210...Data collection device, 212...Wearable device, 214...Vital sign measurement device, 216...Imaging device, 220...Life log collection device, 230...Personal data setting device, 30...Server device, 32...Patient information storage device, 322...Life log storage device, 324...Personal data storage device, NW...Network

Claims

1. A medical information processing device that receives questions from medical professionals to patients and answers them on behalf of the patient using a digital twin that reproduces the patient's condition, The digital twin is an AI model trained by adding the patient's information to at least a large-scale language model. The reception desk that accepts the aforementioned questions, At a minimum, an information acquisition unit that acquires medical condition information representing the patient's medical condition, An update unit updates the digital twin based on the aforementioned medical condition information, A response unit that uses the digital twin to output response information representing the answer to the question received, A display control unit that generates display data for presenting the aforementioned response information and displays it on a display device, A medical information processing device equipped with [a specific feature].

2. A missing information extraction unit determines whether the answer to the received question can be provided using the current digital twin, and if it is determined that the answer to the question cannot be provided, it extracts the missing information necessary to answer the question. Furthermore, When the missing information is extracted, the information acquisition unit acquires new disease condition information corresponding to the patient. The update unit updates the digital twin, including the new disease condition information. The medical information processing device according to claim 1.

3. A difference extraction unit that extracts the difference between the patient's previous medical condition information and the patient's new medical condition information. Furthermore, The update unit updates the digital twin to include the new medical condition information when it detects a difference between the previous medical condition information and the new medical condition information. The medical information processing device according to claim 2.

4. The aforementioned medical condition information is, The patient information recorded when the aforementioned patient was examined, Test information representing the results of tests performed on the aforementioned patient, including, A medical information processing device according to any one of claims 1 to 3.

5. The aforementioned patient information is The first patient information, which is the patient information recorded when the patient was examined, The second patient information is patient information based on lifestyle records output by a collection device that collects the patient's lifestyle, including, The medical information processing device according to claim 4.

6. The second patient information mentioned above is: Third patient information concerning the aforementioned patient that was not recorded when the patient was examined, including, The medical information processing device according to claim 5.

7. The AI ​​model, which receives questions from medical professionals to patients and learns by adding the patient's information to at least a large-scale language model, and the computer of the medical information processing device that answers on behalf of the patient using a digital twin that reproduces the patient's condition, We have received the above question, At a minimum, obtain disease information representing the patient's medical condition, Based on the aforementioned medical condition information, the digital twin is updated. Using the aforementioned digital twin, answer information representing the answer to the received question is output. To generate display data for presenting the aforementioned response information and display it on a display device, Medical information processing method.

8. An AI model that receives questions from medical professionals to patients and learns by adding the patient's information to at least a large-scale language model, and a computer in a medical information processing device that answers on behalf of the patient using a digital twin that reproduces the patient's condition, We will accept the aforementioned questions, At a minimum, obtain medical condition information representing the patient's medical condition, Based on the aforementioned medical condition information, update the digital twin. Using the aforementioned digital twin, the system outputs answer information representing the answer to the received question. To generate display data for presenting the aforementioned response information and to display it on a display device, program.

Images