JP7299047B2 - LEARNING MODEL GENERATION METHOD, COMPUTER PROGRAM AND INFORMATION PROCESSING DEVICE - Google Patents

Description

The present invention relates to a learning model generation method, a computer program, and an information processing apparatus.

In medical facilities such as hospitals, patients are examined and diagnosed based on the results of the examination. Conventionally, techniques for performing diagnosis based on test results using information processing devices have been developed. Patent Literature 1 discloses a technique for diagnosis using a neural network.

JP 2007-52774 A

In medical facilities, a simple test is first performed, and a detailed test may be performed depending on the test results. For example, when a simple test reveals that a patient may be suffering from a specific disease, a detailed test determines the presence or absence of the disease. A simple test yields results within a short period of time, for example, within several tens of minutes to several hours. On the other hand, it takes a long time, for example, several days to one month, until the result of the detailed examination is obtained. A definitive diagnosis is withheld pending further work-up results. However, the situation could deteriorate while the diagnosis is pending. If the disease is an infectious disease, the infection may spread while the diagnosis is pending. Therefore, it is necessary to make a prompt diagnosis.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide a learning model generation method, a computer program, and an information processing apparatus that enable prompt diagnosis.

A learning model generation method according to the present invention is executed by a computer using a first test result of a first test performed on a patient at a medical facility and a specimen collected within a predetermined area including the medical facility. Regional information including a second test result of a second test that requires a longer test period than the first test, patient information about the patient, and treatment information representing the content of treatment for the patient or the second test result about the patient A plurality of training data containing is acquired, and a computer, based on the acquired training data, inputs the first test result, the regional information, and the patient information, and treatment information representing the details of the recommended treatment for the patient, or It is characterized by generating a learning model that outputs an examination prediction that predicts a second examination result for the patient.

In the present invention, when a first test result for a patient, region information including a second test result of a second test requiring a long test period related to an area including a medical facility, and patient information are input, a recommended recommendation for the patient is generated. Based on the teacher data, a learning model is generated that outputs treatment information representing the details of the treatment to be performed or test prediction that predicts the second test result regarding the patient. Based on the first test results, patient information and regional information, a learning model can be used to obtain treatment information or test predictions.

In the learning model generation method according to the present invention, the regional information is the number of occurrences of a specific disease within the predetermined region, the number of occurrences of the specific disease within the predetermined region by patient age or gender, or It is characterized by including a time-series change in occurrence of the specific disease within the predetermined area.

In one embodiment of the present invention, the regional information includes the number of disease occurrences in a predetermined region, the number of disease occurrences by patient age or sex, or time-series changes in the occurrence of specific diseases. A second test result representing the patient's condition is appropriately predicted according to the occurrence of the disease in the area.

The learning model generation method according to the present invention is characterized in that the specific disease is an infectious disease.

In one aspect of the invention, the specified disease is an infectious disease. The regional information includes the outbreak status of infectious diseases within a predetermined region. There is a high correlation between the occurrence of infectious diseases and the possibility that the patient is suffering from infectious diseases, and it is possible to obtain test predictions according to regional information.

The learning model generation method according to the present invention is characterized in that the first examination includes microscopic examination, gene amplification examination, or immunological examination of a sample collected from the patient.

In one aspect of the invention, the first test includes a microscopic test, gene amplification test, or immunological test for the sample. These first tests provide some insight into the patient's condition in a short period of time.

The learning model generation method according to the present invention is characterized in that the second inspection includes an inspection for isolating and culturing pathogens.

In one aspect of the invention, the second test includes a test for isolating and culturing pathogens. Although requiring a longer examination period, this second examination provides more insight into the patient's condition.

The learning model generation method according to the present invention is characterized in that the patient information includes age, sex, symptoms, examination history, medication history, treatment history, or action history.

In one form of the invention, the patient information includes the patient's age, gender, symptoms, examination history, medication history, treatment history, or behavior history. A second test result representing the patient's condition is appropriately predicted according to this patient information.

In the learning model generation method according to the present invention, after the treatment for the patient is completed and the second test result regarding the patient is obtained, the computer performs the first test result regarding the patient, the regional information, and the Re-learning data including patient information and the second test result is acquired, and the computer re-learns the learning model using the re-learning data.

In one aspect of the present invention, after the treatment for the patient is completed and the second test result is obtained, the relearning data including the first test result, the regional information, the patient information and the second test result are used. , to retrain the learning model. Retraining updates the learning model so that it can more accurately predict the second test result.

In the learning model generation method according to the present invention, after the treatment for the patient is completed and the second test result regarding the patient is obtained, the computer performs the first test result regarding the patient, the regional information, and the Acquiring relearning data including patient information and treatment information representing details of treatment that should have been performed on the patient in accordance with the second test result, and causing a computer to use the relearning data to perform the learning model is re-learned.

In one aspect of the present invention, after the treatment for the patient is completed and the second test result is obtained, the patient should be treated according to the first test result, the regional information, the patient information, and the second test result. The learning model is re-learned using the re-learning data including the treatment information representing the contents of the treatment that was performed. Re-learning updates the learning model so that more appropriate processing information is obtained.

A computer program according to the present invention acquires a first test result of a first test performed on a patient at a medical facility, and is executed using a specimen collected within a predetermined area including the medical facility. Acquiring regional information including a second test result of a second test that requires a longer test period than the first test, obtaining patient information about the patient, and inputting the first test result, the regional information, and the patient information inputting the obtained first test result, the regional information and the patient information to a learning model that outputs treatment information representing the content of recommended treatment for the patient or test prediction that predicts the second test result related to the patient It is characterized by causing a computer to execute processing for outputting treatment information or examination prediction.

In the present invention, based on the first test result of the patient, patient information, and regional information, a learning model is used to predict the second test result, or treatment information representing the content of the recommended treatment for the patient. is output. Before the results of the second test are obtained, the test prediction or treatment information can be obtained to quickly make an appropriate diagnosis for the patient.

The computer program according to the present invention refers to a storage unit storing treatment information corresponding to a second examination result and outputs treatment information corresponding to the examination prediction when the learning model outputs the examination prediction. It is characterized in that the processing is further executed by a computer.

In one aspect of the present invention, the treatment information corresponding to the second test result is stored in the storage unit, and the treatment information is acquired from the storage unit according to the second test result obtained using the learning model. be. Once the second test result is obtained, the content of the recommended treatment for the patient is determined.

A computer program according to the present invention performs processing for outputting information for displaying an image including an obtained first examination result, examination prediction output by the learning model, and treatment information according to the examination prediction. Furthermore, it is characterized in that it is executed by a computer.

In one form of the invention, an image is displayed that includes a first exam prediction, an exam prediction, and treatment information. An operator such as a doctor confirms the content of the image, diagnoses the patient with reference to the examination prediction, and determines the content of treatment for the patient with reference to the treatment information.

The computer program according to the present invention includes the acquired patient information, a second test result group of a second test performed using a sample collected in the predetermined area, and the area information including a map showing the predetermined area. The computer is further caused to execute a process of outputting information for displaying an image including and.

In one form of the invention, an image is displayed that includes patient information and regional information. An operator such as a doctor can confirm the contents of the image and judge the validity of the examination prediction or treatment information based on the regional information and the patient information.

An information processing apparatus according to the present invention includes: a first acquisition unit that acquires a first test result of a first test performed on a patient at a medical facility; a second acquisition unit that acquires region information including a second test result of a second test that is performed using a second test that requires a longer test period than the first test, and a third acquisition unit that obtains patient information about the patient; , a learning model that, when inputting a first test result, the region information, and the patient information, outputs treatment information representing details of recommended treatment for the patient or test prediction that predicts a second test result for the patient; and an output unit that inputs the obtained first test result, the regional information, and the patient information to the learning model and outputs treatment information or test prediction.

In the present invention, based on the first test result of the patient, patient information, and regional information, a learning model is used to predict the second test result, or treatment information representing the content of the recommended treatment for the patient. is output. Before the results of the second test are obtained, the test prediction or treatment information can be obtained to quickly make an appropriate diagnosis for the patient.

The present invention has excellent effects, such as obtaining predictions of test results before obtaining test results that require a long test period, and making it possible to quickly make an appropriate diagnosis for a patient. .

It is a schematic diagram which shows the structure of a diagnostic system. 3 is a block diagram showing a functional configuration example of an input/output device; FIG. FIG. 4 is a conceptual diagram showing an example of contents of patient information; 3 is a block diagram showing a functional configuration example of a storage device; FIG. FIG. 3 is a conceptual diagram showing an example of contents of area information; 2 is a block diagram showing a functional configuration example of the information processing apparatus according to the first embodiment; FIG. 4 is a conceptual diagram showing an example of content of teacher data according to the first embodiment; FIG. FIG. 4 is a conceptual diagram showing an example of contents of a treatment database; FIG. 4 is a conceptual diagram showing a functional configuration example of a learning model; 4 is a flow chart showing a procedure of processing for learning a learning model executed by an information processing apparatus; 4 is a flowchart showing a procedure of diagnostic processing performed by an information processing apparatus; 4 is a schematic diagram showing an example of an image including a first examination result, examination prediction, and treatment information according to Embodiment 1. FIG. FIG. 4 is a schematic diagram showing an example of an image including patient information, a second test result group, and area information including a map showing a predetermined area. 4 is a flow chart showing a procedure of re-learning processing performed by the information processing apparatus; FIG. 10 is a block diagram showing an example functional configuration of an information processing apparatus according to a second embodiment; FIG. 10 is a conceptual diagram showing an example of contents of teacher data according to Embodiment 2; FIG. 10 is a schematic diagram showing an example of an image including a first examination result, examination prediction, and treatment information according to Embodiment 2;

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be specifically described below with reference to the drawings showing its embodiments.
<Embodiment 1>
FIG. 1 is a schematic diagram showing the configuration of a diagnostic system 100. As shown in FIG. The diagnosis system 100 is a system for diagnosing the details of treatment to be given to a patient 42 who has been examined at a medical facility 4 such as a hospital or a clinic, using a learning model. The diagnostic system 100 executes a learning model generation method. In Embodiment 1, an example of diagnosing an MRSA (methicillin-resistant Staphylococcus aureus) infection will be mainly described.

In the medical facility 4, a simple examination is performed on the patient 42 so that the examination results can be obtained quickly. Hereinafter, the simple inspection will be referred to as the first inspection. In the first test, a specimen is collected from the patient 42 and tested using the specimen. The first test includes microscopy, gene amplification test, or immunological test on the sample, and may use a test kit. Immunological tests include enzyme immunoassay (EIA) (e.g. direct competitive ELISA, indirect competitive ELISA, sandwich ELISA), fluorescence immunoassay (FIA), membrane assay, luminescence immunoassay, spin immunoassay, Western blotting, including immunohistochemical staining. Membrane assay tests include flow-through membrane assay tests and lateral flow immunochromatographic tests. Infectious disease immunological tests for directly detecting pathogens (antigens) such as bacteria or viruses causing infectious diseases include tests by membrane assays, especially immunochromatographic tests.

In the first test, test results are obtained within a short period of time, for example, within several tens of minutes to several hours. Hereinafter, the inspection result of the first inspection will be referred to as the first inspection result. A first test results in a first test result indicating that patient 42 may be suffering from a disease. For example, a positive first test result indicates that the disease is likely, and a negative first test result indicates that the disease is unlikely.

If the first test results indicate that the patient 42 is likely to have the disease, a workup is performed to make a definitive diagnosis. Hereinafter, the detailed inspection will be referred to as the second inspection. The second inspection is an inspection that requires a longer inspection period than the first inspection. A sample is sent from the medical facility 4 to the testing institution, and the second testing is performed at the testing institution. The second inspection includes an inspection for isolating and culturing pathogens such as bacteria. In the case of diagnosing MRSA infection, the second examination includes, for example, isolation culture, pure culture, and drug susceptibility testing. The second test requires a long period of time, for example, several days to a month, until test results are obtained. Hereinafter, the inspection result of the second inspection will be referred to as the second inspection result. Until the second test result is obtained, the situation may worsen, such as the spread of infection. The diagnostic system 100 performs processing for making a diagnosis before the second test result is obtained.

The diagnostic system 100 includes an information processing device 1 that executes processing for diagnosis. The information processing device 1 is connected to a communication network 5 such as the Internet. The medical facility 4 is provided with an input/output device 2 for inputting/outputting information. The input/output device 2 is operated by an operator 41 such as a doctor. The input/output device 2 is connected to the information processing device 1 via the communication network 5 . From the input/output device 2 , the first test result regarding the patient 42 and patient information regarding the patient 42 are input to the information processing device 1 through the communication network 5 .

The inspection agency is provided with a storage device 3 for storing information. The storage device 3 stores area information including second test results of a second test using a sample collected in a predetermined area including the medical facility 4 . Storage device 3 is connected to information processing device 1 via communication network 5 . The information processing device 1 acquires regional information from the storage device 3 through the communication network 5 . The information processing apparatus 1 acquires the first test result, patient information, and area information, and performs a process of outputting test prediction that predicts the second test result before the second test result is obtained.

FIG. 2 is a block diagram showing a functional configuration example of the input/output device 2. As shown in FIG. The input/output device 2 is configured using a computer such as a personal computer or a tablet computer. The input/output device 2 includes an arithmetic unit 21, a memory 22 for storing temporary data generated by arithmetic operations, and a non-volatile storage unit 23 such as a hard disk. The calculation unit 21 is configured using, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a multi-core CPU. Moreover, the calculation unit 21 may be configured using a quantum computer. The memory 22 is, for example, a RAM (Random Access Memory). The storage unit 23 stores computer programs 231 . The computing unit 21 executes processing necessary for the input/output device 2 according to the computer program 231 .

The input/output device 2 also includes an operation unit 24 that receives operations from the operator 41 , a display unit 25 that displays images, and a communication unit 26 . The operation unit 24 receives information such as text by receiving an operation from the operator 41 . The operating unit 24 is, for example, a keyboard or pointing device. The display unit 25 is, for example, a liquid crystal display or an EL display (Electroluminescence Display). The communication unit 26 is connected to the communication network 5 and communicates with devices outside the input/output device 2 through the communication network 5 .

The storage unit 23 stores the data of the first examination result of the first examination performed on the patient 42 at the medical facility 4 . The first inspection result is input from the operation unit 24 by the operator 41 operating the operation unit 24 , and data of the first inspection result is stored in the storage unit 23 . The first test result may be input through the communication unit 26 or may be input through an interface unit (not shown). The storage unit 23 stores first test result data for each of the plurality of patients 42 .

The storage unit 23 also stores patient information about the patient 42 . FIG. 3 is a conceptual diagram showing an example of contents of patient information. The patient information includes the patient's 42 age, sex, medically-discovered symptoms, or examination history. Patient information may include treatment history. FIG. 3 exemplifies a history of MRSA infection and carriage, a history of hospitalization, a history of artificial dialysis, and a history of surgery as treatment histories included in the patient information. Patient information may include other medical or treatment history, such as history of admission to a nursing home or long-term care facility, history of kidney disease or diabetes, or history of indwelling catheters or medical devices inserted into the body through the skin. . As shown in FIG. 3, the patient information may include antimicrobial medication history. In addition, patient information may include contact with MRSA-infected persons or MRSA-carriers, whether the patient is in a crowded living environment, or whether the facility is usually used (nursery school, school, company, commercial facility, etc.). ), etc., may include an action history.

At least part of the patient information may be input from the operation unit 24 and stored in the storage unit 23 by the operator 41 operating the operation unit 24 . At least part of the patient information may be stored in advance in the storage unit 23 in the form of medical charts or the like when the patient 42 had a medical examination at the medical facility 4 in the past. At least part of the patient information may be stored in a storage device (not shown) outside the input/output device 2 and acquired from the storage device through the communication section 26 or an interface section (not shown). The storage unit 23 stores patient information for each of the plurality of patients 42 .

FIG. 4 is a block diagram showing a functional configuration example of the storage device 3. As shown in FIG. The storage device 3 is configured using a computer such as a server device. The storage device 3 includes an arithmetic unit 31 , a memory 32 for storing temporary data generated by arithmetic operation, a nonvolatile storage unit 33 such as a hard disk, and a communication unit 34 . The calculation unit 31 is, for example, a CPU, and the memory 32 is, for example, a RAM. The storage unit 33 stores a computer program 331 . The calculation unit 31 executes processing necessary for the storage device 3 according to the computer program 331 . The communication unit 34 is connected to the communication network 5 and communicates with devices outside the storage device 3 through the communication network 5 . The storage device 3 may be composed of a plurality of computers.

The storage unit 33 stores area information. The regional information includes second test results of second tests performed in the past using specimens collected within a predetermined region including the medical facility 4 . For example, a plurality of medical facilities including the medical facility 4 exist within a predetermined area. 2 area information including test results is stored in the storage device 3; A second test result of a second test performed using specimens collected within a region gives the number of occurrences of a particular disease within the region.

FIG. 5 is a conceptual diagram showing an example of contents of area information. The area information includes location information indicating the location of a predetermined area including the medical facility 4 . In the example shown in FIG. 5, the location information includes the address of the medical facility 4, the name of the area including the medical facility 4, GPS information representing the location of the medical facility 4 used in GPS (Global Positioning System), Contains latitude and longitude. The location information may include the range of addresses in a given area, the names of multiple lands included in the given area, the range of GPS information representing each position in the given area, or the latitude and longitude range of the given area. The location information may include other information indicating the location of the predetermined area.

Regional information includes the number of occurrences of a particular disease within a given region, or the number of occurrences of a particular disease within a given region by patient age or gender. The regional information may include the number of occurrences for each category of facility where the disease occurred, such as a nursery school, school, or company. Regional information may include chronological changes in the incidence of specific diseases. In the example shown in FIG. 5, the number of disease outbreaks in a region is indicated by the number of affected people, and by recording the number of affected people for each week, chronological changes in the disease outbreak status are recorded. . The example shown in FIG. 5 shows time-series changes in the number of disease occurrences in a region, the number of occurrences by age, the number of occurrences by gender, and the number of occurrences by facility.

The storage unit 33 stores area information for each of a plurality of areas. The storage unit 33 may store area information for each type of diagnosis, such as when diagnosing different types of diseases. Also, the predetermined area including the medical facility 4 may be an area such as an administrative division, in which the land is fixedly divided in advance. For example, there may be multiple medical facilities 4 that include the same region. Alternatively, the predetermined area including the medical facility 4 may be set for each medical facility 4 . For example, an area within a predetermined distance from the medical facility 4 may be set as the predetermined area. The storage unit 33 does not store area information in advance, but stores information to be included in the area information, and the area information is generated from the information stored in the storage unit 33 as necessary. may

FIG. 6 is a block diagram showing a functional configuration example of the information processing device 1 according to the first embodiment. The information processing device 1 is configured using a computer such as a server device. The information processing apparatus 1 includes a calculation unit 11, a memory 12 for storing temporary data generated by calculation, a nonvolatile storage unit 13 such as a hard disk, and a drive for reading information from a recording medium 10 such as an optical disc. A unit 14 and a communication unit 16 are provided. The calculation unit 11 is, for example, a CPU, and the memory 12 is, for example, a RAM. The communication unit 16 is connected to the communication network 5 and communicates with devices external to the information processing device 1 through the communication network 5 . The information processing device 1 may be composed of a plurality of computers.

The calculation unit 11 causes the drive unit 14 to read the computer program 131 recorded on the recording medium 10 and causes the storage unit 13 to store the read computer program 131 . The calculation unit 11 executes processing necessary for the information processing apparatus 1 according to the computer program 131 . Note that the computer program 131 may be downloaded from the outside of the information processing device 1 . In this case, the information processing device 1 does not have to include the drive unit 14 .

The information processing device 1 further includes a learning model 15 used to predict the second test result. The learning model 15 is trained to output a test prediction that predicts a second test result for the patient 42 according to the first test result for the patient 42, patient information, and local information. The learning model 15 is implemented by the computing unit 11 executing information processing according to the computer program 131 . Also, the learning model 15 may be configured using hardware. For example, the learning model 15 may include a processor and a memory that stores necessary programs and data. Also, the learning model 15 may be implemented using a quantum computer.

The storage unit 13 stores a teacher database including teacher data for learning the learning model 15 . A teacher database contains a plurality of teacher data. FIG. 7 is a conceptual diagram showing an example of contents of teacher data according to the first embodiment. One piece of teacher data includes the first examination result of the first examination and the second examination result of the second examination performed on a certain patient 42, patient information of the same patient 42, and the medical facility 4 at which the patient 42 visited. Contains local information about the containing region. In the example shown in FIG. 7, the first inspection result and the second inspection result include the inspection date and time and the inspection result. As the second test result, the probability that the infected bacteria is MRSA is expressed numerically. The second test result may be non-probability information, such as positive or negative. As shown in FIG. 7 , the teacher data includes information related to patient information of the patient 42 among the information included in the regional information regarding the region including the medical facility 4 . For example, information related to the sex and age of the patient 42 is extracted from the regional information and included in the training data. Location information is also included in the regional information.

The plurality of teacher data may all be teacher data for different patients 42, or a plurality of teacher data for the same patient 42 may be included in the teacher database. For example, there may be a plurality of teacher data for the same patient 42 who performed the first examination and the second examination at different times. The regional information included in each training data may be regional information about the same period or regional information about a different period. The regional information included in each training data may be regional information regarding the same region or regional information regarding a different region.

The storage unit 13 stores a treatment database containing treatment information representing details of treatments for patients. FIG. 8 is a conceptual diagram showing an example of contents of a treatment database. The second test result reveals the patient's medical condition, such as the presence or absence of a specific disease. In the example shown in FIG. 8, the second test result indicates the probability that the infectious bacterium is MRSA. The content of the treatment to be given to the patient is determined according to the patient's medical condition. For this reason, treatment information representing the content of treatment is determined according to the second examination result. In the treatment database, the second examination result and treatment information representing the details of the treatment to be given to the patient according to the second examination result are recorded in association with each other. In the example shown in FIG. 8, treatment information is associated with each of the plurality of second examination results. For each value of the probability that the infected bacterium is MRSA, as treatment information, the recommended anti-MRSA drug prescription, the need for a detailed examination, the need for isolation of the ward, and the high need for sterilization response is associated with

The first test result, although not as conclusive as the second test result, indicates that the patient 42 may be suffering from a particular disease. Thus, there is a correlation between a first test result and a second test result that indicates that patient 42 may be suffering from a particular disease. Of the patient information, the symptoms and examination history of the patient 42 are correlated with the second examination result. Depending on the medical history, treatment history, or medication history of the patient 42 among the patient information, there is a high possibility that the patient 42 is suffering from a specific disease. For example, a history of MRSA infection and colonization increases the likelihood that patient 42 has MRSA. For example, if the patient 42 has a history of hospitalization, dialysis, or surgery, there is a chance that the patient 42 has had an infection in the past, and the patient 42 is more likely to have an infection. For example, the possibility of contracting a disease changes depending on the contents of drugs included in the medication history. Among the patient information, there is a high possibility that the patient 42 will contract an infectious disease depending on the behavior history of the patient 42, such as contact with an infected person or carrier of an infectious disease, or being in a crowded living environment. Become.

Further, when the number of occurrences of a specific disease in the predetermined area including the medical facility 4 is high among the area information, the possibility that the patient 42 is afflicted with the disease increases. If the incidence of the disease is high for the gender or age to which the patient 42 belongs, the likelihood that the patient 42 is afflicted increases. If the facility that the patient 42 usually uses has a large number of disease occurrences, the possibility that the patient 42 is afflicted with the disease increases. Also, when the time-series change in the number of occurrences tends to increase, the possibility that the patient 42 is afflicted with the disease increases, and when it tends to decrease, the possibility that the patient 42 is afflicted with the disease decreases. If the specific disease is an infectious disease, there is a higher correlation between the number of disease occurrences or time series changes in the number of occurrences in the region and the likelihood that the patient 42 is afflicted. Thus, there is a correlation between the patient information and the regional information and the second test result indicating the possibility that the patient 42 is suffering from a specific disease. Therefore, it is possible to obtain an examination prediction that predicts the second examination result according to the first examination result, the patient information and the area information.

FIG. 9 is a conceptual diagram showing a functional configuration example of the learning model 15. As shown in FIG. The learning model 15 uses a neural network with an input layer, an intermediate layer and an output layer each having a plurality of nodes. The input layer has a plurality of nodes 151 into which first test results, patient information and local information regarding the patient 42 are input. For example, the first test result is input to one node 151 and each of a plurality of pieces of information included in patient information and area information is input to one of the nodes 151 . Non-numerical information among the first test result, patient information, and regional information may be converted into numerical values and then input to each node 151 . Each numerical value included in the time series of the number of occurrences of diseases included in the regional information may be input to each node 151, or a part of the numerical values included in the time series of the number of occurrences may be input to the node 151. , a value obtained by averaging the numerical values included in the time series of the number of occurrences may be input to the node 151 . In addition, image data of a graph showing time-series changes in the number of disease occurrences with the horizontal axis as time and the vertical axis as the number of disease occurrences may be input to the node 151 as data representing time-series changes in the number of disease occurrences. good.

The intermediate layer of the learning model 15 has a plurality of nodes 152 that perform calculations using parameters in the data input from the node 151 of the input layer. The output layer has a plurality of nodes 153 that receive data calculated from nodes 152 in the intermediate layer and output test predictions that predict the second test results. For example, each node 153 corresponds to a respective numerical value from 0% to 100% of the probability that the patient has a particular disease. Each node 153 outputs a score, and the learning model 15 outputs a test prediction with the probability corresponding to the node 153 having the maximum score value as the test prediction output. If the second test result is non-probability information, the learning model 15 may output a numerical value corresponding to the non-probability information, or may convert the numerical value to non-probability information. For example, the learning model 15 may treat negatives as 0% probability and positives as 100% probability.

Although FIG. 9 shows an example in which the intermediate layer is one layer, the intermediate layer may be multiple layers. The learning model 15 may use a convolutional neural network (CNN: Convolutional Neural Network) or a recurrent neural network (RNN: Recurrent Neural Network) as a neural network. Note that the learning model 15 may be a learning model other than a neural network.

The information processing device 1 performs machine learning of the learning model 15 . FIG. 10 is a flow chart showing the procedure of learning processing of the learning model 15 executed by the information processing apparatus 1 . A step is abbreviated as S below. The computing unit 11 executes the following processes according to the computer program 131 . The calculation unit 11 acquires a plurality of teacher data (S11). In S<b>11 , for example, the calculation unit 11 acquires a plurality of teacher data by reading a plurality of teacher data included in a teacher database stored in the storage unit 13 in advance to the memory 12 . For example, the communication unit 16 receives teacher data through the communication network 5, the calculation unit 11 stores a teacher database including the received teacher data in the storage unit 13, and reads the teacher data to the memory 12, thereby allowing a plurality of teachers to Get data. Further, for example, the communication unit 16 receives the first test result, the patient information, and the region information, the calculation unit 11 extracts the second test result from the region information, A plurality of teacher data are obtained by combining the information and the second inspection result to generate teacher data.

The computing unit 11 generates a learning model 15 that outputs test prediction when the first test result, patient information, and area information are input, based on the acquired plurality of teacher data (S12). In S<b>12 , the calculation unit 11 inputs the first test result, patient information, and area information included in the teacher data to the input layer of the learning model 15 . A score is output from each node 153 of the output layer by the learning model 15 . The calculation unit 11 calculates an error using an error function using the output score and the score corresponding to the second inspection result included in the teacher data as variables, and learns so that the error is minimized by the error backpropagation method. Adjust the parameters of the calculation of each node of the model 15 . The calculation unit 11 performs machine learning of the learning model 15 by repeating processing for each of the plurality of teacher data and adjusting the parameters of the learning model 15 . The calculation unit 11 stores the parameters in the storage unit 13, adjusts the computer program 131, or adjusts the configuration of the learning model 15 so as to reflect the adjusted final parameters. In this way, a learning model 15 is generated that has been trained to output an examination prediction when the first examination result, patient information, and area information are input. The calculation unit 11 ends the learning process.

Note that the learning model 15 includes a learning model for each region, and the learning model may be learned for each region. When learning a learning model for a certain region, the computing unit 11 obtains the first test result, patient information, and second test result for a patient 42 who has undergone a medical examination at a medical facility 4 included in a certain region, and regional information for the certain region. Machine learning is performed using teacher data containing In addition, the learning model 15 includes a learning model for each type of diagnosis, and the learning model may be learned for each type of diagnosis. When learning a learning model for a specific type of diagnosis, the computing unit 11 uses teacher data including first test results, region information, patient information, and second test results for a specific type of diagnosis, Do machine learning. Moreover, the learning model 15 may include a learning model for each part of the patient information. For example, learning models 15 may include patient age-specific learning models. When learning a learning model for a specific age, the calculation unit 11 uses teacher data including first test results, regional information, patient information, and second test results for patients of a specific age to perform machine learning I do.

The diagnostic system 100 uses the learning model 15 to perform processing for making a diagnosis before the second test result is obtained. After the first examination is performed on the patient 42 at the medical facility 4 and the first examination result is obtained, the first examination result is input to the input/output device 2 . For example, the operator 41 operates the operation unit 24 to input the first inspection result to the input/output device 2 . Before the second test result is obtained, the input/output device 2 transmits the first test result of the patient 42 , area information, and patient information of the patient 42 to the information processing device 1 via the communication network 5 .

For example, in accordance with a transmission instruction input by the operator 41 by operating the operation unit 24, the calculation unit 21 transmits the first test result data and the patient information regarding the patient 42 stored in the storage unit 23. It causes the unit 26 to transmit to the information processing apparatus 1 . Further, the calculation unit 21 causes the communication unit 26 to acquire regional information about the region including the medical facility 4 from the storage device 3 via the communication network 5 and transmit the acquired regional information to the information processing device 1 . At this time, the storage device 3 may transmit regional information stored in the storage unit 33 in advance, or may generate and transmit regional information about a predetermined region including the medical facility 4 . Further, the input/output device 2 may receive patient information when the operator 41 operates the operation unit 24 , and the calculation unit 21 may cause the communication unit 26 to transmit the received patient information to the information processing apparatus 1 . . Further, the calculation unit 21 causes the communication unit 26 to transmit an instruction to transmit regional information about the area including the medical facility 4 to the storage device 3, and the storage device 3 transmits the regional information to the information processing device 1 according to the transmission instruction. . The storage device 3 may transmit regional information stored in the storage unit 33 in advance, or may generate and transmit regional information about a predetermined region including the medical facility 4 .

FIG. 11 is a flow chart showing the procedure of diagnostic processing performed by the information processing apparatus 1 . The computing unit 11 executes the following processes according to the computer program 131 . When the communication unit 16 receives the first test result, the region information, and the patient information regarding the patient 42, the calculation unit 11 acquires the first test result, the region information, and the patient information (S21). The calculation unit 11 causes the memory 12 or the storage unit 13 to store the acquired first test result, area information, and patient information. The processing of S21 corresponds to the first acquisition unit, the second acquisition unit, and the third acquisition unit.

The calculation unit 11 inputs the acquired first test result, region information, and patient information to the learned learning model 15, and causes the learning model 15 to output test prediction that predicts the second test result (S22). For example, as a test prediction, a numerical value of the probability that the bacterium that infected the patient 42 is MRSA is output. The processing of S22 corresponds to the output unit. Next, the calculation unit 11 outputs treatment information representing details of recommended treatment for the patient 42 according to the examination prediction (S23). In S23, the calculation unit 11 refers to the treatment database stored in the storage unit 13, reads treatment information recorded in association with the second examination result corresponding to the examination prediction output in S22, and reads out the treatment information. The treatment information is output as treatment information.

Next, the calculation unit 11 generates image information for displaying an image showing the diagnosis result on the input/output device 2, and causes the communication unit 16 to transmit the generated image information to the input/output device 2, thereby outputting the generated image information. (S24). In S24, the calculation unit 11 generates and outputs image information for displaying an image including the obtained first examination result, examination prediction, and treatment information. Further, the computing unit 11 obtains the acquired patient information, a group of second test results of a second test performed using a sample collected in a predetermined area including the medical facility 4, and an area including a map showing the predetermined area. Image information for displaying an image containing information may be generated and output.

The input/output device 2 receives the image information transmitted from the information processing device 1 at the communication unit 26, and the calculation unit 21 causes the display unit 25 to display an image based on the image information. FIG. 12 is a schematic diagram illustrating an example of an image including a first examination result, examination prediction, and treatment information according to the first embodiment; As the first test result, the result of the bacteriological test is displayed, and as the test prediction, the probability that the infectious bacterium is MRSA is displayed as a numerical value. Further, as treatment information, recommended prescriptions such as anti-MRSA drugs or the need for further work-up are displayed. An operator 41 such as a doctor confirms the content of the image displayed on the display unit 25, diagnoses the patient 42 with reference to the examination prediction, and determines the content of treatment for the patient 42 with reference to the treatment information. .

FIG. 13 is a schematic diagram showing an example of an image including patient information, a second test result group, and area information including a map showing a predetermined area. A map of a predetermined area including the medical facility 4 is displayed, and the occurrence of diseases within the area is displayed. In addition, the occurrence status by facility category such as nursery school, school, or company is displayed. In addition, patient information for patient 42 is displayed, such as patient 42's treatment history and medication history. An operator 41 such as a doctor can confirm the content of the image displayed on the display unit 25 and judge the validity of the examination prediction and treatment information based on the area information and patient information. The contents of the images shown in FIGS. 12 and 13 are examples. The information processing apparatus 1 ends the diagnosis process.

As described above, the learning model 15 is used to diagnose the patient 42, and the patient 42 is treated according to the diagnosis result. Treatment for patient 42 is performed before the second test results are obtained.

After the second test result is obtained, diagnostic system 100 may retrain learning model 15 . After obtaining the second test result for the patient 42 , the second test result is input to the input/output device 2 . For example, the operator 41 operates the operation unit 24 to input the second inspection result to the input/output device 2 . The second test result may be input to the input/output device 2 by receiving the data of the second test result transmitted via the communication network 5 by the communication unit 26 . The input/output device 2 transmits the first test result, the regional information, the patient information and the second test result regarding the patient 42 to the information processing device 1 via the communication network 5 .

FIG. 14 is a flow chart showing the procedure of re-learning processing performed by the information processing apparatus 1 . The computing unit 11 executes the following processes according to the computer program 131 . The calculation unit 11 acquires relearning data including the first test result, area information, patient information, and second test result (S31). For example, when the communication unit 16 receives the first test result, the region information, the patient information, and the second test result regarding the patient 42, the calculation unit 11 receives the first test result, the region information, the patient information, and the second test result. Get retraining data containing For example, the calculation unit 11 temporarily stores the received first test result, region information, patient information, and second test result in the storage unit 13, and outputs the first test result, the region information, the patient information, and the second test result from the storage unit 13. The relearning data is obtained by reading the inspection result and generating the relearning data.

The calculation unit 11 re-learns the learning model 15 using the acquired re-learning data (S32). In S32, the calculation unit 11 inputs the first test result, the region information, and the patient information included in the relearning data to the input layer of the learning model 15, and performs the test corresponding to the second test result included in the relearning data. Machine learning is performed to adjust the parameters of each node of the learning model 15 so that predictions can be output from the output layer of the learning model 15 . The calculation unit 11 stores the parameters in the storage unit 13, adjusts the computer program 131, or adjusts the configuration of the learning model 15 so as to reflect the adjusted final parameters. In this way, the learning model 15 re-learned by machine learning is generated. The calculation unit 11 ends the re-learning process. The process of S32 may be performed each time one piece of relearning data is obtained, may be performed when a predetermined number of relearning data are obtained, or may be performed periodically. Re-learning updates the learning model 15 so that it can more accurately predict the second test result. Therefore, it is possible to output more appropriate inspection prediction.

As described in detail above, the information processing apparatus 1 uses the learning model 15 based on the first test result, the patient information, and the area information regarding the patient 42 to output test prediction that predicts the second test result. An inspection prediction is obtained that predicts the second inspection result before the second inspection result of the second inspection requiring a long inspection period is obtained. By making a diagnosis according to the examination prediction, it is possible to quickly make an appropriate diagnosis for the patient 42 . In addition, the information processing apparatus 1 outputs treatment information representing details of recommended treatment for the patient 42 in accordance with the examination prediction. Appropriate treatment according to the treatment information can be administered to the patient 42 before the second test result is obtained. It is possible to prevent the situation from worsening, such as the spread of infection or deterioration of the condition of the patient 42, until the second test result is obtained.

In this embodiment, an example of diagnosing an MRSA infection has been mainly described, but the diagnostic system 100 can diagnose other infectious diseases. For example, a diagnosis of pandemic influenza may be performed. A patient 42 with influenza-like clinical features is subjected to an influenza test as a first test, and if the first test result is negative but the case definition of pandemic influenza is met, it is determined whether it is a pandemic influenza. For this reason, a second test for identifying the influenza virus subtype may be performed, and a diagnosis using the diagnostic system 100 may be performed. In this case, the regional information includes the outbreak status of new strains of influenza. The patient information may include symptoms such as fever, muscle pain, arthralgia, or malaise, body temperature, or a test result image of an influenza rapid diagnostic kit. The patient information may also include the infection status in the vicinity of the patient 42 such as family members, or the infection status in facilities such as workplaces or schools related to the patient 42 . Treatment information includes prescribing and selection of anti-influenza drugs such as Tamiflu, Relenza, Inavir, Xofluza, or Rapiacta, whether a work-up is required, or whether isolation is required. The test prediction includes the probability of having influenza or the probability of having pandemic influenza.

For example, tuberculosis drug resistance may be diagnosed. Patient 42 is tested for tuberculosis bacteria as a first test, and if the first test result is positive, a second test is tuberculosis for drugs such as rifampicin, isoniazid, pyrazinamide, streptomycin, ethambutol, or kanamycin. resistance test. Regional information includes the occurrence status of drug-resistant tuberculosis bacteria. Patient information includes tuberculosis infection history and treatment history, medication history, contact history with resistant tuberculosis patients, expected number of tuberculosis patients at medical facility 4, accommodation capacity of negative pressure isolation private rooms, or X-ray diagnosis It may also include the results of diagnostic imaging. The treatment information includes the need for consideration of isolation of the patient 42, whether or not a rapid drug susceptibility test method (genetic test) is indicated, or the recommended selection of an anti-tuberculosis drug. The test prediction includes the probability of having drug-resistant M. tuberculosis.

The diagnostic system 100 may perform processing according to the type of diagnosis. The information processing apparatus 1 uses the learning model for each type of diagnosis included in the learning model 15 to execute the processes of S21 to S24 and the processes of S31 to S32 for each type of diagnosis. The input/output device 2 displays information such as examination prediction on the display unit 25 for each type of diagnosis. The diagnostic system 100 can also diagnose diseases other than infectious diseases.

In the present embodiment, the information processing apparatus 1 acquires the treatment information representing the content of the recommended treatment. good too. In this form, the treatment database is stored in the storage unit 23 of the input/output device 2 . The information processing device 1 transmits the test prediction data to the input/output device 2 without performing the process of S23. The input/output device 2 receives the test prediction data at the communication unit 26, and the calculation unit 21 refers to the treatment database stored in the storage unit 23, and associates the data with the second test result corresponding to the test prediction. The recorded treatment information is acquired as treatment information. Also in this form, it is possible to perform appropriate treatment on the patient 42 according to the treatment information before obtaining the second test result.

<Embodiment 2>
In the second embodiment, the information processing apparatus 1 uses the learning model 15 to output treatment information. FIG. 15 is a block diagram showing a functional configuration example of the information processing device 1 according to the second embodiment. The storage unit 13 does not store a treatment database. The teacher data included in the teacher database stored in the storage unit 13 includes treatment information in addition to first test results, second test results, patient information, and area information.

FIG. 16 is a conceptual diagram showing an example of contents of teacher data according to the second embodiment. One piece of teacher data includes the first test result of a first test and the second test result of a second test performed on a certain patient 42, patient information of the same patient 42, and the medical facility 4 where the patient 42 visited. It includes regional information about the region and treatment information representing the content of the treatment to be given to the patient 42 according to the second test result. In the example shown in FIG. 16, treatment information is recorded that indicates the details of the treatment given to the patient 42 for whom the second examination result was obtained. The treatment information includes the need for prescription of anti-MRSA drugs, the need for work-up, the need for ward isolation, and the need for decolonization measures. The contents of the treatment information may be represented by probabilities.

As an example, the learning model 15 uses a neural network as in the first embodiment. The output layer of the learning model 15 has a plurality of nodes 153 that receive data calculated from nodes 152 in the intermediate layer and output examination predictions that predict second examination results and treatment information. For example, the output layer includes a plurality of nodes 153 corresponding to respective numerical values from 0% to 100% of the probability that the patient has a particular disease. The output layer is the probability of the need for prescription of anti-MRSA drugs, the need for detailed examination, the need for isolation of the ward, and the need for disinfection response, each of which is a numerical value of 0% to 100%. contains a plurality of nodes 153 corresponding to . Each node 153 outputs a score, and the probability that the node 153 with the largest score value among the plurality of nodes 153 corresponding to the test prediction corresponds is output as the test prediction. Also, the probability that the node 153 having the maximum score value among the plurality of nodes 153 corresponding to each piece of treatment information corresponds is output as the treatment information. The configuration of other parts of the information processing apparatus 1 is the same as that of the first embodiment. Also, the configuration of the diagnostic system 100 other than the information processing device 1 is the same as that of the first embodiment.

The information processing apparatus 1 performs machine learning of the learning model 15 by processing including the same procedures as S11 to S12 shown in the flowchart of FIG. The computing unit 11 executes the following processes according to the computer program 131 . The calculation unit 11 acquires a plurality of teacher data each including a first test result, a second test result, patient information, area information, and treatment information (S11). The computing unit 11 then creates a learning model that outputs test prediction and treatment information corresponding to the second test result when the first test result, patient information, and region information are input based on the acquired plurality of teacher data. 15 is generated (S12). Of the first test result, patient information, area information, second test result, and treatment information, non-numeric information is processed after being converted into a numerical value.

In S<b>12 , the calculation unit 11 inputs the first test result, patient information, and area information included in the teacher data to the input layer of the learning model 15 . The calculation unit 11 calculates an error using an error function whose variables are the score output from each node 153 of the output layer of the learning model 15 and the score corresponding to the second inspection result and the treatment information included in the teacher data, The operation parameters of each node of the learning model 15 are adjusted so that the error is minimized by the error backpropagation method. The calculation unit 11 performs machine learning of the learning model 15 by repeating processing for each of the plurality of teacher data. The calculation unit 11 stores the parameters in the storage unit 13, adjusts the computer program 131, or adjusts the configuration of the learning model 15 so as to reflect the adjusted final parameters. In this way, a learning model 15 is generated that has been learned to output examination prediction and treatment information when the first examination result, patient information and area information are input. The calculation unit 11 ends the learning process. The processing executed by the input/output device 2 is the same as that of the first embodiment.

The information processing apparatus 1 performs diagnostic processing through processing including procedures similar to those of S21 to S24 shown in the flowchart of FIG. The computing unit 11 executes the following processes according to the computer program 131 . The calculation unit 11 acquires the first test result, the region information, and the patient information by receiving the first test result, the region information, and the patient information regarding the patient 42 with the communication unit 16 (S21). The calculation unit 11 inputs the acquired first test result, region information, and patient information to the learned learning model 15, causes the learning model 15 to output test prediction that predicts the second test result (S22), The learning model 15 is caused to output treatment information (S23). Next, the calculation unit 11 generates image information for causing the input/output device 2 to display an image showing a diagnosis result including treatment information, and causes the communication unit 16 to transmit the generated image information to the input/output device 2. is output (S24).

FIG. 17 is a schematic diagram illustrating an example of an image including a first examination result, examination prediction, and treatment information according to the second embodiment; As the first test result, the result of the bacteriological test is displayed, and as the test prediction, the probability that the infectious bacterium is MRSA is displayed as a numerical value. Furthermore, as treatment information, the prescription of anti-MRSA drugs, the need for detailed examinations, the need for ward isolation, and the need for sterilization measures are displayed in terms of probabilities. An operator 41 such as a doctor diagnoses the patient 42 with reference to the examination prediction, and determines the details of the treatment for the patient 42 with reference to the treatment information. The calculation unit 11 ends the diagnosis process. The processing executed by the input/output device 2 is the same as that of the first embodiment.

After the second test result is obtained, diagnostic system 100 may retrain learning model 15 . After the result of the second examination is obtained, the condition of the patient 42 at the time of the second examination becomes clear, and the treatment that should have been given to the patient 42 is also clarified. After obtaining the second test result for the patient 42 , the second test result is input to the input/output device 2 . Furthermore, treatment information representing the content of the treatment that should have been given to the patient 42 according to the second examination result is input to the input/output device 2 . For example, the operator 41 operates the operation unit 24 to input the second examination result and treatment information to the input/output device 2 . The input/output device 2 transmits, via the communication network 5, the first test result, area information, patient information, second test result, and treatment information indicating the details of the treatment that should have been performed on the patient 42. It is transmitted to the information processing device 1 .

The information processing apparatus 1 performs the relearning process by the process including the same procedure as S31 to S32 shown in the flowchart of FIG. The computing unit 11 executes the following processes according to the computer program 131 . The calculation unit 11 acquires relearning data including the first test result, area information, patient information, second test result, and treatment information (S31). For example, when the communication unit 16 receives the first test result, the region information, the patient information, the second test result, and the treatment information regarding the patient 42, the calculation unit 11 receives the first test result, the region information, the patient information, the first 2 Acquire relearning data including test results and treatment information.

The calculation unit 11 re-learns the learning model 15 using the acquired re-learning data (S32). In S32, the calculation unit 11 inputs the first test result, the region information, and the patient information included in the relearning data to the input layer of the learning model 15, and converts the second test result and treatment information included in the relearning data into Machine learning is performed to adjust the parameters of each node of the learning model 15 so that the corresponding test prediction and treatment information can be output from the output layer of the learning model 15 . The calculation unit 11 stores the parameters in the storage unit 13, adjusts the computer program 131, or adjusts the configuration of the learning model 15 so as to reflect the adjusted final parameters. In this way, the learning model 15 re-learned by machine learning is generated. The calculation unit 11 ends the re-learning process. Re-learning updates the learning model 15 so that more appropriate treatment information is obtained.

As described in detail above, in the second embodiment, the information processing apparatus 1 uses the learning model 15 to predict the second test result based on the first test result, the patient information, and the regional information regarding the patient 42. Outputs test prediction and treatment information. Also in the second embodiment, it is possible to make a diagnosis according to the test prediction and to perform appropriate treatment on the patient 42 according to the treatment information before the second test result is obtained. Therefore, it is possible to prevent the situation from becoming worse during the period until the second inspection result is obtained.

In Embodiments 1 and 2, the diagnosis system 100 has shown a form in which one input/output device 2 is provided, but the diagnosis system 100 may be provided with a plurality of input/output devices 2 . That is, there are a plurality of medical facilities 4 , and each medical facility 4 is provided with an input/output device 2 . If the medical facility 4 is different, the area including the medical facility 4 may be different and the area information may be different. The information processing apparatus 1 uses the regional learning model included in the learning model 15 to perform the processes of S21 to S24 and the processes of S31 to S32 for each of the patients 42 who have been examined at different medical facilities 4. .

In Embodiments 1 and 2, an example in which the second examination is performed at a testing institution has been shown, but the second examination may be performed within the medical facility 4 . In Embodiments 1 and 2, the form in which the storage device 3 is provided in the inspection institution has been shown, but the storage device 3 may be provided in a place different from the inspection institution. For example, the storage device 3 may be provided within the medical facility 4 . A plurality of storage devices 3 may be provided. At least part of the functions of the storage device 3 may also be performed by the information processing device 1 .

In Embodiments 1 and 2, the input/output device 2 performs input and output of information, but the input/output device 2 is divided into a device used for inputting information and a device used for outputting information. may be separate. The information processing device 1 may be provided within the examination period or within the medical facility 4 .

The present invention is not limited to the contents of the above-described embodiments, and various modifications are possible within the scope of the claims. That is, the technical scope of the present invention also includes embodiments obtained by combining technical means appropriately modified within the scope of the claims.

Reference Signs List 1 information processing device 10 recording medium 11 computing unit 13 storage unit 15 learning model 100 diagnostic system 131 computer program 2 input/output device 3 storage device 4 medical facility 41 operator 42 patient 5 communication network

Claims (13)

A first test result of a first test performed on a patient by a computer at a medical facility, which is performed using a specimen collected within a predetermined area including the medical facility and requires a longer test period than the first test. acquiring a plurality of teacher data including regional information including second test results of a second test, patient information about the patient, and treatment information representing details of treatment for the patient or second test results about the patient;
A computer predicts treatment information representing details of a recommended treatment for a patient or a second examination result for the patient when the first examination result, the region information, and the patient information are input based on the acquired teacher data. A learning model generation method characterized by generating a learning model that outputs a test prediction. The regional information includes the number of occurrences of a specific disease within the predetermined region, the number of occurrences of the specific disease within the predetermined region by patient age or gender, or the number of occurrences of the specific disease within the predetermined region. 2. The learning model generation method according to claim 1, wherein the time-series change in the occurrence of is included. 3. The learning model generation method according to claim 2, wherein the specific disease is an infectious disease. The learning model generation method according to any one of claims 1 to 3, wherein the first examination includes microscopic examination, gene amplification examination, or immunological examination of a sample collected from the patient. The learning model generation method according to any one of claims 1 to 4, wherein the second inspection includes an inspection for isolating and culturing pathogens. The learning model generating method according to any one of claims 1 to 5, wherein the patient information includes age, sex, symptoms, examination history, medication history, treatment history, or action history. After the treatment for the patient is completed and the second test result for the patient is obtained, a computer generates the first test result for the patient, the regional information, the patient information for the patient, and the second test result for the patient. Get the retraining data containing
The learning model generation method according to any one of claims 1 to 6, wherein a computer re-learns the learning model using the re-learning data. After the treatment for the patient is finished and the second test result for the patient is obtained, a computer stores the first test result for the patient, the regional information, and the patient information for the patient, and the second test result for the patient. Acquiring relearning data including treatment information representing the contents of the treatment that should have been given to the patient according to the
The learning model generation method according to any one of claims 1 to 6, wherein a computer re-learns the learning model using the re-learning data. obtaining a first test result of a first test performed on a patient at a medical facility;
Obtaining regional information including a second test result of a second test that is performed using a sample collected in a predetermined area including the medical facility and requires a longer test period than the first test;
obtaining patient information about the patient;
To a learning model that outputs treatment information representing the content of recommended treatment for a patient when the first test result, the region information, and the patient information are input, or an inspection prediction that predicts the second test result related to the patient A computer program for causing a computer to execute a process of inputting a first examination result, the region information and patient information and outputting treatment information or examination prediction. further causing a computer to execute a process of outputting treatment information according to said inspection prediction by referring to a storage unit storing treatment information according to a second inspection result when said learning model outputs said inspection prediction; 10. A computer program as claimed in claim 9, characterized by: further causing the computer to execute a process of outputting information for displaying an image including the obtained first examination result, the examination prediction output by the learning model, and the treatment information according to the examination prediction. 11. The computer program according to claim 10, wherein displaying an image including the acquired patient information, a second test result group of a second test performed using a specimen collected in the predetermined area, and the area information including a map showing the predetermined area; 12. The computer program according to any one of claims 9 to 11, further causing the computer to execute a process of outputting information for. a first acquisition unit that acquires a first test result of a first test performed on a patient at a medical facility;
a second acquisition unit that acquires area information including a second test result of a second test that is performed using a sample collected in a predetermined area including the medical facility and requires a longer test period than the first test;
a third acquisition unit that acquires patient information about the patient;
a learning model that, when a first test result, the region information, and the patient information are input, outputs treatment information representing details of recommended treatment for the patient or test prediction that predicts the second test result for the patient;
an output unit that inputs the obtained first test result, the regional information, and the patient information to the learning model and outputs treatment information or test prediction.

Images