Medical chart creation assistance device and medical chart creation assistance method

Abstract

A server 1 according to one embodiment of the present invention comprises: an initial plan creation unit 11 that creates an initial plan relating to treatment of a patient on the basis of basic data, which is basic information relating to the patient obtained from the patient, and stores the initial plan in an initial plan database 53; a course record creation unit 12 that stores, in a course record database 54, information on the course of the treatment of the patient performed on the basis of the initial plan; and an abnormality inference unit 13 that infers, on the basis of the initial plan of the patient and the information on the course of the treatment, an abnormality assumed to be a cause of a deviation between the initial plan and the course of the treatment, and outputs an inference result of the abnormality.

Description

Dental Chart Creation Support Device and Dental Chart Creation Support Method

[0001] The present invention relates to a dental chart creation support device and a dental chart creation support method.

[0002] Many dentists study medicine such as diagnosis and treatment methods at dental universities and the like, but often do not have the opportunity to deeply study how to write a dental chart. Therefore, many dentists graduate from dental universities and the like without learning how to write a dental chart and immediately start clinical practice.

[0003] However, in order to provide high-quality dental care, it is not possible to achieve this only by improving treatment techniques. It is also an important factor to record patient information accurately and without delay in the dental chart. In a dental clinic, since dentists may change, it is ideal to record a dental chart that allows any dentist to understand the patient's condition as well as the attending dentist.

[0004] As a method for maintaining the quality of the created dental chart at a certain level, the "SOAP format", which is the entry format of nursing records in a medical institution, is known. The SOAP format is a method of entering information obtained from a patient through examination and the like into four items: S (subjective), O (objective), A (assessment), and P (plan).

[0005] "S" indicates subjective information, and in S, the content of the chief complaint by the patient and the like are classified. "O" indicates objective information, and in O, objective information obtained from examination and tests and the like are classified. "A" indicates an evaluation, and in A, the doctor's diagnosis, comprehensive evaluation as a result of analysis or interpretation based on the content of S and O, and the like are classified. "P" indicates a plan, and in P, the treatment policy and plan determined based on "A" and the like are classified.

[0006] For example, Patent Document 1 describes a medical record creation support device comprising: a text division unit that divides free text containing information obtained from a patient into phrases; and a phrase classification unit that classifies each of the phrases divided by the text division unit into one of several categories based on the nursing record entry format according to the content indicated by the phrase, and outputs the classification result. According to the technology described in Patent Document 1, it is possible to appropriately create medical records based on medical record entry formats such as the "SOAP format".

[0007] Patent No. 7501883

[0008] Incidentally, there is a known medical record method called POMR (Problem-Oriented Medical Record) that records patient information gathered from interviews, physical examinations, tests, and imaging data, categorized according to the SOAP classification for each problem. POMR is a medical record adopted in POS (Problem-Oriented System). POS is a system that represents a series of operations and mechanisms based on the idea of ​​providing medical care that focuses on the medical problems a patient has, while also valuing the patient's QOL (Quality of Life), and aiming for the most effective solution to those problems.

[0009] Based on the POM (Problem-Oriented Medical Report), the data collected for each problem, categorized as "S" and "O," is analyzed ("A"), evaluated, and strategies are formulated ("P"), completing one PDCA cycle (plan-do-check-act cycle). In this way, by running the PDCA cycle based on the POM, the content of examinations and advice given to patients can be improved, thereby enhancing the quality of medical care.

[0010] However, if there are oversights or misdiagnoses during the examination, even if a medical record is created based on POMR, the patient's symptoms may not improve, or in some cases, may even worsen. The cause of a patient's oral disease may lie in internal conditions such as diabetes or kidney disease, or in lifestyle habits, and an experienced dentist can examine these medical issues in conjunction with the symptoms. On the other hand, an inexperienced dentist may find it difficult to discern such a connection. If inappropriate treatment is given due to oversights or misdiagnoses during the examination, the course of treatment is likely to deviate from the original treatment plan.

[0011] This invention was made in consideration of the above circumstances, and its purpose is to enable the output of information on abnormalities that are assumed to be the cause of discrepancies between the treatment plan and the course of treatment.

[0012] A medical record creation support device according to one aspect of the present invention comprises: an initial plan creation unit that creates an initial treatment plan for a patient based on basic data, which is basic information about the patient obtained from the patient, and stores it in an initial plan database; a progress record creation unit that stores information on the patient's treatment progress carried out based on the initial plan in a progress record database; and an abnormality inference unit that infers abnormalities that are assumed to be the cause of discrepancies between the initial plan and the treatment progress, based on the patient's basic data, initial plan, and treatment progress information, and outputs the result of the abnormality inference.

[0013] According to at least one aspect of the present invention, it is possible to output information on abnormalities that are assumed to be the cause of discrepancies between the treatment plan and the treatment progress. Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments.

[0014] This figure shows an example of the schematic configuration of a medical record creation support system according to one embodiment of the present invention. This is a block diagram showing an example of the configuration of a computer, which is the hardware of the server, medical record management server, and terminal device according to one embodiment of the present invention. This is a block diagram showing an example of the functional configuration of a server according to one embodiment of the present invention. This figure shows an example of the configuration of basic data according to one embodiment of the present invention. This figure shows an example of the configuration of a problem list DB according to one embodiment of the present invention. This figure shows an example of the configuration of an initial plan DB according to one embodiment of the present invention. This figure shows an example of the configuration of a progress record DB according to one embodiment of the present invention. This figure shows an example of the configuration of an abnormality DB according to one embodiment of the present invention. This figure shows an example of the structure of a problem list generation prompt input to a problem list AI according to one embodiment of the present invention. This figure shows an example of the configuration of training data to be trained on a problem list AI according to one embodiment of the present invention. This figure shows an example of the execution result of inference by a problem list AI according to one embodiment of the present invention. This figure shows an example of the structure of an initial plan generation prompt input to an initial plan AI according to one embodiment of the present invention. This figure shows an example of the configuration of training data to be trained on an initial plan AI according to one embodiment of the present invention. This figure shows an example of the execution result of inference by an initial plan AI according to one embodiment of the present invention. This figure shows an example of the structure of an abnormality inference prompt input to an abnormality inference AI according to one embodiment of the present invention. This figure shows an example of the configuration of training data to be trained on an abnormality inference AI according to one embodiment of the present invention. This figure shows an example of the execution result of inference by an abnormality inference AI according to one embodiment of the present invention. This figure shows an example of the structure of a summary creation prompt input to the summary AI according to one embodiment of the present invention. This figure shows an example of the execution result of summary creation by the summary AI according to one embodiment of the present invention. This figure shows an example of the configuration of a medical record creation support screen according to one embodiment of the present invention. This figure shows an example of the configuration of a problem list creation screen according to one embodiment of the present invention. This figure shows an example of the configuration of a medical record creation support screen displaying a problem list according to one embodiment of the present invention. This figure shows an example of the configuration of an initial plan creation screen according to one embodiment of the present invention. This figure shows an example of the configuration of a medical record creation support screen displaying an initial plan according to one embodiment of the present invention. This figure shows an example of the configuration of a progress record creation screen according to one embodiment of the present invention.This figure shows an example configuration of a medical record creation support screen displaying progress records related to one embodiment of the present invention. This figure shows an example configuration of an anomaly confirmation screen related to one embodiment of the present invention. This figure shows an example configuration of a medical record creation support screen displaying anomaly inference results related to one embodiment of the present invention. This figure shows an example configuration of a summary creation screen related to one embodiment of the present invention. This flowchart shows an example of the procedure for creating a problem list related to one embodiment of the present invention. This flowchart shows an example of the procedure for creating an initial plan related to one embodiment of the present invention. This flowchart shows an example of the procedure for creating progress records related to one embodiment of the present invention. This flowchart shows an example of the procedure for anomaly inference processing related to one embodiment of the present invention. This flowchart shows an example of the procedure for creating a summary related to one embodiment of the present invention.

[0015] Hereinafter, examples of embodiments for carrying out the present invention (hereinafter referred to as "embodiments") will be described with reference to the accompanying drawings. Various numerical values ​​and other figures in the embodiments of the present invention are merely illustrative, and the present invention is not limited to the embodiments described below. Furthermore, in this specification and drawings, the same reference numerals will be used for identical components or components having substantially the same function, and redundant explanations will be omitted.

[0016] <Outline Configuration of the Medical Record Creation Support System> First, with reference to Figure 1, the configuration of the medical record creation support system 100 according to one embodiment of the present invention will be described. Figure 1 is a diagram showing an example of the outline configuration of the medical record creation support system 100.

[0017] As shown in Figure 1, the medical record creation support system 100 includes a server 1 (an example of a medical record creation support device), a medical record management server 2, and terminal devices 3-1 to 3-n (where n is a natural number greater than or equal to 2). Server 1, the medical record management server 2, and each of the terminal devices 3-1 to 3-n are connected to each other via a network N. In the following description, when it is not necessary to distinguish between terminal devices 3-1 to 3-n, they will be collectively referred to as terminal device 3.

[0018] Server 1, for example, is located in a cloud environment and creates the information necessary for creating an electronic medical record (hereinafter also simply referred to as "medical record") in accordance with the POMR (Proof of Medical Record). Server 1 then transmits the created information to the medical record management server 2. The medical record management server 2 uses its medical record management function to digitize the information transmitted from Server 1 and create an electronic medical record. This electronic medical record becomes an insurance medical record that records the treatments performed on the patient and the insurance points.

[0019] More specifically, Server 1 creates a problem list from the pre-created basic data based on examinations, etc., and creates an initial plan for each problem defined in the problem list. The basic data is fundamental information about the patient, and this basic data includes the patient's medical and dental data obtained based on examinations, tests, etc., and psychosocial data describing the patient's psychological and / or social life. An example of the configuration of the basic data DB 51 that stores the basic data is explained in Figure 4.

[0020] The problem list is a compilation of patient problems extracted from the basic data and is stored in the problem list DB52. An example of the structure of the problem list DB52 is explained in Figure 5.

[0021] The initial plan consists of a diagnostic plan, a treatment plan, and a patient education plan, and is stored in the initial plan DB53. An example of the configuration of the initial plan DB53 is illustrated in Figure 6.

[0022] Furthermore, Server 1 records the details of examinations and other procedures performed by the dentist based on the initial plan created, corresponding to each category of SOAP, in the progress record DB 54 for each problem in the problem list. An example of the configuration of the progress record DB 54 is explained in Figure 7.

[0023] Furthermore, based on the information regarding the difference between the initial plan and the progress record, Server 1 infers abnormalities that are likely to cause discrepancies between the treatment plan and the treatment progress recorded in the progress record. The results of the abnormality inference are stored in the abnormality DB 55. An example of the configuration of the abnormality DB 55 is explained in Figure 8.

[0024] Furthermore, Server 1 creates a summary when the patient's treatment is completed, etc. This summary is a document used to provide information to external staff, such as care managers and doctors at other hospitals.

[0025] The medical record management server 2 is a server with medical record management functionality implemented, operated, for example, by a service company. Like server 1, medical record management server 2 is deployed in a cloud environment. For example, staff working at a dental clinic can access medical record management server 2 and view electronic medical records by entering their user ID, password, etc., on the authentication screen of medical record management server 2 using their respective terminal devices 3. Medical record management server 2 creates medical records based on various information transmitted from server 1 and stores them in a medical record database (not shown). Medical record management server 2 also transmits information about the created medical records to terminal devices 3.

[0026] Terminal device 3 is a device consisting of a PC (Personal Computer) installed in the dental clinic and a mobile terminal such as a smartphone. Based on instructions entered by a user such as a dentist via the operation input unit 23 (see Figure 2), terminal device 3 requests the server 1 to acquire various data, displays the acquired data on the screen of the output unit 24, and requests the server 1 to record the information entered by the user via the operation input unit 23.

[0027] <Configuration of the control system for the medical record creation support system> Next, with reference to Figure 2, the configuration of the hardware constituting each of the server 1, medical record management server 2, and terminal device 3 that make up the medical record creation support system 100 will be described. Figure 2 is a block diagram showing an example configuration of the computer 200, which is the hardware of server 1, medical record management server 2, and terminal device 3.

[0028] As shown in Figure 2, the computer 200 comprises a control unit 21, a storage unit 22, an operation input unit 23, an output unit 24, and a communication interface unit 25, all connected to bus B. The control unit 21 is an arithmetic unit including a CPU (Central Processing Unit) 211, a ROM (Read Only Memory) 212, and a RAM (Random Access Memory) 213.

[0029] The CPU 211 reads the program code of the software that implements each function of the server 1 according to this embodiment from the ROM 212, loads it into the RAM 213, and executes it. Note that the server 1 may also include a processing unit such as an MPU (Micro-Processing Unit) instead of the CPU 211. Variables and parameters generated during the calculation process by the CPU 211 are temporarily written to the RAM 213.

[0030] For the storage unit 22, for example, an HDD (Hard Disk Drive), SSD (Solid State Drive), flexible disk, optical disk, magneto-optical disk, CD-ROM, CD-R, non-volatile memory card, etc. can be used. In addition to the OS (Operating System) and various parameters, the storage unit 22 stores a program for making the server 1 function.

[0031] The program for operating server 1 may be stored in ROM 212. The program is stored in the form of computer-readable program code. The CPU 211 sequentially executes operations according to the program code. In other words, ROM 212 or storage unit 22 is an example of a computer-readable, non-transient recording medium that stores a program executed by a computer.

[0032] The operation input unit 23 is composed of, for example, a mouse or keyboard, and generates operation signals in response to user operations and supplies them to the CPU 211. The output unit 24 is a monitor composed of, for example, an LCD (Liquid Crystal Display). The operation input unit 23 and the output unit 24 may be integrated as a touch panel.

[0033] Communication I / F25 consists of communication devices and the like that control communication with external devices.

[0034] <Server Functionality Block> Next, with reference to Figure 3, the functional configuration of the server 1 of the medical record creation support system 100 according to this embodiment will be described. Figure 3 is a block diagram showing an example of the functional configuration of the server 1.

[0035] As shown in Figure 3, Server 1 is connected to the medical record management server 2 and includes a basic data DB (Database) 51, a problem list creation unit 10, a problem list DB 52, an initial plan creation unit 11, an initial plan DB 53, a progress record creation unit 12, and a progress record DB 54. Server 1 also includes an anomaly inference unit 13, an anomaly DB 55, and a summary creation unit 14.

[0036] The problem list creation unit 10 includes a problem list creation support unit 101 and a problem list AI 102. When the user instructs the creation of a patient's problem list via the operation input unit 23 (see Figure 2), the problem list creation support unit 101 reads all the basic data associated with the patient's information (patient ID) from the basic data DB 51. The problem list creation support unit 101 then outputs the read information to the problem list AI 102 along with a prompt that is an instruction to the problem list AI 102.

[0037] Furthermore, the problem list creation support unit 101 displays the problem list data transmitted from the problem list AI 102 on the screen of the output unit 24 (see Figure 2), and modifies the contents of the problem list according to the user's instructions entered from the operation input unit 23. Then, the problem list creation support unit 101 stores the created (modified) problem list in the problem list DB 52.

[0038] The problem list AI 102 is composed of a large language model (LLM (Large Language Model)) or the like. Based on the basic data and prompts associated with the patient ID input from the problem list creation support unit 101, the problem list AI 102 extracts patient problems from the information input from the problem list creation support unit 101 and outputs the extracted problems to the problem list creation support unit 101. The problem list AI 102 may also be composed of a deep learning model (hereinafter simply referred to as "learning model") that has learned from past cases (clinical data).

[0039] Learning models that have learned from past cases can be constructed using, for example, a neural network for multi-class classification, a learning model employing the Random Forest algorithm, or a finely tuned LLM.

[0040] As described above, the problem list is a list of each patient problem extracted from the basic data. Therefore, when creating a problem list without using the technology according to the present invention, dentists need to determine which items from the items described in the basic data could be problems for the patient and extract them appropriately. However, for dentists with little experience or knowledge, such determination and problem extraction can be difficult. Even dentists with extensive experience and knowledge may overlook or select the wrong items to extract as problems. In contrast, the problem list AI 102 according to this embodiment can create a problem list by appropriately extracting and listing items that could be problems for the patient based on the patient's basic data.

[0041] An example of the configuration of prompt Pp1 input to problem list AI102 is shown in Figure 9, and an example of training data Dt1 for problem list AI102 is shown in Figure 10. An example of inference result Ri1 by problem list AI102 is shown in Figure 11.

[0042] Problem list DB52 is a database that manages problems extracted by problem list AI102, associating them with patient IDs that identify patients.

[0043] The initial plan creation unit 11 includes an initial plan creation support unit 111 and an initial plan AI 112. When the creation of the patient's initial plan is instructed by the user via the operation input unit 23, the initial plan creation support unit 111 reads out the problems that match the patient ID of that patient from the problem list DB 52. Then, the initial plan creation support unit 111 outputs the problem list of the read problems to the initial plan AI 112 together with a prompt which is an instruction for the initial plan AI 112.

[0044] Also, the initial plan creation support unit 111 causes the data of the initial plan transmitted from the initial plan AI 112 to be displayed on the screen of the output unit 24 (see FIG. 2), and modifies the content of the initial plan according to the instruction of the user input from the operation input unit 23. Then, the initial plan creation support unit 111 stores the modified initial plan in the initial plan DB 53.

[0045] The initial plan AI 112 is composed of a large language model or the like. The initial plan AI 112 creates an initial plan based on the prompt input from the initial plan creation support unit 111, and outputs the created initial plan to the initial plan creation support unit 111. The initial plan AI 112 formulates an "initial plan" from each of the viewpoints of "diagnosis", "treatment", and "education" for each problem in the problem list. For this reason, the initial plan AI 112 is formulated by three learning models, a diagnosis plan model, a treatment plan model, and an education plan model (all not shown).

[0046] Note that the initial plan AI 112 may be composed of a learning model that has learned past cases. The learning model that has learned past cases can be composed of, for example, a neural network of a multi-class classification model, a learning model that employs the algorithm of a random forest, a fine-tuned LLM, etc.

[0047] As described above, the initial plan is created based on the problem list. Each element (problem) constituting the problem list has various combinations for each patient. And to that combination, the patient's allergies, medication history, patient requests, etc. are added as parameters. Therefore, the dentist needs to formulate a plan for each problem after considering all the elements in those problem lists. However, it is often difficult for dentists with little experience and knowledge to extract problems without excess or deficiency and to formulate a plan for each extracted problem. Also, even experienced and knowledgeable dentists may overlook problems or make incorrect selections. In contrast, the initial plan AI 112 according to this embodiment can create an appropriate diagnosis plan, treatment plan, and education plan for the patient based on the problem list.

[0048] A configuration example of the prompt Pp2 input to the initial plan AI 112 is described in FIG. 12, and an example of the learning data Dt2 of the initial plan AI 112 is described in FIG. 13. Also, an example of the inference result Ri2 by the initial plan AI 112 is described in FIG. 14.

[0049] The initial plan DB 53 is a database that manages the initial plan created by the initial plan AI 112 in association with the patient ID.

[0050] The progress record creation unit 12 stores the content of the examination (treatment progress) performed by the dentist or the like based on the initial plan stored in the initial plan DB 53 in the progress record DB 54 in association with each section of SOAP. The association between the content of the examination and each section of SOAP is performed by the user such as the dentist through an operation on the progress record creation screen Pu3. A configuration example of the progress record creation screen Pu3 is described in FIG. 25.

[0051] The functions of the progress record creation unit 12 may also be realized by a learning model using the technology described in Patent Document 1. In the medical record creation support device described in Patent Document 1, the text division unit divides the free text containing information obtained from the patient into phrases. The phrase classification unit then classifies each of the divided phrases into one of several categories based on the medical record entry format, according to the content indicated by the phrase, and outputs the classification result. The phrase classification unit is composed of a learning model that has learned the correspondence information between phrases contained in the free text and the categories to which the phrases should be classified.

[0052] The abnormality inference unit 13 includes an abnormality confirmation unit 131 and an abnormality inference AI 132. When the user instructs the abnormality confirmation unit 131 to perform abnormality inference for a patient via the operation input unit 23, the abnormality confirmation unit 131 reads an initial plan matching the patient's patient ID from the initial plan DB 53. The abnormality confirmation unit 131 also reads a progress record matching the patient ID from the progress record DB 54. In this embodiment, "abnormality" refers to the discrepancy (inconsistency) that occurred between the initial plan and the treatment progress, which is the result of the treatment actually performed. The "abnormality inference" then infers the cause of such a discrepancy between the initial plan and the treatment progress.

[0053] The anomaly detection unit 131 outputs the information from the read initial plan and progress records to the anomaly inference AI 132 along with a prompt that is an instruction to the anomaly inference AI 132. The anomaly detection unit 131 also displays the anomaly inference result data transmitted from the anomaly inference AI 132 on the screen of the output unit 24 (see Figure 2) and corrects the content of the anomaly inference result according to the user's instructions entered from the operation input unit 23. The anomaly detection unit 131 then stores the corrected anomaly inference result in the anomaly DB 55.

[0054] The anomaly inference AI 132 is composed of a large-scale language model, etc. Based on prompts input from the anomaly confirmation unit 131, the anomaly inference AI 132 infers an anomaly based on the initial plan and progress records input from the anomaly confirmation unit 131, and outputs the inferred anomaly information to the anomaly confirmation unit 131. The anomaly inference AI 132 may also be composed of a learning model that has learned from past cases.

[0055] A learning model that has learned from past cases can be constructed using, for example, a neural network for multi-class classification, a learning model employing the Random Forest algorithm, or a finely tuned LLM. An example of the configuration of the prompt Pp3 input to the anomalous inference AI 132 is explained in Figure 15, and an example of the training data Dt3 for the anomalous inference AI 132 is explained in Figure 16. Furthermore, an example of the inference result Ri3 by the anomalous inference AI 132 is explained in Figure 17.

[0056] Alternatively, the anomaly inference AI 132 may be composed of a learning model that detects differences in the strings between the initial plan and the treatment progress in the progress record as anomalies. In conventional technology, it is assumed that the strings written in the initial plan and treatment progress include various expressions and that the expressions are not standardized. For example, even terms that mean the same thing may be expressed differently depending on the medical record maker, etc. In such cases, when the initial plan and treatment progress include synonyms with different pronunciations, the anomaly inference AI 132 can appropriately infer anomalies by training its learning model with learning data that associates synonyms with appropriate terms.

[0057] The Anomaly DB55 is a database that manages anomaly information inferred by the Anomaly Inference AI132, associating it with patient IDs.

[0058] The summary creation unit 14 includes a summary creation support unit 141 and a summary AI 142. When the user instructs the creation of a summary via the operation input unit 23, the summary creation support unit 141 reads basic data from the basic data DB 51 that matches the patient's patient ID. The summary creation support unit 141 also reads the patient's progress records from the progress record DB 54 and reads abnormalities that match the patient ID from the abnormality DB 55.

[0059] The summary creation support unit 141 then outputs each piece of information it has read to the summary AI 142 along with a prompt that is an instruction to the summary AI 142. The summary creation support unit 141 also displays the summary data sent from the summary AI 142 on the screen of the output unit 24 (see Figure 2) and modifies the contents of the summary according to the user's instructions entered from the operation input unit 23. Finally, the summary creation support unit 141 sends the modified summary to the medical record management server 2.

[0060] The summary AI 142 creates a summary using the patient's basic data, progress records, and abnormal information input from the summary creation support unit 141, and outputs the created summary to the summary creation support unit 141. The summary AI 142 may also output a marked-up summary. By configuring the summary AI 142 in this way, the summary creation support unit 141 can also present the summary in a user-friendly format.

[0061] Furthermore, the summary AI 142 may be composed of a learning model that has learned from past cases. A learning model that has learned from past cases can be composed of, for example, a neural network of a multi-class classification model, a learning model employing the Random Forest algorithm, or a fine-tuned LLM. An example of the configuration of the prompt Rp4 input to the summary AI 142 is explained in Figure 18, and an example of the inference result Ri4 by the summary AI 142 is explained in Figure 19.

[0062] When attempting to create a summary manually, it is anticipated that extracting and listing all the necessary information from the patient's basic data, treatment progress, etc., would take an enormous amount of time. In contrast, the summary AI 142 according to this embodiment can extract all the items necessary for creating a summary and create a summary appropriately.

[0063] <Configuration of the Basic Data Database> Next, the configuration of the basic data database 51 will be explained with reference to Figure 4. Figure 4 is a diagram showing an example of the configuration of the basic data database 51. As shown in Figure 4, the basic data database 51 has the following items: "Patient ID", "Name", "Date of Birth", "Gender", "Occupation", "Economic Environment", "Chief Complaint", "Current Illness", "Past Medical History and Family History", "Allergies", "Comorbidities", "Current Condition", and "Images".

[0064] The "Patient ID" field stores the patient's identification ID. The "Name" field stores the patient's name. The "Date of Birth" field stores the patient's date of birth. The "Gender" field stores the patient's gender. The "Occupation" field stores information about the patient's occupation. The "Economic Background" field stores information about the patient's economic background. The "Chief Complaint" field stores information about the patient's chief complaint obtained through interviews, examinations, etc.

[0065] The "Current Illness" section stores information about the patient's current illness. This section describes the onset and progression of symptoms currently occurring. The "Past Medical History and Family History" section stores information about the patient's past illnesses and family history. Past medical history shows information about illnesses the patient has had in the past. Family history shows information about the health (illnesses) of the patient and their close relatives. The "Allergies" section stores information about whether the patient has any allergic diseases and the nature of those allergic diseases.

[0066] The "Addiction" field stores information about the addiction the patient is suffering from. The "Current Symptoms" field stores information about the symptoms the patient is currently being treated for. The "Images" field stores images such as X-rays and intraoral images, for example, in a BLOB (Binary Large Object) data type. Note that the example shown in Figure 4 is just one example, and the basic data DB 51 does not need to have all the items shown in Figure 4, and may also have items not included in Figure 4.

[0067] <Structure of the Problem List DB> Next, the structure of the Problem List DB 52 will be explained with reference to Figure 5. Figure 5 is a diagram showing an example of the structure of the Problem List DB 52. As shown in Figure 5, the Problem List DB 52 has the following items: "Patient ID", "Problem Number", "Problem Category", "Onset Date", "Problem", and "Status (Active or Inactive)".

[0068] The "Patient ID" field stores the patient's ID. The "Problem Number" field stores the problem number assigned to each patient's problem extracted by the problem list AI102 from the basic data DB51. The "Problem Category" field stores information about the classification of the problem. Problem categories include, for example, dental problems, medical problems, social / family problems, habits, or risk factors.

[0069] The "Onset Date" field stores information about the onset date of the symptoms of patients categorized as a problem. If onset date information is stored in the basic data DB51, the information in the onset date field may be copied from there, or it may be manually entered by the user. The "Problem" field stores information about the patient's problem extracted from the basic data DB51 by the problem list AI102.

[0070] The "Status (Active or Inactive)" field stores information indicating whether the status of the problem stored in the "Problem" field is active or inactive. Active problems are those that must always be considered when treating an ongoing illness. Inactive problems are problems that have already been resolved, such as past medical history or illnesses or conditions that may be related to the disease being treated. The "Status (Active or Inactive)" field is accessed by users such as dentists through operations on the problem list creation screen Pu1. An example of the configuration of the problem list creation screen Pu1 is explained in Figure 21.

[0071] Furthermore, the initial planning AI 112 may create an initial plan by referring to the information stored in the "Status (Activity or Inactivity)" item. For example, the initial planning AI 112 can create an initial plan using only problems classified under "Activity". Also, the example shown in Figure 5 is just one example, and the problem list DB 52 does not need to have all the items shown in Figure 5, and may also have items not included in Figure 5.

[0072] <Configuration of Initial Planning DB> Next, the configuration of the initial planning DB 53 will be explained with reference to Figure 6. Figure 6 is a diagram showing an example of the configuration of the initial planning DB 53. As shown in Figure 6, the initial planning DB 53 has the following items: "Patient ID", "Problem Number", "Diagnostic Plan", "Treatment Plan", and "Education Plan".

[0073] The "Patient ID" field stores the patient's ID. The "Problem Number" field stores the problem number. The "Diagnostic Plan" field stores information about the diagnostic plan to be performed on the patient. The "Treatment Plan" field stores information about the treatment plan to be performed on the patient. The "Education Plan" field stores information about the education plan to be performed on the patient.

[0074] <Configuration of the Progress Record DB> Next, the configuration of the progress record DB 54 will be explained with reference to Figure 7. Figure 7 is a diagram showing an example of the configuration of the progress record DB 54. As shown in Figure 7, the progress record DB 54 has the following items: "Patient ID", "Date", "Problem Number", and "Therapy and Treatment".

[0075] The "Patient ID" field stores the patient ID. The "Date" field stores the date the treatment details were recorded in the medical record. The "Problem Number" field stores the problem number. The "Location" field stores information about the location in the patient's oral cavity where the problem occurred. The "Therapy and Treatment" field stores the details of the examination and treatment performed on the patient, corresponding to each category of SOAP. In addition, if there are comments from the dentist, etc., comments such as "I explained that oral problems can develop systemically and cautioned that treatment should be prioritized above all else" may also be stored. The "Therapy and Treatment" field can be entered by the dentist, etc. via the operation input unit 23 (see Figure 2). Alternatively, if the technology described in Patent Document 1 is used, the information classified by the document classification unit is stored in the "Therapy and Treatment" field.

[0076] <Configuration of the Abnormal Database> Next, the configuration of the Abnormal Database 55 will be explained with reference to Figure 8. Figure 8 is a diagram showing an example of the configuration of the Abnormal Database 55. As shown in Figure 8, the Abnormal Database 55 has the following items: "Patient ID", "Problem Number", and "Abnormal Inference Result".

[0077] The "Patient ID" field stores the patient ID. The "Problem Number" field stores the problem number. The "Abnormality Inference Result" field stores information about abnormalities inferred by the abnormality inference AI 132. The "Abnormality Inference Result" field also stores comments regarding the discrepancies between the initial plan and the progress record, and the content of the abnormality (problem) that is assumed to be the cause of the discrepancy between the treatment plan and the treatment progress.

[0078] The comments include, in the first line of the "Abnormal Inference Result" section, "The number of times medication has been applied has been greater than initially planned." As for the nature of the abnormality, for example, in the second line of the "Abnormal Inference Result" section, it says, "Dental problem: Hidden root canal (Some teeth may have unexpected additional root canals, and complete treatment may not be possible if they are not found)."

[0079] <Structure of Input / Output Data of Problem List AI> Next, referring to Figures 9 to 11, the structure of the data input to Problem List AI 102 and the data output from Problem List AI 102 will be explained. The data input to Problem List AI 102 includes the problem list generation prompt Pp1 and the training data Dt1, while the data output from Problem List AI 102 includes the inference execution result Ri1.

[0080] [Structure of the Problem List Generation Prompt] Figure 9 shows an example of the structure of the problem list generation prompt Pp1 that is input to the problem list AI 102. As shown in Figure 9, the first line of the problem list generation prompt Pp1 contains the sentence "You are a top-class dentist who is well-versed in the concept of POS" as a conditional statement for the answer from the problem list AI 102.

[0081] Furthermore, the second line of the problem list generation prompt Pp1 contains the following instruction for the problem list AI102: "From the 'basic data' entered by the user, extract the 'dental problems,' 'medical problems,' 'psychological problems,' 'social and family problems,' and 'habits or risk factors' that the patient is said to have." The lines from the third line onward define terms such as "dental problems" and "medical problems," and specify how the data should be output. The data output method is specified by the following statement: "Each extracted problem should be output as a single line of data in the format 'line number, type of problem, extracted problem, line number where a relationship was found.'"

[0082] [Structure of Training Data] Figure 10 shows an example of the structure of training data Dt1 to be trained on the problem list AI 102. The training data Dt1 shown in Figure 10 is a sequence of problems and problem categories. Each "problem" item consists of a single phrase of text that makes up the basic data stored in the basic data DB 51. A single phrase of text can be extracted from the basic data by dividing it into sections using punctuation marks or by performing morphological analysis.

[0083] For example, in the first row of the training data Dt1 shown in Figure 10, the problem category "dental problem" is associated with the problem "acute suppurative apical periodontitis." Similarly, in the second row, the problem category "dental problem" is associated with the problem "acute subperiosteal abscess." The problem list AI 102 then infers the problem category for each problem based on the training data. The problem list AI 102 can improve the accuracy of its inference by training on a large amount of such training data generated based on examples.

[0084] [Structure of the inference execution result] Figure 11 shows an example of the inference execution result Ri1 by the problem list AI 102. As shown in Figure 11, the inference execution result Ri1 by the problem list AI 102 is output in the format of "line number, type of problem, extracted problem, line number where a relationship was found" based on the provisions written in the last line of the problem list generation prompt Pp1 shown in Figure 9. For example, the first line of the inference execution result Ri1 shows the execution result "#1, upper right 6, acute suppurative apical periodontitis, none".

[0085] <Structure of Input / Output Data of Initial Planning AI> Next, with reference to Figures 12 to 14, the structure of the data input to and output from the initial planning AI 112 will be explained. The data input to the initial planning AI 112 includes the initial plan generation prompt Pp2 and the training data Dt2, while the data output from the initial planning AI 112 includes the execution result Ri2 of the initial plan creation.

[0086] [Structure of the Initial Plan Generation Prompt] Figure 12 shows an example of the structure of the initial plan generation prompt Pp2 that is input to the initial plan AI 112. As shown in Figure 12, the first line of the initial plan generation prompt Pp2 contains a sentence specifying conditions for the response from the initial plan AI 112.

[0087] Furthermore, the second line of the initial plan generation prompt Pp2 contains the instruction to the initial plan AI 112: "Devise a 'diagnostic plan,' 'treatment plan,' and 'educational plan' for each problem from the 'problem list' entered by the user." The lines from the third line onward define terms such as "diagnostic plan," "treatment plan," and "educational plan," and specify how the data should be output. The data output method is specified by the statement: "The output format should be 'problem number, diagnostic plan, treatment plan, educational plan.'"

[0088] [Structure of Training Data] Figure 13 shows an example of the structure of training data Dt2 to be trained on the initial planning AI 112. Figure 13A shows an example of the structure of training data Dt21 for the initial planning AI 112's medical examination plan model. Figure 13B shows an example of the structure of training data Dt22 for the initial planning AI 112's treatment plan model. Figure 13C shows an example of the structure of training data Dt23 for the initial planning AI 112's education plan model.

[0089] The learning data Dt21 shown in Figure 13A corresponds to the problem and the diagnostic plan. The learning data Dt22 shown in Figure 13B corresponds to the problem and the treatment plan. The learning data Dt23 shown in Figure 13C corresponds to the problem and the educational plan.

[0090] For example, in the first row of the training data Dt21 shown in Figure 13A, the diagnostic plan of "dental X-ray examination" is associated with the problem of "acute suppurative apical periodontitis." In the second row, although some of the diagrams are omitted, the diagnostic plan of "bacterial examination of tumor, blood test" is associated with the problem of "acute subperiosteal abscess." The initial planning AI 112 then infers a diagnostic plan for the problem based on the training data. Therefore, the initial planning AI 112 can improve the accuracy of its inference by training the diagnostic planning model with a large amount of such training data generated based on cases.

[0091] Furthermore, for example, in the first row of the training data Dt22 shown in Figure 13B, for the problem of "acute suppurative apical periodontitis," although some of the illustrations are omitted, a treatment plan for infected root canal treatment, "X-ray examination → infected root canal treatment → medication → root filling → PZ imp BT → Set crown," is associated. Similarly, in the second row, a treatment plan of "anti-inflammatory therapy (anti-inflammatory surgery / antibiotics), analgesic treatment" is associated with the problem of "acute subperiosteal abscess." The treatment planning model of the initial planning AI 112 then infers a treatment plan for each problem based on the training data. By training the treatment planning model with a large amount of such training data generated based on case examples, the initial planning AI 112 can improve the accuracy of its inference.

[0092] Furthermore, for example, in the first row of the learning data Dt23 shown in Figure 13C, the problem "acute suppurative apical periodontitis" is associated with the educational plan "avoid brushing this area." In the second row, although some of the illustrations are omitted, the problem "acute subperiosteal abscess" is associated with the educational plan "rest, oral hygiene, and explanation of prognosis." The educational plan model of the initial planning AI 112 then infers an educational plan for each problem based on the learning data. The initial planning AI 112 can improve the accuracy of its inference by training the educational plan model with a large amount of such learning data generated based on cases.

[0093] [Structure of the inference execution result] Figure 14 shows an example of the inference execution result Ri2 by the initial planning AI 112. As shown in Figure 14, the inference execution result Ri2 by the initial planning AI 112 is output in the format of "Problem number, Diagnostic plan, Treatment plan, Education plan" based on the provisions written in the last line of the initial plan generation prompt Pp2 shown in Figure 12. For example, the first line of the inference execution result Ri2 shows the execution result "#1, Dental X-ray examination, Infected root canal treatment, Avoid brushing this area".

[0094] <Structure of Input / Output Data for Anomaly Inference AI> Next, with reference to Figures 15 to 17, the structure of the data input to the Anomaly Inference AI 132 and the data output from the Anomaly Inference AI 132 will be explained. The data input to the Anomaly Inference AI 132 includes the Anomaly Inference Prompt Pp3 and the Training Data Dt3, while the data output from the Anomaly Inference AI 132 includes the Inference Execution Result Ri3.

[0095] [Structure of the Abnormal Inference Prompt] Figure 15 shows an example of the structure of the abnormal inference prompt Pp3 that is input to the abnormal inference AI 132. As shown in Figure 15, the first line of the abnormal inference prompt Pp3 contains a sentence specifying the conditions for the response from the abnormal inference AI 132.

[0096] Furthermore, the second and third lines of the abnormal reasoning prompt Pp3 contain instructions for the abnormal reasoning AI 132, stating: "Compare the 'treatment plan' entered by the user with the 'treatment progress' actually performed on the patient, and present any possible 'concerns.' If any concerns are found, explain what discrepancies exist between the 'treatment plan' and the 'treatment progress.'"

[0097] The lines from the fourth line onward specify the data output method and content. The data output method and content are specified by the following sentence: "The output format will be JSON format with 'Deviation' and 'Concerns' as property names. If no concerns are found, simply output 'There are no possible concerns.'" Note that the "JSON (JavaScript Object Notation) format" specified as the data output format in the abnormal inference prompt Pp3 is just one example, and the data output format for the abnormal inference AI 132 may be in other formats.

[0098] [Structure of Training Data] Figure 16 shows an example of the structure of training data Dt3 to be trained on the abnormality inference AI 132. The training data Dt3 shown in Figure 16 is a combination of "treatment plan" and "treatment progress" to which "concerns" are associated. The "treatment plan" is the "treatment plan" obtained from the initial plan DB 53, and the "treatment progress" is the treatment progress information described in the "therapy and procedures" section obtained from the progress record DB 54. "Concerns" are abnormalities that are assumed to cause discrepancies between the treatment plan and the treatment progress, that is, matters that are a concern, such as oversights during examination.

[0099] For example, the training data Dt3 shown in Figure 16 illustrates an example where the treatment plan is "X-ray examination → infected root canal treatment → medication → root canal filling → PZ imp BT → Set crown," and the treatment progress is "X-ray examination → infected root canal treatment → medication → medication → medication." In other words, the training data Dt3 shows an example where there is a discrepancy between the plan and the progress, where the treatment does not proceed to "root canal filling" after the "medication" procedure, but instead the "medication" process is repeated.

[0100] Furthermore, the training data Dt3 lists abnormalities such as "Dental problems: Hidden root canals (some teeth may have unexpected additional root canals, and complete treatment may not be possible if they are not found)" as "concerns" for this combination.

[0101] The anomaly inference AI 132 then infers anomalies that can be expected from combinations of treatment plans and treatment progress, based on the training data it has learned. The anomaly inference AI 132 can improve the accuracy of its inferences by learning a large amount of such training data generated based on case examples.

[0102] [Structure of the inference execution result] Figure 17 shows an example of the inference execution result Ri3 by the abnormal inference AI 132. As shown in Figure 17, the inference execution result Ri3 by the abnormal inference AI 132 is output in "JSON format with "Deviation" and "Concerns" as property names" based on the provisions described in the last line of the abnormal inference prompt Pp3 shown in Figure 15. For example, the inference execution result Ri3 outputs the abnormal inference execution result "The number of times the patch has been applied has been more than the initial plan" with the property name "Deviation".

[0103] Furthermore, the inference result Ri3 outputs the inference result "Hidden root canals (Some teeth may have unexpected additional root canals, and complete treatment may not be possible if they are not found)" under the property names "Concerns" and "Dental problems".

[0104] <Structure of Input / Output Data of Summary AI> Next, with reference to Figures 18 and 19, the structure of the data input to and output from Summary AI 142 will be described. The data input to Summary AI 142 includes the summary creation prompt Pp4, and the data output from Summary AI 142 includes the summary creation execution result Ri4.

[0105] [Structure of the Summary Creation Prompt] Figure 18 shows an example of the structure of the summary creation prompt Pp4 that is input to the summary AI 142. As shown in Figure 18, the first line of the summary creation prompt Pp4 contains a sentence specifying the conditions for the response from the summary AI 142.

[0106] Furthermore, the second line of the summary creation prompt Pp4 contains the following instruction for the summary AI 142: "Based on the 'basic data,' 'progress record,' and 'abnormalities' entered by the user, summarize the information in a way that is easy for medical staff who have never been involved in the treatment of this patient to understand." The third line of the summary creation prompt Pp4 contains the following instruction for the summary AI 142: "When summarizing, first describe the contents of the 'basic data,' then, referring to the 'progress record,' summarize the progress of the treatment in chronological order, with a maximum of 200 characters per day."

[0107] Furthermore, the fourth and fifth lines of the summary creation prompt Pp4 contain text specifying conditions for the response from the summary AI 142. In addition, the last line of the summary creation prompt Pp4 defines the data output format from the summary AI 142.

[0108] [Structure of Summary Creation Execution Results] Figure 19 shows an example of the summary creation execution result Ri4 by the summary AI 142. As shown in Figure 19, the summary creation execution result Ri4 by the summary AI 142 is output in "JSON format with each item as a property name" based on the specifications written on the last line of the summary creation prompt Pp4 shown in Figure 18. For example, the summary creation execution result Ri4 outputs the summary creation execution result such as "Patient ID: "P-0001", "Name": "Taro Nippon" as the property name for "Patient Data".

[0109] <Screen Configuration> Next, with reference to Figures 20 to 29, examples of the configurations of various screens presented to the user via the output unit 24 (see Figure 2) will be described.

[0110] [Basic Data Display Screen] Figure 20 shows an example of the configuration of the medical record creation support screen Sc. The medical record creation support screen Sc shown in Figure 20 is the screen displayed on the output unit 24 of the terminal device 3 when creating a patient's medical record. Menus such as "File," "Edit," and "View" are displayed at the top of the medical record creation support screen Sc. Below the area where the menus are displayed, there are three areas from left to right on the screen: the oral information display area Ar1, the medical record creation area Ar2, and the patient information display area Ar3.

[0111] The oral information display area Ar1 shows information about the patient's oral cavity. The medical record creation area Ar2 displays information related to the creation of the medical record. The patient information display area Ar3 displays information about the patient, such as the patient's age, gender, and medical history. In the example shown in Figure 20, the patient's basic data is displayed in the patient information display area Ar3. The basic data is data stored in the basic data DB51 (see Figure 4).

[0112] At the top of the patient information display area Ar3, the patient name "Taro Nippon" and a Problem AI button B1 are provided. This Problem AI button B1 is a menu that causes the Problem List Creation Support Unit 101 to generate a problem list in the Problem List AI 102. When the Problem AI button B1 is pressed by the user, the Problem List Creation Screen Pu1 shown in Figure 21 pops up on the Medical Record Creation Support Screen Sc.

[0113] [Problem List Creation Screen] Next, the configuration of the problem list creation screen Pu1 will be explained with reference to Figure 21. Figure 21 is a diagram showing an example of the configuration of the problem list creation screen Pu1. At the upper right corner of the problem list creation screen Pu1 shown in Figure 21, there is a button labeled "Create problem list for this patient". When this button is pressed by the user, the problem list creation support unit 101 retrieves basic data associated with the patient ID from the basic data DB 51 and sends the retrieved basic data and prompt to the problem list AI 102.

[0114] The problem list AI 102 then creates a patient problem list based on the input basic data and prompts, and sends the created problem list to the problem list creation support unit 101. The problem list creation support unit 101 displays the problem list sent from the problem list AI 102 on the problem list creation screen Pu1.

[0115] The problem list displayed on the problem list creation screen Pu1 shows the following items: "Problem Category," "Problem Number," "Onset Date," "Problem," and "Status." The contents of each of these items can be modified by the user. A pull-down button is provided at the far right of the display field for the "Status" item. When this button is pressed by the user, the options "Active" and "Inactive" are displayed as a pull-down list. The user can select the appropriate status for the patient's problem from the options.

[0116] In the lower right corner of the area where the problem list is displayed, there are two buttons: a Cancel button B2 and a Save button B3. The Cancel button B2 is used to cancel the creation of the problem list. The Save button B3 is used to save the created problem list to the problem list DB 52. When the user presses the Save button B3, the created problem list is saved to the problem list DB 52 and displayed on the medical record creation support screen Sc.

[0117] Figure 22 shows an example configuration of the medical record creation support screen Sc with a problem list displayed. In the medical record creation area Ar2 of the medical record creation support screen Sc shown in Figure 22, the problem list is displayed. A planning AI button B4 is located at the upper left corner of the area where the problem list is displayed. When this planning AI button B4 is pressed by the user, the initial planning creation screen Pu2 shown in Figure 23 pops up on the medical record creation support screen Sc.

[0118] [Initial Plan Creation Screen] Next, the configuration of the initial plan creation screen Pu2 will be explained with reference to Figure 23. Figure 23 is a diagram showing an example of the configuration of the initial plan creation screen Pu2. At the upper right corner of the initial plan creation screen Pu2 shown in Figure 23, there is a button labeled "Create initial plan for this patient". When this button is pressed by the user, the initial plan creation support unit 111 retrieves a list of problems associated with the patient ID from the problem list DB 52 and sends the retrieved problem list and prompts to the initial plan AI 112. The initial plan AI 112 then creates an initial plan for the patient based on the input problem list and prompts.

[0119] More specifically, the diagnostic plan learning model (not shown) of the initial planning AI 112 creates a diagnostic plan based on the input problem list and prompts, the treatment plan learning model (not shown) creates a treatment plan, and the education plan learning model creates an education plan. The initial planning AI 112 then transmits these created initial plans to the initial plan creation support unit 111. The initial plan creation support unit 111 displays the initial plans transmitted from the initial planning AI 112 on the initial plan creation screen Pu2.

[0120] The initial plan creation screen Pu2 displays the following items in the initial plan section: "Problem Number," "Problem," "Diagnostic Plan," "Treatment Plan," and "Education Plan." The content of each of these items can be modified by the user. In the example shown in Figure 23, the underlined section "Explained that #1 was the cause" in the "Education Plan" item for problem number #2 is the part that was modified (added) by the user.

[0121] In the lower right corner of the area where the initial plan is displayed, there are two buttons: a Cancel button B2 and a Save button B3. The Cancel button B2 is used to cancel the creation of the initial plan. The Save button B3 is used to save the created initial plan to the Initial Plan DB 53. When the user presses the Save button B3, the created initial plan is saved to the Initial Plan DB 53 and displayed on the Medical Record Creation Support Screen.

[0122] Figure 24 shows an example of the configuration of the medical record creation support screen Sc with the initial plan displayed. In the medical record creation area Ar2 of the medical record creation support screen Sc shown in Figure 24, the initial plan is displayed, and a progress record button B5 is located at the upper left corner of the area where the initial plan is displayed. When this progress record button B5 is pressed by the user, the progress record creation screen Pu3 shown in Figure 25 pops up on the medical record creation support screen Sc.

[0123] [Progress Record Creation Screen] Next, the configuration of the progress record creation screen Pu3 will be explained with reference to Figure 25. Figure 25 is a diagram showing an example of the configuration of the progress record creation screen Pu3. Figure 25 shows the state in which progress records have been written to the progress record creation screen Pu3. Initially, the progress record creation screen Pu3 displays only the items "Date," "Problem Number," "Location," and "Therapy and Treatment." In addition, the "Therapy and Treatment" item also displays the categories "S:," "O:," "A:," and "P:." Users such as dentists can enter the details of the diagnosis corresponding to each category of SOAP.

[0124] Furthermore, when using the technology described in Patent Document 1, a button for creating progress records may be provided at the upper right corner of the progress record creation screen Pu3, and when this button is pressed by the user, the medical record creation screen of Patent Document 1 may be displayed. In the text input area of ​​the medical record creation screen of Patent Document 1, when the classification button is pressed after the examination details, etc. have been entered, text is entered into the text input area. The entered text is divided into phrases by the text division unit and then classified into one of the SOAP categories by the phrase classification unit. The classified text can then be recorded in the progress record DB 54 by the progress record creation unit 12 according to this embodiment.

[0125] The contents of each item displayed on the progress record creation screen Pu3 can be modified by the user. A cancel button B2 and a save button B3 are located in the lower right corner of the area where the progress record is displayed. The cancel button B2 is a button that instructs the user to cancel the creation of the progress record. The save button B3 is a button that instructs the user to save the created progress record to the progress record DB 54. When the user presses the save button B3, the created initial plan is saved to the progress record DB 54, and the created progress record is displayed on the medical record creation support screen Sc.

[0126] Figure 26 shows an example configuration of the medical record creation support screen Sc with the progress record displayed. In the medical record creation area Ar2 of the medical record creation support screen Sc shown in Figure 26, the progress record is displayed, and an abnormal AI button B6 is located at the upper left corner of the area where the progress record is displayed. When this abnormal AI button B6 is pressed by the user, the abnormal confirmation screen Pu4 shown in Figure 27 pops up on the medical record creation support screen Sc.

[0127] [Anomaly Confirmation Screen] Next, the configuration of the anomaly confirmation screen Pu4 will be explained with reference to Figure 27. Figure 27 is a diagram showing an example of the configuration of the anomaly confirmation screen Pu4. At the upper right corner of the anomaly confirmation screen Pu4 shown in Figure 27, there is a button labeled "Infer Anomaly". When this button is pressed by the user, the anomaly confirmation unit 131 obtains the initial plan associated with the patient ID from the initial plan DB 53 and obtains the progress record associated with the patient ID from the progress record DB 54.

[0128] The anomaly detection unit 131 then transmits the acquired initial plan, progress record, and prompt to the anomaly inference AI 132. Based on the input initial plan, progress record, and prompt, the anomaly inference AI 132 infers anomalies that are likely to be the cause of the discrepancy between the treatment plan and the treatment progress. The anomaly inference AI 132 transmits the content of the inferred anomaly to the anomaly detection unit 131. The anomaly detection unit 131 displays the content of the anomaly transmitted from the anomaly inference AI 132 on the anomaly confirmation screen Pu4.

[0129] On the abnormality confirmation screen Pu4, the inferred content of the abnormality is displayed in the "Inference Result" item, associated with the problem number. In the example shown in Figure 27, the top line of the "Inference Result" item displays a message regarding the discrepancy between the initial plan and the progress record, such as "The number of times the medication has been applied has been greater than initially planned." In the subsequent lines, abnormalities that are assumed to be the cause of the discrepancy between the treatment plan and the treatment progress are displayed, associated with problem categories such as "Dental Problem" and "Medical Problem."

[0130] In the lower right corner of the area where the anomaly inference is displayed, there are two buttons: Cancel button B2 and Save button B3. Cancel button B2 is a button that instructs the user to cancel the confirmation of the anomaly. Save button B3 is a button that instructs the user to save the inferred anomaly to the anomaly database 55. When the user presses Save button B3, the inferred anomaly is saved to the anomaly database 55, and the inferred anomaly is displayed on the medical record creation support screen Sc.

[0131] Figure 28 shows an example configuration of the medical record creation support screen Sc where the anomaly inference results are displayed. In the medical record creation area Ar2 of the medical record creation support screen Sc shown in Figure 28, the anomaly inference results are displayed, and a summary AI button B7 is located at the upper left corner of the area where the anomaly inference results are displayed. When this summary AI button B7 is pressed by the user, the summary creation screen Pu5 shown in Figure 29 is displayed as a pop-up on the medical record creation support screen Sc.

[0132] [Summary Creation Screen] Next, the configuration of the summary creation screen Pu5 will be explained with reference to Figure 29. Figure 29 is a diagram showing an example of the configuration of the summary creation screen Pu5. At the upper right corner of the summary creation screen Pu5 shown in Figure 29, there is a button labeled "Create summary for this patient". When this button is pressed by the user, the summary creation support unit 141 obtains basic data associated with the patient ID from the basic data DB 51, obtains progress records from the progress record DB 54, and obtains the anomaly inference results from the anomaly DB 55.

[0133] The summary creation support unit 141 then transmits the acquired basic data, progress records, anomaly inference results, and prompts to the summary AI 142. The summary AI 142 then creates a patient summary based on the input basic data, progress records, anomaly inference results, and prompts. The summary AI 142 then transmits the created summary to the summary creation support unit 141. The summary creation support unit 141 displays the summary transmitted from the summary AI 142 on the summary creation screen Pu5.

[0134] The upper half of the summary creation screen Pu5 displays basic data such as the patient's ID, name, and gender, while the lower half displays the treatment progress and concerns. The "Treatment Progress" section displays chronological information from the progress records DB54, which is associated with the patient ID. The "Concerns" section displays information about abnormalities associated with the patient ID in the abnormality DB55. The content of each of these items that make up the summary creation screen Pu5 can be modified by the user.

[0135] A close button B8 is located in the lower right corner of the area where the summary is displayed. When the close button B8 is pressed by the user, the summary creation screen Pu5 is closed. Note that the buttons provided on the summary creation screen Pu5 are not limited to just the close button B8. Other buttons may be provided, such as a save button to instruct the user to save the created summary to the summary DB (not shown), a print button to instruct the user to print the summary, and a QR code button to instruct the user to create a QR code (registered trademark) of the summary.

[0136] <Medical Record Creation Support Method> Next, with reference to Figures 30 to 34, a medical record creation support method using the medical record creation support system 100 according to this embodiment will be described. In the medical record creation support method using the medical record creation support system 100 according to this embodiment, a problem list creation process, an initial plan creation process, a progress record creation process, anomaly inference processing, and a summary creation process are performed.

[0137] [Problem List Creation Process] Figure 30 is a flowchart showing an example of the procedure for the problem list creation process. The problem list creation process is started when the pressing of the problem AI button B1 is detected on the medical record creation support screen Sc shown in Figure 20.

[0138] When the pressing of the Problem AI button B1 is detected on the medical record creation support screen Sc, the patient ID is sent from the medical record management server 2 (see Figure 1) to the problem list creation support unit 101 (step S1). Next, the problem list creation support unit 101 obtains basic data associated with the patient ID sent from the medical record management server 2 from the basic data DB 51 (step S2). Next, the problem list creation support unit 101 sends the obtained basic data and prompts to the problem list AI 102 (step S3). Next, the problem list AI 102 extracts the patient's problems from the basic data sent from the problem list creation support unit 101 and returns the extracted problems as a problem list to the problem list creation support unit 101 (step S4).

[0139] Next, the problem list creation support unit 101 displays the problem list sent from the problem list AI 102 on the medical record creation support screen Sc and accepts editing by the user (step S5). Then, the problem list creation support unit 101 stores the edited or unedited problem list in the problem list DB 52 (step S6). After the processing in step S6, the problem list creation process by the problem list creation support unit 101 and the problem list AI 102 is completed.

[0140] [Initial Plan Creation Process] Next, the initial plan creation process by the initial plan creation support unit 111 and the initial plan AI 112 will be described with reference to Figure 31. Figure 31 is a flowchart showing an example of the procedure for the initial plan creation process. The initial plan creation process is started when the pressing of the plan AI button B4 is detected on the medical record creation support screen Sc shown in Figure 22.

[0141] When the pressing of the planning AI button B4 on the medical record creation support screen Sc is detected, the patient ID is sent from the medical record management server 2 to the initial planning creation support unit 111 (step S11). Next, the initial planning creation support unit 111 obtains a problem list associated with the patient ID sent from the medical record management server 2 from the problem list DB 52 (step S12). Then, the initial planning creation support unit 111 sends the obtained problem list and prompts to the initial planning AI 112 (step S13). Next, the initial planning AI 112 creates an initial plan based on the problem list sent from the initial planning creation support unit 111 and returns the created initial plan to the initial planning creation support unit 111 (step S14).

[0142] Next, the initial plan creation support unit 111 displays the initial plan transmitted from the initial plan AI 112 on the medical record creation support screen Sc and accepts editing by the user (step S15). Next, the initial plan creation support unit 111 stores the edited or unedited initial plan in the initial plan DB 53 (step S16). After the processing in step S16, the initial plan creation process by the initial plan creation support unit 111 and the initial plan AI 112 is completed.

[0143] [Progress Record Creation Process] Next, the progress record creation process by the progress record creation unit 12 will be described with reference to Figure 32. Figure 32 is a flowchart showing an example of the procedure for the progress record creation process. The progress record creation process is started when the pressing of the progress record button B5 is detected on the medical record creation support screen Sc shown in Figure 24.

[0144] When the press of the progress record button B5 is detected on the medical record creation support screen Sc, the patient ID is sent from the medical record management server 2 to the progress record creation unit 12 (step S21). Next, the progress record creation unit 12 displays the progress record creation screen Pu3 (Figure 25) superimposed on the medical record creation support screen Sc and accepts the user's input of progress records (step S22). Next, the progress record creation unit 12 stores the entered progress records in the progress record DB 54 (step S23). After the processing in step S23, the progress record creation process by the progress record creation unit 12 is completed.

[0145] [Anomaly Inference Processing] Next, the anomaly inference processing by the anomaly confirmation unit 131 and the anomaly inference AI 132 will be described with reference to Figure 33. Figure 33 is a flowchart showing an example of the procedure for anomaly inference processing. Anomaly inference processing is started when the pressing of the anomaly AI button B6 is detected on the medical record creation support screen Sc shown in Figure 26.

[0146] When the abnormal AI button B6 is pressed on the medical record creation support screen Sc, the medical record management server 2 (see Figure 2) transmits the patient ID and problem number to the abnormality confirmation unit 131 (step S31). Next, the abnormality confirmation unit 131 obtains the initial plan corresponding to the patient ID and problem number transmitted from the medical record management server 2 from the initial plan DB 53 and obtains the progress record from the progress record DB 54 (step S32).

[0147] The anomaly inference AI 132 retrieves the treatment plan from the initial plan transmitted from the anomaly confirmation unit 131 and retrieves the latest treatment progress from the progress record (step S33). Next, the anomaly confirmation unit 131 transmits the acquired treatment plan, the latest treatment progress, and a prompt to the anomaly inference AI 132 (step S34).

[0148] The abnormality inference AI 132 infers abnormalities from the treatment plan and treatment progress transmitted from the abnormality confirmation unit 131 and returns the abnormality inference result to the abnormality confirmation unit 131 (step S35). Next, the abnormality confirmation unit 131 displays the abnormality inference result transmitted from the abnormality inference AI 132 on the medical record creation support screen Sc and accepts editing by the user (step S36). Next, the abnormality confirmation unit 131 stores the edited or unedited abnormality inference result in the abnormality DB 55 (step S37). After the processing in step S37, the abnormality inference processing by the abnormality confirmation unit 131 and the abnormality inference AI 132 is completed.

[0149] [Summary Creation Process] Next, the summary creation process by the summary creation support unit 141 and the summary AI 142 will be described with reference to Figure 34. Figure 34 is a flowchart showing an example of the procedure for the summary creation process. The summary creation process is started when the press of the summary AI button B7 is detected on the medical record creation support screen Sc shown in Figure 28.

[0150] When the Summary AI button B7 is pressed on the medical record creation support screen Sc, the patient ID is transmitted from the medical record management server 2 to the summary creation support unit 141 (step S41). Next, the summary creation support unit 141 obtains the basic data associated with the patient ID transmitted from the medical record management server 2 from the basic data DB 51, obtains the progress record from the progress record DB 54, and obtains the anomaly inference result from the anomaly DB 55 (step S42).

[0151] The summary creation support unit 141 transmits the acquired basic data, progress records, anomaly inference results, and prompts to the summary AI 142 (step S43). Next, the summary AI 142 creates a summary based on the basic data, progress records, and anomaly inference results transmitted from the summary creation support unit 141, and returns the created summary to the summary creation support unit 141 (step S44).

[0152] Next, the summary creation support unit 141 displays the summary sent from the summary AI 142 on the medical record creation support screen Sc (step S45). After the processing in step S45, the summary creation process by the summary creation support unit 141 and the summary AI 142 is completed.

[0153] In the embodiment described above, the initial plan AI 112 of the initial plan creation unit 11 creates an initial treatment plan for the patient based on basic data obtained from the patient and stores it in the initial plan DB 53. The progress record creation unit 12 stores information on the patient's treatment progress carried out based on the initial plan in the progress record DB 54. Furthermore, the abnormality inference AI 132 of the abnormality inference unit 13 infers abnormalities that are assumed to be the cause of the discrepancy between the initial plan and the treatment progress, based on the patient's basic data, the initial plan, and the treatment progress information, and outputs the abnormality inference result. Therefore, according to this embodiment, even if an oversight occurs during the consultation or treatment, or if there is an error in judgment due to a lack of knowledge or experience on the part of the dentist, the dentist will be able to perform appropriate treatment based on the abnormality information in the inference result.

[0154] Furthermore, in the embodiment described above, the abnormality inference unit 13 includes an abnormality inference AI 132, and the abnormality inference AI 132 is composed of a large-scale language model, or a learning model that has learned the correspondence between a patient's treatment plan, the treatment progress based on the treatment plan, and abnormalities.

[0155] Therefore, for example, if the anomaly inference AI 132 is composed of a large-scale language model, the anomaly confirmation unit 131 can appropriately infer anomalies by appropriately configuring the content of the prompts that the anomaly confirmation unit 131 inputs to the anomaly inference AI 132. Also, if the anomaly inference AI 132 is composed of a learning model, the anomaly inference AI 132 can accurately infer anomalies based on information obtained in the clinical setting by creating training data for the learning model based on past clinical data.

[0156] Furthermore, in the embodiment described above, the problem list AI 102 of the problem list creation unit 10 extracts elements that are problematic for the patient from the basic data and creates a problem list.The initial plan stored in the initial plan DB 53, the treatment progress stored in the progress record DB 54, and the abnormalities inferred by the abnormality inference AI 132 are managed in association with each problem included in the problem list.Currently, the accuracy of judgment regarding what may be problematic in the information shown in the basic data varies greatly depending on the dentist's knowledge and experience, but according to this embodiment, the problem list AI 102 can appropriately extract problems from the patient's basic data.

[0157] Furthermore, the problem list in the above-described embodiment includes problem classifications, and these classifications include at least dental problems and medical problems. In other words, according to this embodiment, abnormalities are inferred by the abnormality inference AI 132 using information on problems classified as medical problems as well as dental problems. In other words, according to this embodiment, abnormalities are inferred and presented based on a multifaceted perspective that is not limited to dental problems alone, so even dentists with little medical knowledge can make an appropriate diagnosis based on the abnormality information output from the abnormality inference unit 13.

[0158] Furthermore, in the embodiment described above, the problem list creation unit 10 includes a problem list AI 102, which is composed of a learning model that has learned the correspondence between patient problems included in the basic data of past clinical data and the problem categories of those problems. Therefore, for example, if the problem list AI 102 is composed of a large-scale language model, the problem list AI 102 can output problems included in the basic data in correspondence with appropriate problem categories by appropriately configuring the content of the prompts that the problem list creation support unit 101 inputs to the problem list AI 102. Also, if the problem list AI 102 is composed of a learning model, the problem list AI 102 can accurately extract problems from the basic data based on information obtained in the clinical setting by creating learning data to train the learning model based on past clinical data.

[0159] Furthermore, in the embodiment described above, the problem list creation unit 10 creates the problem list in a predetermined format. Therefore, according to this embodiment, regardless of the type of specialist the patient's attending physician is, and regardless of their experience level, the patient's condition can be appropriately understood by checking the problem list.

[0160] Furthermore, in the embodiment described above, the initial plan AI 112 of the initial plan creation unit 11 creates initial plans for patient diagnosis, treatment, and education. Therefore, according to this embodiment, dentists can appropriately diagnose patients based on the diagnostic plan created by the initial plan creation unit 11, and appropriately treat patients based on the treatment plan created by the initial plan creation unit 11. In addition, dentists can appropriately educate patients based on the treatment plan created by the initial plan creation unit 11. As a result, patients can develop habits of autonomously caring for their oral cavity based on appropriate education, which can reduce the frequency of visits to the dental clinic and save patients time and medical expenses.

[0161] Furthermore, according to the embodiment described above, the server 1 creates a problem list from the patient's basic data, formulates an initial plan, records the treatment progress based on a description format such as SOAP, and creates a summary. In other words, the medical record is created based on the POMR, which is a description format in line with the POS. Therefore, according to this embodiment, the information recorded in the medical record is organized, so information can be shared accurately and reliably among dentists, dental assistants, or dental hygienists.

[0162] Furthermore, acquiring the skills to create medical records based on POS and POMR requires not only "communication skills" to extract valuable information from patients, "analytical skills" to analyze that information from multiple perspectives, and "writing skills" to output it as written text in the medical record, but also sufficient "time" to perform the actual work. According to this embodiment, even dentists who lack such skills or time can easily create high-quality medical records.

[0163] Furthermore, in this embodiment, the readability of the medical record is improved because the medical record is created based on the POM (Position-Oriented Medical Record) using a medical record format such as SOAP. Therefore, according to this embodiment, even if a patient requests disclosure or explanation of evidence of treatment, dentists can respond appropriately, thereby preventing disputes with patients. In addition, by creating the medical record based on the POM and SOAP format, the created medical record can withstand insurance guidance (audits) in insurance-covered medical treatment.

[0164] Furthermore, in this embodiment, since medical records are created according to the POMR format, which is based on POS, it becomes easier to realize the goal of providing medical care that aims to solve the problem most effectively while also valuing the patient's quality of life (QOL), as advocated by POS. Therefore, according to this embodiment, it becomes possible to provide optimal medical care to each individual patient, thereby improving the patient's QOL and contributing to society. In addition, because diagnosis, treatment, and education are carried out based on POS, it becomes possible to realize comprehensive treatment and care for patients, making it easier for dentists to gain the trust of patients and enabling the dental clinic to aim for stable management.

[0165] Furthermore, according to this embodiment, it becomes possible to use medical records created based on POM (Patient-Assisted Medical Record) to provide training to personnel. Specifically, for example, it becomes possible to use medical records created based on POM to provide training on how to communicate with patients, how to propose optimal treatment (care), and how to make decisions from multiple perspectives.

[0166] In the embodiments described above, an example was given in which the medical record was entered in SOAP format, but the present invention is not limited to this. The medical record may be entered in other formats such as SOAPIE (Subject Object Assessment Plan Intervention Evaluation).

[0167] Furthermore, while the above-described embodiment example showed that the treatment progress created by the progress record creation unit 12 is a medical record (electronic medical record) in dental practice, the present invention can also be applied when creating medical records and claims for medical treatment by doctors, nurses, physical therapists, judo therapists, etc. Moreover, the medical record creation support system of the present invention can be applied not only to dentistry but also to various medical fields such as medicine, pharmacy, and veterinary medicine.

[0168] Furthermore, while the above-described embodiment shows an example in which the problem list AI 102, initial planning AI 112, anomaly inference AI 132, and summary AI 142 are all constructed using AI, the present invention is not limited to this. Some or all of these functional units may be constructed using methods other than AI, such as programming.

[0169] Furthermore, the medical record creation support system 100 according to this embodiment is configured to enable the abnormality inference unit 13 to appropriately infer abnormalities. For this reason, it is configured to include all components: the problem list creation unit 10, the initial plan creation unit 11, the progress record creation unit 12, and the abnormality inference unit 13 (and the summary creation unit 14). However, from the viewpoint of enabling even dentists with little experience or knowledge to appropriately create problem lists, initial plans, summaries, etc., the medical record creation support device of the medical record creation support system according to the present invention may also adopt the following configuration.

[0170] (1) A medical record creation support device comprising: a learning model that has learned learning data in which health problems faced by patients are associated with problem categories into which those problems should be classified; a problem list inference unit (problem list AI) that takes basic data, which is basic information about the patient obtained from the patient, as input, infers the correspondence between the patient's problems and problem categories included in the basic data, and outputs the inference result; and a problem list creation (support) unit that uses the inference result output from the problem list inference unit to create a problem list in which information including the correspondence between the patient's problems and problem categories is displayed in a list. (2) A medical record creation support device comprising: a learning model that has learned learning data in which health problems faced by patients are associated with plans for addressing those problems; an initial plan inference unit (initial plan AI) that takes the patient's problems as input, infers and outputs a plan for addressing those problems; and an initial plan creation (support) unit that uses the plan output from the initial plan inference unit to create and output a plan list in which plans for the patient are displayed in a list. (3) A medical record creation support device that takes as input basic data, which is basic information about the patient obtained from the patient, progress records, which are records of the treatment progress of the patient, and abnormal information, which is created based on the information of a problem list that lists the health problems the patient has, which is generated from the basic data, and shows the discrepancy between the initial plan for responding to the patient and the treatment progress, as a summary creation unit and outputs a summary of the treatment for the patient.

[0171] In the configuration described in (2) above, the problem list necessary for creating the initial plan may be created using methods such as programming, rather than by the problem list creation unit 10. Also, in the configuration described in (3) above, the information on anomaly inference results necessary for creating the summary may be created using methods such as programming, rather than by the anomaly inference unit 13.

[0172] Furthermore, the embodiments described above are intended to explain the configuration of the apparatus (server, terminal device) and system (medical record creation support system) in detail and specifically in order to make the present invention easier to understand, and are not necessarily limited to those comprising all the configurations described.

[0173] Furthermore, the control lines or information lines shown as solid lines in Figures 1 and 2 are those deemed necessary for explanation and do not necessarily represent all control lines or information lines in the actual product. In practice, it can be assumed that almost all components are interconnected.

[0174] 1...Server, 2...Medical record management server, 3...Terminal device, 10...Problem list creation unit, 11...Initial plan creation unit, 12...Progress record creation unit, 13...Anomaly inference unit, 14...Summary creation unit, 51...Basic data DB, 52...Problem list DB, 53...Initial plan DB, 54...Progress record DB, 55...Anomaly DB, 100...Medical record creation support system, 101...Problem list creation support unit, 102...Problem list AI, 111...Initial plan creation support unit, 112...Initial plan AI, 131...Anomaly confirmation unit, 132...Anomaly inference AI, 141...Summary creation support unit, 142...Summary AI

Claims

1. A medical record creation support device comprising: an initial plan creation unit that creates an initial treatment plan for a patient based on basic data, which is basic information about the patient obtained from the patient, and stores it in an initial plan database; a progress record creation unit that stores information on the patient's treatment progress carried out based on the initial plan in a progress record database; and an abnormality inference unit that infers abnormalities that are assumed to be the cause of discrepancies between the initial plan and the treatment progress, based on the patient's initial plan and treatment progress information, and outputs the inference results of the abnormalities.

2. The medical record creation support device according to claim 1, wherein the abnormality inference unit includes a large-scale language model, or a learning model that has learned the correspondence between the patient's treatment plan, the treatment progress based on the treatment plan, and the abnormality.

3. The medical record creation support device according to claim 2, further comprising a problem list creation unit that extracts elements that are problematic for the patient from the basic data and creates a problem list, wherein the initial plan stored in the initial plan database, the treatment progress stored in the progress record database, and the abnormalities inferred by the abnormality inference unit are managed in association with each problem included in the problem list.

4. The medical record creation support device according to claim 3, wherein the problem list includes problem categories for the problems, and the problem categories include at least dental problems and medical problems.

5. The medical record creation support device according to claim 4, wherein the problem list creation unit includes a large-scale language model or a learning model that has learned the correspondence between patient problems included in the basic data of past clinical data and the problem categories.

6. The medical record creation support device according to claim 4, wherein the initial planning unit creates initial plans for the diagnosis, treatment, and education of the patient.

7. The medical record creation support device according to claim 6, wherein the progress record creation unit stores the patient's treatment progress in the progress record database in correspondence with each category determined by the medical record entry method.

8. A medical record creation support method comprising: a procedure in which an initial plan creation unit creates an initial treatment plan for a patient based on basic data, which is basic information about the patient obtained from the patient, and stores it in an initial plan database; a procedure in which a progress record creation unit stores information on the patient's treatment progress implemented based on the initial plan in a progress record database; and a procedure in which an abnormality inference unit infers abnormalities that are assumed to be the cause of discrepancies between the initial plan and the treatment progress, based on the patient's initial plan and treatment progress information, and outputs the result of the abnormality inference.

Images