Injury and disease name estimation device, injury and disease name estimation method, and recording medium

Abstract

This injury and disease name estimation device comprises an acquisition unit, an extraction unit, an estimation unit, and an output unit. The acquisition unit acquires medical accounting data. The extraction unit extracts time-series medical care information for each patient from the medical accounting data. The estimation unit uses an estimation model for estimating the injury and disease name of a patient from the time-series medical care information, and estimates the injury and disease names of each of the patients on the basis of the acquired time-series medical care information of each of the patients. The output unit outputs information on the estimated injury and disease names. The injury and disease name estimation device can, for example, estimate the injury and disease name of a patient from the medical accounting data and thereby support decision making based on the injury or disease which is the main cause of the symptoms appearing in the patient.

Description

Injury / disease name estimation device, injury / disease name estimation method, and recording medium

[0001] The present disclosure relates to an injury / illness name estimation device and the like.

[0002] In order to understand the use of medications or to confirm the details of medical fee claims based on prescriptions, investigations into the illness or injury of a patient may be conducted. For example, investigations into the illness or injury of a patient may be conducted by entities that cannot directly review the medical records. When an entity that cannot directly review the medical records conducts an investigation into the illness or injury, for example, there is often insufficient information to identify the illness or injury the patient is suffering from. In such cases, the investigation is conducted by inferring the illness or injury the patient is suffering from based on information that the investigating entity can obtain, or information that can be obtained from sources other than the medical records.

[0003] The disease name inference system of Patent Document 1 generates a learning model by machine learning based on learning data in which prescription drugs in prescription data are associated with disease name data, and then infers the name of the disease for which the prescription drug is prescribed from the prescription data input to the learning model.

[0004] Japanese Patent Application Laid-Open No. 2021-60932

[0005] The technology described in Patent Document 1 may not be able to accurately estimate the illness or injury that a patient is suffering from.

[0006] In order to solve the above-mentioned problems, the present disclosure aims to provide an injury or illness name estimation device that can improve the accuracy of estimating the injury or illness that a patient is suffering from.

[0007] In order to solve the above problems, an illness or injury name estimation device according to one embodiment of the present disclosure includes an acquisition means for acquiring medical accounting data, an extraction means for extracting time-series medical information for each patient from the medical accounting data, an estimation means for estimating the name of the illness or injury for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the illness or injury for each patient from the time-series medical information, and an output means for outputting information related to the estimated name of the illness or injury.

[0008] A method for estimating the name of an illness or injury according to one aspect of the present disclosure acquires medical accounting data, extracts time-series medical information for each patient from the medical accounting data, estimates the name of the illness or injury for each patient using an estimation model that estimates the name of the patient's illness or injury from the time-series medical information based on the acquired time-series medical information for each patient, and outputs information related to the estimated name of the illness or injury.

[0009] A recording medium according to one embodiment of the present disclosure non-temporarily records an illness or injury name estimation program that causes a computer to execute the following processes: acquiring medical accounting data; extracting time-series medical information for each patient from the medical accounting data; estimating the name of each patient's illness or injury based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the patient's illness or injury from the time-series medical information; and outputting information related to the estimated name of the illness or injury.

[0010] According to the present disclosure, it is possible to improve the accuracy of estimating the injury or illness that a patient is suffering from.

[0011] 1 is a diagram illustrating an example of the configuration of a medical information system in the present disclosure; FIG. 2 is a diagram illustrating an example of the configuration of an injury or illness name estimation device in the present disclosure; FIG. 3 is a diagram illustrating an example of an estimation result of an injury or illness name in the present disclosure; FIG. 4 is a diagram illustrating an example of an estimation result of an injury or illness name in the present disclosure; FIG. 5 is a diagram illustrating an example of an estimation result of an injury or illness name in the present disclosure; FIG. 6 is a diagram illustrating an example of an operation flow of an injury or illness name estimation device in the present disclosure; FIG. 7 is a diagram illustrating an example of the hardware configuration of an injury or illness name estimation device in the present disclosure.

[0012] An embodiment of the present disclosure will be described in detail with reference to the drawings. FIG. 1 is a diagram illustrating an example of the configuration of a medical information system. The medical information system includes, for example, an injury / illness name estimation device 10, a medical accounting device 20, a medical data management device 30, and a terminal device 40. The injury / illness name estimation device 10 is connected to the medical accounting device 20, for example, via a network. The injury / illness name estimation device 10 is also connected to the terminal device 40, for example, via a network. The medical accounting device 20 is also connected to the medical data management device 30, for example, via a network. There may be multiple medical accounting devices 20, multiple medical data management devices 30, and multiple terminal devices 40. The number of medical accounting devices 20, multiple medical data management devices 30, and multiple terminal devices 40 can be set as appropriate.

[0013] The medical information system estimates the name of a patient's illness or injury based on, for example, medical accounting data. For example, when a medication is prescribed for a patient, the medical information system estimates the name of the illness or injury for which the medication is prescribed based on the medical accounting data. For example, the medical information system estimates the name of the illness or injury that is the main cause of the symptoms for which the medication is prescribed as the name of the illness or injury for which the medication is prescribed. In other words, the medical information system estimates the name of the illness or injury for which treatment is actually being performed. The illness or injury for which treatment is actually being performed is, for example, the illness or injury that is causing the symptoms for which treatment is being performed. The illness or injury for which treatment is actually being performed may be, for example, the main illness.

[0014] For example, when a patient is treated to relieve pain, the illness or injury that is the primary cause of the symptoms is the illness or injury that is causing the pain. For example, a patient with colon cancer may experience severe lower back pain. A doctor may prescribe a compress containing an analgesic and anti-inflammatory agent to a patient experiencing severe lower back pain due to colon cancer as a medication for the lower back pain. In this case, the illness or injury that is the primary cause of the symptoms is colon cancer. Furthermore, when claiming medical fees from insurance for the cost of the medication, if the illness or injury that is the primary cause of the symptoms is not covered by insurance, the medical fee payment claim may be submitted using the actual symptoms as the illness or injury. For example, a compress containing an analgesic and anti-inflammatory agent may be covered by insurance for lower back pain, but colon cancer is not. In this case, the medical fee payment claim is submitted for the medication for lower back pain, but the illness or injury that is the primary cause of the symptoms is, for example, colon cancer. In such cases, the medical information system, for example, estimates the actual illness or injury for which the medication is prescribed.

[0015] The medical information system, for example, estimates the name of an illness or injury based on time-series medical information included in medical accounting data. The medical accounting data is, for example, data for billing patients for medical expenses. The medical accounting data includes, for example, medical information about the patient. The patient's medical information includes, for example, the details of medical procedures performed on the patient and information about the date of the medical procedures. The details of the medical procedures include, for example, information indicating the name of the patient's illness or injury, examination, surgery, testing, injections, medication, treatment, rehabilitation, counseling, and the details of meals provided to the patient. The details of the medical procedures may include information indicating the body part to be treated. Treatment procedures include, for example, irradiation with radiation, shock waves, or ultraviolet rays, suction, removal, excision, or suturing. Treatment procedures are not limited to the above. Medical procedures are not limited to the above. The medical accounting data also includes data extracted from a portion of the medical accounting data. For example, the medical accounting data may be receipt or prescription data. A receipt is also called, for example, a medical fee statement, a dispensing fee statement, or a home nursing care fee statement.

[0016] Chronological medical information, for example, is information indicating the chronological order of medical procedures performed during a treatment period for an injury or illness and the details of the medical procedures. For example, if medical procedures include a physical examination, an abdominal X-ray, a colonoscopy, a biopsy, radiation therapy, administration of an anticancer drug, and prescription of an anti-inflammatory analgesic compress for lower back pain, the chronological medical information is information listing the details of the medical procedures in the above order. In this case, the anti-inflammatory analgesic compress may be treated as a prescription for lower back pain, for example. However, given that radiation therapy and administration of an anti-cancer drug have been performed, the lower back pain is likely caused by colon cancer. In such a case, the medical information system estimates that the illness suffered by a patient who has been prescribed an anti-inflammatory analgesic for lower back pain is colon cancer. In this way, the medical information system estimates, for example, the name of the illness that is the primary cause of the symptoms experienced by a patient who has been prescribed medication. Chronological medical information may, for example, be information indicating the date and time of each medical procedure performed during a treatment period for an injury or illness and the details of the medical procedure. The time-series medical information is not limited to the above.

[0017] Here, an example of the configuration of the injury / illness name estimation device 10 will be described. Fig. 2 is a diagram showing an example of the configuration of the injury / illness name estimation device 10. The injury / illness name estimation device 10 basically includes an acquisition unit 11, an extraction unit 12, an estimation unit 13, and an output unit 15. The injury / illness name estimation device 10 may further include, for example, a generation unit 14 and a storage unit 16.

[0018] The acquisition unit 11 acquires medical accounting data. The medical accounting data includes, for example, the date on which a medical procedure was performed and the details of the medical procedure. The acquisition unit 11 acquires, for example, medical accounting data for each patient whose illness or injury name is to be estimated. The acquisition unit 11 may further acquire at least one of electronic medical record data and test data. The acquisition unit 11 acquires the medical accounting data from, for example, the medical accounting device 20.

[0019] The acquisition unit 11 may acquire information specifying a patient whose illness or injury name is to be estimated. The information specifying the patient whose illness or injury name is to be estimated is, for example, information specifying data to be used for estimating the illness or injury name. For example, the acquisition unit 11 acquires information specifying a medication as information specifying the patient whose illness or injury name is to be estimated. In this case, the name of the illness or injury suffered by the patient is estimated based on the medical accounting data of the patient for whom the specified medication has been prescribed. The medication is specified using, for example, one or more pieces of information including the name of the medication, the name of the pharmaceutical company, the active ingredient, and the use of the medication. The medication may also be specified using information other than the above. The acquisition unit 11 may also acquire information specifying a patient whose illness or injury name is to be estimated. The information specifying the patient whose illness or injury name is to be estimated is, for example, one or more pieces of information including the patient's gender, age, medical history, the hospital where the patient is receiving treatment, the medical department where the patient is receiving treatment, the health insurance policy enrolled, and the medication prescribed. The information specifying the patient whose illness or injury name is to be estimated is not limited to the above. The acquisition unit 11 may acquire, for example, from the terminal device 40, information specifying a target for which the name of an injury or illness is to be estimated.

[0020] The extraction unit 12 extracts, for example, chronological medical information from the medical accounting data of each patient. The extraction unit 12 extracts, for example, chronological medical information based on the details of medical procedures recorded in the medical accounting data and the dates of the medical procedures. Examples of medical procedures include examinations, surgeries, injections, tests, medication, and rehabilitation. The details of medical procedures are, for example, information about at least one of the body part that is the subject of medical treatment and what was done during the medical treatment. For example, if a polyp that has developed in the large intestine is removed, the details of the medical procedure are colon polypectomy surgery. The details of medical procedures are not limited to the above.

[0021] The time-series medical information is, for example, a chronological history of medical procedures performed on each patient. The time-series medical information includes, for example, the content of the medical procedures and the order of the medical procedures. The time-series medical information may also include the content of the medical procedures, the order of the medical procedures, and the intervals or dates and times between the medical procedures.

[0022] The extraction unit 12 extracts medical information from the medical accounting data using, for example, an extraction model. The extraction model uses, for example, a dictionary containing terms related to medical procedures, the contents of the medical procedures, and body parts, and converts the medical accounting data and each of the terms included in the dictionary into an embedding vector. The extraction model extracts terms related to medical information from the medical accounting data based on, for example, the Euclidean distance or cosine similarity between the embedding vector converted from the medical accounting data and the embedding vector converted from the terms included in the dictionary. The extraction unit 12 then generates time-series medical information using dates associated with the terms and the extracted terms related to the medical procedures. Furthermore, for example, Word2Vec can be used for the conversion to a feature vector. Other language models may also be used for the conversion to a feature vector.

[0023] The extraction of medical information may also be performed using a language model. For example, a large-scale language model is used to extract the medical information. For example, the extraction unit 12 outputs a prompt to the language model requesting that the medical information be extracted from the medical accounting data and output in chronological order. Then, the extraction unit 12 acquires, for example, the chronological medical information as the output of the language model. For example, the language model used to extract the medical information may be GPT-2 (Generative Pre-trained Transformer-2), GPT-3, GPT-3.5, or GPT-4. For example, the language model used to extract the medical information may be T5 (Text-to-Text Transfer Transformer), BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), or ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately). The language model used to extract the medical information is not limited to the above.

[0024] For example, when medical accounting data is compiled by account, the extraction unit 12 extracts medical accounting data for each medical expense account for the same patient. Furthermore, when medical accounting data is compiled by visit, the extraction unit 12 extracts medical accounting data for each visit by the same patient. Furthermore, when medical accounting data is compiled by period, the extraction unit 12 extracts medical accounting data for the same patient for each of multiple periods. Furthermore, when medical accounting data is compiled by hospitalization, if a patient has repeated hospitalizations and discharges, the extraction unit 12 extracts medical accounting data for each hospitalization of the patient. Furthermore, when a patient has been transferred to another hospital, the extraction unit 12 extracts medical accounting data for each hospital where the patient received treatment.

[0025] When the acquisition unit 11 acquires information specifying a target for which the name of an injury or illness is to be estimated, the extraction unit 12 may extract time-series medical information from the medical accounting data corresponding to the specified target. For example, when the name of a medication is specified, the extraction unit 12 extracts time-series medical information from the medical accounting data of patients for whom the specified medication has been prescribed. Furthermore, for example, when a target patient is specified, the extraction unit 12 extracts time-series medical information from the medical accounting data of patients corresponding to the specified content. For example, when the medical department in which the patient is receiving treatment is specified, the extraction unit 12 extracts time-series medical information from the medical accounting data of patients receiving treatment in the specified medical department.

[0026] The estimation unit 13 uses an estimation model to estimate the name of the patient's illness or injury based on the time-series medical information extracted by the extraction unit 12. The name of the patient's illness or injury estimated by the estimation unit 13 using the estimation model is, for example, the name of the illness or injury that is the main cause of the symptoms the patient is experiencing. The estimation model is a machine learning model that estimates the name of the patient's illness or injury from the time-series medical information. The estimation unit 13 estimates the name of the patient's illness or injury by, for example, using the time-series medical information extracted by the extraction unit 12 as input to the estimation model and obtaining the name of the patient's illness or injury as output from the estimation model.

[0027] The estimation unit 13 estimates the name of the patient's illness or injury using, for example, an estimation model generated in a system external to the illness or injury name estimation device 10. The estimation unit 13 may estimate the name of the patient's illness or injury using an estimation model generated by the generation unit 14. A method for generating an estimation model will be described later.

[0028] The estimation unit 13, for example, converts each piece of time-series medical information into a feature vector. Then, the estimation unit 13 estimates the name of the patient's illness or injury using, for example, the feature vector converted from the time-series medical information as input to an estimation model. For example, Word2Vec can be used to convert the time-series medical information into a feature vector and to convert the output of the estimation model into the name of the illness or injury. For converting the time-series medical information into a feature vector and to convert the output of the estimation model into the name of the illness or injury, a language model other than Word2Vec may also be used.

[0029] The estimation unit 13 may estimate candidate names of the patient's injury or illness using an estimation model. For example, the estimation unit 13 estimates multiple names of injury or illness as candidate names of the patient's injury or illness using the estimation model. For example, the estimation unit 13 estimates names of injury or illness whose accuracy in estimation by the estimation model is equal to or higher than a standard as candidate names of the patient's injury or illness. The accuracy in estimation of the name of injury or illness is, for example, the probability of each name of injury or illness output by the estimation model. In other words, the accuracy in estimation of the name of injury or illness is the probability that each patient's injury or illness corresponds to each name of injury or illness when the estimation model classifies the name of the patient's injury or illness. The standard for estimating candidate names of injury or illness is set using, for example, a value of accuracy that, if the standard is exceeded, the name of injury or illness can be considered correct. Furthermore, the estimation unit 13 may estimate names of injury or illness whose accuracy in estimation by the estimation model is ranked from the top to a predetermined rank as candidate names of the patient's injury or illness. The predetermined ranking is set, for example, so that a person in charge who uses the estimation results of the candidate injury / illness names can verify the candidate injury / illness names and select a correct injury / illness name from the candidate injury / illness names. The estimation unit 13 may estimate, as the candidate injury / illness names of the patient, injury / illness names whose accuracy in estimation of the injury / illness name by the estimation model is equal to or exceeds a standard and whose accuracy in estimation of the injury / illness name is in the top to a predetermined ranking.

[0030] The estimation unit 13 may estimate the name of an illness or injury for each of a plurality of symptoms that the patient is experiencing. For example, the estimation unit 13 uses an estimation model to estimate the name of an illness or injury for each of a plurality of symptoms that the patient is experiencing. In this case, the estimation unit 13 estimates the name of the patient's illness or injury using, for example, an estimation model that estimates the names of a plurality of illnesses or injuries from time-series medical information included in the medical accounting data.

[0031] The estimation unit 13 may estimate the name of an injury or illness that has developed as a complication. For example, the estimation unit 13 uses an estimation model to estimate the name of an injury or illness that is the cause of a complication that a patient has developed and the name of an injury or illness that has developed as a complication. The estimation model estimates the name of an injury or illness that is the cause of a complication and the name of an injury or illness that has developed as a complication, for example, from time-series medical information included in the medical accounting data. For example, if a patient is prescribed a drug to control blood sugar levels and a drug to suppress arteriosclerosis simultaneously, there is a high possibility that the patient has developed arteriosclerosis as a complication of diabetes, along with diabetes. In this case, the estimation model estimates that arteriosclerosis has developed as a complication of diabetes, based on the time-series medical information included in the medical accounting data. Furthermore, the estimation unit 13 may estimate the name of an injury or illness that has developed as a complication based on multiple injury or illness names estimated by the estimation model. For example, the estimation unit 13 compares multiple injury or illness names included in the estimation result of the estimation model with a list including names of injuries or illnesses related to complications. The list including names of injuries or illnesses related to complications is, for example, a list that associates the names of injuries or illnesses that are the source of complications with the names of injuries or illnesses that have developed as complications. When multiple injury or illness names included in the estimation results of the estimation model are included in the list including names of injuries or illnesses related to complications, the estimation unit 13 estimates that the injury or illness names are the names of injuries or illnesses that are the source of complications and the names of injuries or illnesses that have developed as complications.

[0032] The estimation unit 13 may estimate the name of an illness or injury using the prescription history of multiple medications. For example, the estimation unit 13 estimates the name of an illness or injury of a patient by further using the prescription history of multiple medications as input to an estimation model. The prescription history of multiple medications may be included in chronological medical information. The estimation unit 13 may also estimate the name of an illness or injury of a patient by using the prescription history of multiple medications as input to an estimation model. In this case, for example, the estimation model estimates the name of an illness or injury using information indicating the prescription order of multiple medications as the prescription history. For example, if a patient who has been prescribed a medication to control blood glucose levels is newly prescribed a kidney medication, the patient may have developed kidney damage as a complication of diabetes. In such a case, the estimation model estimates that the patient has developed diabetes. In such a case, the estimation model may also estimate that the patient has developed kidney dysfunction as a complication of diabetes. The prescription history may also include the prescribed amount of each medication.

[0033] The estimation unit 13 may further use the symptom description to estimate the name of the patient's illness or injury. For example, the estimation unit 13 further uses the symptom description as an input to an estimation model to estimate the name of the patient's illness or injury. The estimation unit 13 may also estimate the name of the patient's illness or injury using the symptom description as chronological medical information. For example, the estimation unit 13 estimates the name of the patient's illness or injury using the symptom description written by a doctor for each patient for each month as chronological medical information. For example, the estimation unit 13 estimates the name of the patient's illness or injury using the chronological symptom description as an input to an estimation model. The symptom description is, for example, an explanation of the content of a medical procedure. The symptom description may include, for example, an explanation of the legitimacy of a medical procedure in a medical fee claim. For example, the symptom description may include a description of why the medical procedure or the materials used were medically necessary. The symptom description is, for example, created by a doctor. The symptom description is included, for example, in medical receipt data. The symptom description may also be included in medical accounting data.

[0034] The estimation unit 13 may further use the patient's test data to estimate the name of the illness or injury. For example, the estimation unit 13 further uses the test data as input to an estimation model to estimate the name of the patient's illness or injury. The estimation unit 13 may also estimate the name of the patient's illness or injury using time-series test data as time-series medical information. For example, the estimation unit 13 estimates the name of the patient's illness or injury using the time-series test data as input to the estimation model. For example, if a patient who has been prescribed an antipyretic analgesic is undergoing an antigen test for the novel coronavirus, the patient is likely to have a viral infection. In such a case, the estimation model estimates that the name of the illness or injury for which the antipyretic analgesic is prescribed is a viral infection.

[0035] The estimation model may estimate the name of the illness or injury using the amount of medical expenses of the patient included in the medical accounting data. For example, the estimation unit 13 estimates the name of the illness or injury of the patient by further using the amount of medical expenses of the patient as an input to the estimation model. The amount of medical expenses of the patient may be time-series data of the amount of medical expenses. For example, the estimation unit 13 estimates the name of the illness or injury of the patient using the time-series amount of medical expenses as time-series medical treatment information. The estimation unit 13 estimates the name of the illness or injury of the patient using the time-series amount of medical expenses as an input to the estimation model. Furthermore, the amount of medical expenses may be insurance points used to calculate medical fees. The insurance points may be time-series data of insurance points. For example, the cost required for treatment may vary greatly depending on the illness or injury. Therefore, using the amount of medical expenses to estimate the name of the illness or injury can improve the accuracy of the estimation of the name of the illness or injury.

[0036] The estimation unit 13 may estimate the name of the patient's illness or injury by comparing the time-series medical information with the list. For example, the estimation unit 13 compares a list in which illness or injury names are associated with the order of medical procedures with the time-series medical information extracted by the extraction unit 12. Then, the estimation unit 13 estimates, as the name of the patient's illness or injury, the name of the illness or injury associated with the order of medical procedures that matches the order of medical procedures in the time-series medical information.

[0037] When generating an estimation model in the injury / illness name estimation device 10, the generation unit 14 generates, for example, the estimation model. The generation unit 14 generates the estimation model by learning, for example, the relationship between the time-series medical information included in the medical accounting data and the actual injury / illness name of the patient. The generation unit 14 may generate the estimation model using the injury / illness name identified from the detailed symptom description as a label.

[0038] The generation unit 14 converts, for example, each piece of time-series medical information and the name of the injury or illness into a feature vector. Then, the generation unit 14 generates an estimation model using, for example, the feature vector converted from the time-series medical information and the feature vector converted from the name of the injury or illness as training data. That is, the estimation model is generated by learning the relationship between the feature vector converted from the time-series medical information and the feature vector converted from the name of the injury or illness through machine learning. For example, Word2Vec can be used for the conversion to the feature vector. A language model other than Word2Vec may also be used for the conversion to the feature vector. The estimation model is generated, for example, by deep learning using a recurrent neural network. The learning algorithm for generating the estimation model is not limited to the above. The estimation model may also be generated by a system external to the injury or illness name estimation device 10.

[0039] When the estimation model estimates the name of a complication, the generation unit 14 converts, for example, each piece of time-series medical information and the name of the injury or illness that is the source of the complication and the name of the injury or illness that has developed as a complication into feature vectors. Then, the generation unit 14 generates an estimation model that estimates the name of the injury or illness that has developed as a complication, using, for example, the feature vectors converted from each piece of time-series medical information and the feature vectors converted from the name of the injury or illness that is the source of the complication and the name of the injury or illness that has developed as a complication as training data. That is, in this case, the estimation model is generated by, for example, learning, through machine learning, the relationship between the feature vectors converted from each piece of time-series medical information and the feature vectors converted from the name of the injury or illness that is the source of the complication and the name of the injury or illness that has developed as a complication.

[0040] When the estimation model estimates the name of an injury or illness using multiple drug prescription histories, the generation unit 14 generates the estimation model using training data including, for example, a feature vector converted from medical information including multiple drug prescription histories. When the estimation model estimates the name of an injury or illness using additional patient test data, the generation unit 14 converts, for example, each piece of test data into a feature vector. The generation unit 14 then generates the estimation model using, for example, feature vectors converted from each piece of time-series medical information, feature vectors converted from the test data, and feature vectors converted from the name of the injury or illness, as training data. When the estimation model estimates the name of an injury or illness using the amount of medical expenses of a patient, the generation unit 14 converts, for example, the amount of medical expenses of the patient into a feature vector. The generation unit 14 then generates the estimation model using, for example, feature vectors converted from the amount of medical expenses of the patient, feature vectors converted from the test data, and feature vectors converted from the name of the injury or illness, as training data.

[0041] The output unit 15 outputs information regarding the name of the illness or injury estimated by the estimation unit 13. That is, the output unit 15 outputs an estimation result regarding the name of the illness or injury estimated by the estimation unit 13 using the estimation model. For example, the output unit 15 outputs the name of the illness or injury of each patient estimated by the estimation unit 13. The output unit 15 may output, for example, the name of the illness or injury of each patient estimated by the estimation unit 13 along with medical information of the patient. The output unit 15 may also output the accuracy of the estimation along with the name of the illness or injury of each patient estimated by the estimation unit 13. The output unit 15 may also output multiple names of the illness or injury as candidate names of the illness or injury based on the accuracy of the name of the illness or injury of each patient estimated by the estimation unit 13. The output unit 15 may output, for each medication prescribed to a patient, the name of the illness or injury estimated by the estimation unit 13 for each patient to whom the medication is prescribed. The output unit 15 may output the number of patients for each name of the illness or injury for each patient to whom the medication is prescribed based on the estimation result of the estimation unit 13. The output unit 15 may output the percentage of the number of patients for each illness or injury for which the medicine was prescribed, based on the estimation result of the estimation unit 13. The output unit 15 outputs information about the illness or injury estimated by the estimation unit 13 to the terminal device 40, for example.

[0042] When the estimation model estimates the name of an injury or illness that has developed as a complication, the output unit 15 outputs, for example, information regarding the name of the injury or illness that has developed as a complication. The output unit 15 may also output the name of the injury or illness that is the source of the complication and the name of the injury or illness that has developed as a complication. The output unit 15 may also output multiple names of injuries or illnesses estimated by the estimation unit 13 using the estimation model.

[0043] For example, when the name of an injury or illness included in the medical accounting data differs from the name of an injury or illness estimated by the estimation unit 13, the output unit 15 outputs information indicating that the names of the injury or illness differ. For example, for a patient for whom the name of an injury or illness included in the medical accounting data differs from the name of an injury or illness estimated by the estimation unit 13, the output unit 15 outputs the name of an injury or illness included in the medical accounting data and the name of an injury or illness estimated by the estimation unit 13. Furthermore, when the name of an injury or illness included in the medical accounting data differs from the name of an injury or illness estimated by the estimation unit 13, the output unit 15 may output, for each name of an injury or illness included in the medical accounting data, the proportion of the number of patients with each name of an injury or illness estimated by the estimation unit 13. When the name of an injury or illness included in the medical accounting data differs from the name of an injury or illness estimated by the estimation unit 13, the output unit 15 may output, for each medication, the name of an injury or illness estimated by the estimation unit 13 for each patient to whom the medication was prescribed.

[0044] The memory unit 16 stores, for example, data related to the estimation of the name of an injury or illness. The memory unit 16 stores, for example, medical accounting data acquired by the acquisition unit 11. The memory unit 16 also stores, for example, time-series medical information extracted by the extraction unit 12. The memory unit 16 also stores, for example, the name of a patient's injury or illness estimated by the estimation unit 13. The memory unit 16 stores, for example, an extraction model. The memory unit 16 also stores, for example, an estimation model. The extraction model and the estimation model may each be stored in a storage means other than the memory unit 16.

[0045] FIG. 3 shows an example of the estimated result of the disease name output by the output unit 15. The example of the estimated result of the disease name in FIG. 3 shows the estimated result of the disease name for a patient who has been prescribed a compress containing an anti-inflammatory analgesic. In the example of the estimated result in FIG. 3, the "patient name" and the "disease name" are associated with each other. The "patient name" is, for example, the patient's name. The "patient name" may also be the patient's identification number. The "disease name" is the name of the disease or injury of each patient estimated by the estimation unit 13. By referring to the estimated result of the disease or injury as shown in the example of the estimated result of the disease or injury in FIG. 3, the person in charge of investigating the disease or injury name can easily identify the disease or injury that is the main cause of the symptoms the patient is experiencing.

[0046] FIG. 4 is an example of an estimation result of the proportion of illnesses and injuries for which a medicine is prescribed, output by the output unit 15. The example estimation result in FIG. 4 shows the proportion of illnesses and injuries for which a compress containing an anti-inflammatory analgesic was prescribed. In the example estimation result in FIG. 4, "name of illness and injury" and "proportion (%)" are associated with each other. In the example of FIG. 4, "name of illness and injury" is the name of the illness and injury for which a medicine is prescribed. Furthermore, "proportion (%)" indicates the proportion, expressed as a percentage, of the number of patients with each illness and injury to the total number of patients for whom a medicine was prescribed.

[0047] FIG. 5 illustrates an example of an estimation result output by the output unit 15 when multiple candidates for the name of an injury or illness are estimated. FIG. 5 illustrates an example of an estimation result when the estimation model estimates multiple candidates for the name of an injury or illness. In the example estimation result of FIG. 5, "Patient Name" is associated with "Candidate 1," "Candidate 2," and "Candidate 3." "Patient Name" is, for example, the patient's name. "Patient Name" may also be the patient's identification number. "Candidate 1," "Candidate 2," and "Candidate 3" are each candidate names for the name of an injury or illness. For example, this indicates that the name of the injury or illness suffered by patient "H" is likely to be "COVID-19 infection," "influenza," or "fever (cold symptoms)." The output unit 15 outputs, for example, the name of an injury or illness estimated by the estimation model with a standard or higher accuracy as the estimation result for the name of the injury or illness suffered by the patient. In the example estimation result of FIG. 5, the output unit 15 may further output the accuracy of each candidate estimation by the estimation model. 5, "-" indicates that the number of illnesses for which the accuracy of the estimation model was higher than the standard was less than three. For example, for patient "H," the only illnesses for which the accuracy of the estimation model was higher than the standard were "influenza" and "novel coronavirus infection," so candidate 3 is indicated by "-."

[0048] FIG. 6 illustrates an example of an estimation result output by the output unit 15 when multiple illnesses that a patient is suffering from are estimated. FIG. 6 illustrates an example of the result of estimating the illness name for each symptom when a patient has multiple symptoms. In the example of the estimation result in FIG. 6, a "patient name" is associated with "Illness 1" and "Illness 2." The "patient name" is, for example, the patient's name. The "patient name" may also be the patient's identification number. In the example of the estimation result in FIG. 6, "Illness 1" and "Illness 2" are illnesses that the patient may be suffering from. In the example of the estimation result in FIG. 6, for example, "Illness 2" is a complication of "Illness 1." Although the example of the estimation result in FIG. 6 illustrates estimation results for two illnesses, estimation results for three or more illnesses may also be illustrated. In this case, the estimation unit 13 estimates three or more illnesses using an estimation model.

[0049] The medical accounting device 20 is, for example, a device that generates medical receipts from a patient's medical information. The medical accounting device 20 may calculate medical expenses to be billed to a patient based on the patient's medical information. The medical accounting device 20, for example, acquires the patient's medical information from the medical data management device 30. The medical accounting device 20 then generates a medical receipt based on the patient's medical information. The medical accounting device 20 may also generate prescription data based on the patient's medical information. The functions of the medical accounting device 20 are not limited to those described above.

[0050] The medical accounting device 20 stores, for example, medical accounting data. The medical accounting device 20 stores, for example, medical treatment information about patients. The medical accounting device 20 may also store medical expense accounting data, receipts, and prescription data. The medical accounting device 20 outputs the stored data to, for example, the acquisition unit 11 of the injury / illness name estimation device 10.

[0051] The medical data management device 30 is, for example, a device that stores patient medical information. The patient medical information includes electronic medical records and test data. The information in the electronic medical records is input by, for example, doctors, nurses, laboratory technicians, physical therapists, and counselors. The medical data management device 30 may also store information on one or more items of the patient's injury / illness information, complication information, biomarkers, disease status, guideline score, medical history, effects, and test results, in addition to the information written in the electronic medical records.

[0052] The terminal device 40 is, for example, a terminal device used by a person who investigates the name of the patient's injury or illness. The terminal device 40 acquires information about the patient's injury or illness, for example, from the output unit 15 of the injury or illness name estimation device 10. The terminal device 40 then outputs the information about the patient's injury or illness to, for example, a display device (not shown).

[0053] The person who investigates the name of a patient's illness or injury is, for example, a person in charge of investigating the use of medicine at a pharmaceutical company, a pharmacy, or a pharmaceutical wholesaler. The person who investigates the name of a patient's illness or injury may be a person in charge of checking the details of medical fees. The person in charge of checking the details of medical fees may be, for example, a person from a health insurance association. The person who investigates the name of a patient's illness or injury may also be a medical professional. The medical professional is, for example, a doctor, nurse, or pharmacist. The medical professional may also be a person who performs audits within a hospital. Medical professionals are not limited to the above.

[0054] The terminal device 40 may acquire information specifying a target whose injury or illness name is to be estimated. The information specifying a target whose injury or illness name is to be estimated is, for example, input to the terminal device 40 by a person investigating the patient's injury or illness name. The terminal device 40 acquires, for example, information specifying a medicine whose injury or illness name is to be estimated. The terminal device 40 may also acquire information specifying a patient whose injury or illness name is to be estimated. The terminal device 40 outputs, for example, information specifying a target whose injury or illness name is to be estimated to the acquisition unit 11 of the injury or illness name estimation device 10.

[0055] An example of a process for estimating the name of an injury or illness will be described below. Fig. 7 is a diagram showing an example of an operation flow of a process for estimating the name of an injury or illness of each of a plurality of patients in the injury or illness name estimation device 10.

[0056] The acquisition unit 11 acquires medical accounting data (step S11). The acquisition unit 11 acquires the medical accounting data from the medical accounting device 20, for example.

[0057] When the medical accounting data is acquired, the extraction unit 12 extracts time-series medical information for each patient from the medical accounting data (step S12).

[0058] When the time-series medical information is extracted from the medical accounting data, the estimation unit 13 estimates the name of the illness or injury of each patient based on the acquired time-series medical information of each patient using an estimation model (step S13). The estimation model is a machine learning model that estimates the name of the illness or injury of a patient from the time-series medical information.

[0059] When the disease names have been estimated for all target patients (Yes in step S14), the output unit 15 outputs information related to the estimated disease names (step S15).

[0060] In step S14, if there are any patients for whom the estimation of the illness or injury name has not been completed (No in step S14), the process returns to step S12, and the extraction unit 12 extracts chronological medical information from the medical accounting data for each patient for whom the extraction of chronological medical information has not been completed.

[0061] The disease name estimation device 10 extracts time-series medical information from medical accounting data. Furthermore, the estimation unit 13 of the disease name estimation device 10 uses an estimation model to estimate the name of each patient's disease or injury based on the acquired time-series medical information for each patient. The estimation model is a machine learning model that estimates the name of a patient's disease or injury from the time-series medical information. The disease name estimation device 10 then outputs information regarding the estimated name of each patient's disease or injury. By estimating the name of a patient's disease or injury based on the time-series medical information for each patient in this manner, the disease name estimation device 10 can estimate the name of a patient's disease or injury based on, for example, the relationship between actions performed in medical care, thereby improving the accuracy of estimating the name of a patient's disease or injury. Furthermore, the disease name estimation device 10 can support decision-making based on the disease or injury that is the main cause of the patient's symptoms, for example, by estimating the name of a patient's disease or injury from medical accounting data.

[0062] Furthermore, the injury / illness name estimation device 10 can accurately estimate the name of an injury or illness without directly using information from an electronic medical record by, for example, estimating the name of an injury or illness using medical information extracted from medical accounting data. Therefore, by using the injury / illness name estimation device 10, it is possible to accurately estimate the name of an injury or illness while protecting the patient's personal information, for example.

[0063] Furthermore, for example, when estimating the name of an illness or injury of a patient to whom a medication has been prescribed, the illness name estimation device 10 can accurately estimate the name of the illness or injury that is the main cause of the patient's symptoms by estimating the name of the illness or injury based on time-series medical information. Furthermore, the illness name estimation device 10 can easily understand the purpose of the medication by, for example, outputting the proportion of each illness or injury that is the target of the medication prescription as an estimation result. Furthermore, by estimating multiple illnesses and injuries that a patient is suffering from based on time-series medical data, the illness name estimation device 10 can estimate, for example, the name of an illness or injury that has developed as a complication.

[0064] Each process in the injury / illness name estimation device 10 may be distributed and executed in multiple information processing devices connected via a network. For example, the processes in the extraction unit 12 and the estimation unit 13 and the generation unit 14 may be performed in different information processing devices. Also, for example, the processes in the extraction unit 12 and the estimation unit 13 may be performed in different information processing devices. Which information processing device performs each process in the injury / illness name estimation device 10 can be set as appropriate.

[0065] Each process in the injury / illness name estimation device 10 can be realized by executing a computer program on a computer. Fig. 8 shows an example of the configuration of a computer 100 that executes a computer program that performs each process in the injury / illness name estimation device 10. The computer 100 includes a CPU (Central Processing Unit) 101, a memory 102, a storage device 103, an input / output I / F (Interface) 104, and a communication I / F 105.

[0066] The CPU 101 reads and executes computer programs for each process from the storage device 103. The CPU 101 may be configured with a combination of multiple CPUs. The CPU 101 may also be configured with a combination of a CPU and another type of processor. For example, the CPU 101 may be configured with a combination of a CPU and a graphics processing unit (GPU). The memory 102 is configured with a dynamic random access memory (DRAM) or the like, and temporarily stores the computer programs executed by the CPU 101 and data being processed. The storage device 103 stores the computer programs executed by the CPU 101. The storage device 103 is configured with, for example, a non-volatile semiconductor storage device. Other storage devices such as a hard disk drive may also be used for the storage device 103. The input / output I / F 104 is an interface that accepts input from an operator and outputs display data, etc. The communication I / F 105 is an interface that transmits and receives data to and from other information processing devices. The medical accounting device 20, the medical data management device 30, and the terminal device 40 may also have a configuration similar to that of the computer 100.

[0067] The computer program used to execute each process can also be stored and distributed on a computer-readable recording medium that non-temporarily stores data. Examples of recording media that can be used include magnetic tapes for recording data and magnetic disks such as hard disks. Optical disks such as CD-ROMs (Compact Disc Read Only Memory) can also be used as recording media. Non-volatile semiconductor storage devices can also be used as recording media.

[0068] Some or all of the above-described embodiments can be described as, but are not limited to, the following supplementary notes.

[0069] [Supplementary Note 1] An injury or illness name estimation device comprising: an acquisition means for acquiring medical accounting data; an extraction means for extracting time-series medical information for each patient from the medical accounting data; an estimation means for estimating the name of the injury or illness for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the injury or illness of a patient from the time-series medical information; and an output means for outputting information related to the estimated name of the injury or illness for each patient.

[0070] [Supplementary Note 2] The injury / illness name estimation device described in Supplementary Note 1, wherein the estimation means estimates the name of an injury or illness that has developed as a complication using the estimation model that estimates multiple injury / illness names of a patient from time-series medical information, and the output means outputs information regarding the name of the injury or illness that has developed as a complication.

[0071] [Appendix 3] The injury / illness name estimation device described in Appendix 1 or 2, wherein the output means outputs information indicating that the injury / illness name is different when the injury / illness name included in the medical accounting data differs from the injury / illness name estimated by the estimation means.

[0072] [Supplementary Note 4] The injury or illness name estimating device according to any one of Supplementary Notes 1 to 3, wherein the estimating means estimates the injury or illness name using a drug prescription history as the time-series medical information.

[0073] [Supplementary Note 5] The injury or illness name estimation device described in any one of Supplementary Notes 1 to 4, wherein the estimation means estimates the name of an injury or illness for which a drug is prescribed using the estimation model, and the output means outputs, for each drug, information regarding the name of an injury or illness of a patient to whom the drug has been prescribed.

[0074] [Supplementary Note 6] The injury or illness name estimating device according to any one of Supplementary Notes 1 to 5, wherein the estimating means further uses detailed symptom descriptions to estimate the injury or illness name.

[0075] [Supplementary Note 7] The injury or illness name estimating device according to any one of Supplementary Notes 1 to 6, wherein the estimating means further uses the amount of medical expenses of the patient included in the medical accounting data to estimate the injury or illness name.

[0076] [Supplementary Note 8] The injury or illness name estimating device according to any one of Supplementary Notes 1 to 7, wherein the estimating means further uses examination data of the patient to estimate the injury or illness name.

[0077] [Supplementary Note 9] The injury or illness name estimation device according to any one of Supplementary Notes 1 to 8, further comprising: a generation means for generating the estimation model by learning the relationship between time-series medical information and the patient's actual injury or illness name through machine learning.

[0078] [Supplementary Note 10] The injury or illness name estimation device according to Supplementary Note 9, wherein the generation means generates the estimation model using a name of an injury or illness identified from a detailed symptom description as an actual name of the patient's injury or illness.

[0079] [Supplementary Note 11] A method for estimating the name of an injury or illness, comprising: acquiring medical accounting data; extracting time-series medical information for each patient from the medical accounting data; estimating the name of an injury or illness for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of a patient's injury or illness from the time-series medical information; and outputting information relating to the estimated name of an injury or illness for each patient.

[0080] [Supplementary Note 12] A recording medium that non-temporarily records an injury or illness name estimation program that causes a computer to execute the following processes: a process of acquiring medical accounting data; a process of extracting time-series medical information for each patient from the medical accounting data; a process of estimating the name of the injury or illness for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the patient's injury or illness from the time-series medical information; and a process of outputting information related to the estimated name of the injury or illness for each patient.

[0081] Furthermore, some or all of the configurations described in Supplementary Notes 2 to 10, which are dependent on Supplementary Note 1, may also be dependent on Supplementary Notes 11 and 12 in the same dependent relationship as Supplementary Notes 2 to 10. Furthermore, not limited to Supplementary Notes 1, 11, and 12, some or all of the configurations described as Supplements may be made dependent on various hardware, software, various recording means for recording software, or systems, within the scope of each of the above-mentioned embodiments.

[0082] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate.

[0083] REFERENCE SIGNS LIST 10 Injury / disease name estimation device 11 Acquisition unit 12 Extraction unit 13 Estimation unit 14 Generation unit 15 Output unit 16 Storage unit 20 Medical accounting device 30 Medical data management device 40 Terminal device 100 Computer 101 CPU 102 Memory 103 Storage device 104 Input / output I / F 105 Communication I / F

Claims

1. An injury or illness name estimation device comprising: an acquisition means for acquiring medical accounting data; an extraction means for extracting time-series medical information for each patient from the medical accounting data; an estimation means for estimating the name of the injury or illness for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the injury or illness for a patient from the time-series medical information; and an output means for outputting information related to the estimated name of the injury or illness.

2. The injury / illness name estimation device according to claim 1, wherein the estimation means estimates the name of an injury or illness that has developed as a complication using the estimation model that estimates multiple injury / illness names of a patient from chronological medical information, and the output means outputs information regarding the name of the injury or illness that has developed as a complication.

3. The injury or illness name estimation device described in claim 1 or 2, wherein the output means outputs information indicating that the injury or illness name is different when the injury or illness name contained in the medical accounting data differs from the injury or illness name estimated by the estimation means.

4. An injury or illness name estimating device according to any one of claims 1 to 3, wherein the estimating means estimates the injury or illness name using a drug prescription history as the time-series medical information.

5. An injury or illness name estimation device as described in any one of claims 1 to 4, wherein the estimation means uses the estimation model to estimate the name of the injury or illness for which a drug is prescribed, and the output means outputs, for each drug, information regarding the name of the injury or illness of the patient to whom the drug has been prescribed.

6. An injury or illness name estimating device according to any one of claims 1 to 5, wherein the estimating means further uses detailed symptom descriptions to estimate the injury or illness name.

7. An injury or illness name estimating device according to any one of claims 1 to 6, wherein the estimating means further uses the amount of medical expenses of the patient contained in the medical accounting data to estimate the injury or illness name.

8. An injury or illness name estimating device according to any one of claims 1 to 7, wherein the estimating means further uses test data of the patient to estimate the injury or illness name.

9. An injury or illness name estimation device according to any one of claims 1 to 8, further comprising: a generation means for generating the estimation model by learning the relationship between chronological medical information and the actual illness or injury name of the patient through machine learning.

10. The injury or illness name estimation device according to claim 9, wherein the generation means generates the estimation model using the name of the injury or illness identified from the detailed symptom description as the actual name of the patient's injury or illness.

11. A method for estimating the name of an injury or illness, comprising: acquiring medical accounting data; extracting time-series medical information for each patient from the medical accounting data; estimating the name of an injury or illness for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of a patient's injury or illness from the time-series medical information; and outputting information related to the estimated name of the injury or illness.

12. A method for estimating the name of an injury or illness as described in claim 11, which uses the estimation model to estimate the names of multiple injury or illness names of a patient from time-series medical information, estimates the name of an injury or illness that has developed as a complication, and outputs information regarding the name of the injury or illness that has developed as a complication.

13. A method for estimating the name of an injury or illness as described in claim 11 or 12, which outputs information indicating that the name of the injury or illness is different when the name of the injury or illness contained in the medical accounting data differs from the estimated name of the injury or illness.

14. A method for estimating the name of an injury or illness according to any one of claims 11 to 13, wherein the estimation model estimates the name of the injury or illness using a drug prescription history as the time-series medical information.

15. A method for estimating the name of an injury or illness described in any one of claims 11 to 14, which uses the estimation model to estimate the name of an injury or illness for which a drug is prescribed, and outputs, for each drug, information regarding the name of the injury or illness of a patient to whom the drug has been prescribed.

16. A method for estimating the name of an injury or illness according to any one of claims 11 to 15, wherein the estimation model further uses detailed symptom descriptions to estimate the name of the injury or illness.

17. A method for estimating the name of an injury or illness described in any one of claims 11 to 16, wherein the estimation model further uses the amount of medical expenses of the patient included in the medical accounting data to estimate the name of the injury or illness.

18. A method for estimating the name of an injury or illness according to any one of claims 11 to 17, wherein the estimation model further uses test data of the patient to estimate the name of the injury or illness.

19. A method for estimating the name of an injury or illness described in any one of claims 11 to 18, wherein the estimation model is generated by learning the relationship between chronological medical information and the patient's actual illness or injury name through machine learning.

20. A recording medium that non-temporarily records an illness or injury name estimation program that causes a computer to execute the following processes: a process of acquiring medical accounting data; a process of extracting time-series medical information for each patient from the medical accounting data; a process of estimating the name of the illness or injury for each patient based on the acquired time-series medical information for each patient using an estimation model that estimates the name of the patient's illness or injury from the time-series medical information; and a process of outputting information related to the estimated name of the illness or injury.

Images