JP2022037803A - Information processing method, diagnostic support device, and computer program - Google Patents

Abstract

To provide an information processing method capable of providing users with information that supports self-diagnosis of health conditions.SOLUTION: An information processing method includes: acquiring health-related information on a user recorded in a distributed database including a block chain; acquiring current self-physical condition information on the user; when health-related information and self-physical condition information are input, inputting the health-related information acquired from the distributed database and the self-physical condition information acquired from the user into a learning model that outputs a diagnosis result related to a user's health condition; thereby, examining the user's health condition; transmitting a diagnosis result of the user's health condition to the user's terminal device; and transmitting the diagnosis result of the user's health condition and the self-physical condition information to an external communication apparatus.SELECTED DRAWING: Figure 2

Description

The present invention relates to an information processing method, a diagnostic support device, and a computer program.

Patent Document 1 discloses an automatic diagnostic system that detects data from a patient using a medical device and performs an initial evaluation of the patient's health using a neural network.

Special Table 2006-507875 Gazette

However, Patent Document 1 describes a technique for comprehensively automatically diagnosing a user's health condition using the user's past to present health-related information recorded on the blockchain and the user's current self-physical condition information. Not disclosed.

An object of the present disclosure is to provide an information processing method, a diagnostic support device, and a computer program capable of providing a user with information for supporting a self-diagnosis of a health condition based on a user's health-related information and self-physical condition information.

In the information processing method according to the present disclosure, the user's health-related information recorded in a distributed database including a blockchain is acquired, the user's current self-physical condition information is acquired, and the health-related information and the self-physical condition information are obtained. When input, the health condition of the user is input by inputting the health-related information acquired from the distributed database and the self-physical condition information acquired from the user into the learning model that outputs the diagnosis result related to the health condition of the user. Is diagnosed, the diagnosis result of the user's health condition is transmitted to the user's terminal device, and the diagnosis result of the user's health condition and the self-physical condition information are transmitted to the external communication device.

The diagnostic support device according to the present disclosure includes a first acquisition unit that acquires user health-related information recorded in a distributed database including a blockchain, a second acquisition unit that acquires the user's current self-physical condition information, and the health. When related information and the self-physical condition information are input, a learning model that outputs a diagnosis result related to the user's health condition, the health-related information acquired from the distributed database, and the self-physical condition information acquired from the user are input. By doing so, the processing unit for diagnosing the user's health condition and the user's health condition diagnosis result are transmitted to the user's terminal device, and the user's health condition diagnosis result and the self-physical condition information are transmitted to the external communication device. It has a communication unit.

The computer program according to the present disclosure acquires the user's health-related information recorded in a distributed database including a blockchain, acquires the user's current self-physical condition information, and inputs the health-related information and the self-physical condition information. If so, the health status of the user is determined by inputting the health-related information acquired from the distributed database and the self-physical condition information acquired from the user into the learning model that outputs the diagnosis result related to the health status of the user. A computer is made to perform a process of diagnosing, transmitting the diagnosis result of the user's health condition to the user's terminal device, and transmitting the diagnosis result of the user's health condition and the self-physical condition information to an external communication device.

According to the present disclosure, an information processing method, a diagnostic support device, and a computer program capable of providing a user with information for supporting a self-diagnosis of a health condition based on a user's health-related information and self-physical condition information are provided.

It is a schematic diagram which shows the structural example of the diagnosis support system which concerns on Embodiment 1. FIG. It is a schematic diagram which shows the outline of the diagnosis support method (information processing method) which concerns on Embodiment 1. It is a block diagram which shows the structural example of the diagnostic support apparatus which concerns on Embodiment 1. FIG. It is a conceptual diagram which shows an example of the user basic information recorded in a distributed database. It is a conceptual diagram which shows an example of the health-related information recorded in a distributed database. It is a conceptual diagram which shows the structural example of the disease diagnosis learning model. It is a conceptual diagram which shows the structural example of the image diagnosis learning model. It is a conceptual diagram which shows an example of the disease basic information. It is a conceptual diagram which shows an example of hospital information. It is a conceptual diagram which shows an example of a pharmacy information. It is a flowchart which shows the collection processing procedure of the health-related information which concerns on Embodiment 1. It is a flowchart which shows the processing procedure of the diagnosis support which concerns on Embodiment 1. It is a flowchart which shows the processing procedure of the diagnosis support which concerns on Embodiment 1. It is a schematic diagram which shows an example of the diagnosis display screen which concerns on Embodiment 1. FIG. It is a schematic diagram which shows an example of the hospital search reservation screen which concerns on Embodiment 1. It is a block diagram which shows the structural example of the diagnostic support apparatus which concerns on Embodiment 2. It is a flowchart which shows the processing procedure about the collection of the infectious disease-related information and the warning notification which concerns on Embodiment 2. It is a block diagram which shows the structural example of the infectious disease diagnosis learning model. It is a flowchart which shows the processing procedure of the diagnosis support which concerns on Embodiment 2. It is a schematic diagram which shows an example of the diagnosis display screen which concerns on Embodiment 2. It is a block diagram which shows the structural example of the diagnostic support apparatus 1 which concerns on Embodiment 3. It is a block diagram which shows the structural example of the interview learning model which concerns on Embodiment 3. It is a block diagram which shows the structural example of the inquiry matter generation learning model which concerns on Embodiment 4. It is explanatory drawing which shows the generation method of the medical examination matter which concerns on Embodiment 4. It is a flowchart which shows the processing procedure of the diagnosis support which concerns on Embodiment 5. It is a conceptual diagram of the disease diagnosis learning model which concerns on Embodiment 6. It is a block diagram which shows the structural example of the disease diagnosis learning model which concerns on Embodiment 6.

Specific examples of the information processing method, the diagnostic support device, and the computer program according to the embodiment of the present invention will be described below with reference to the drawings. It should be noted that the present invention is not limited to these examples, and is indicated by the scope of claims, and is intended to include all modifications within the meaning and scope equivalent to the scope of claims. In addition, at least a part of the embodiments described below may be arbitrarily combined.

(Embodiment 1)
FIG. 1 is a schematic diagram showing a configuration example of the diagnosis support system according to the first embodiment. The diagnostic support system according to the first embodiment is a system for supporting self-diagnosis of a health condition by a user and promoting efficient digital transformation (DX: Digital transformation) of a medical system.

The diagnostic support system according to the first embodiment includes a diagnostic support device 1 and a plurality of information processing devices 2, and each device is connected by P2P (Peer to Peer). The P2P-connected diagnostic support device 1 and the plurality of information processing devices 2 constitute a distributed database 3. The distributed database 3 includes the blockchain 3a (see FIG. 2), and all or part of the information to be recorded, or the hash value of the information to be recorded can be recorded in the blockchain 3a. A user terminal device 4, a doctor terminal device (external communication device) 6, and the like are connected to the diagnosis support device 1 via a communication line. Further, the wearable terminal 5 worn by the user is wirelessly connected to the user terminal device 4.

FIG. 2 is a schematic diagram showing an outline of the diagnosis support method (information processing method) according to the first embodiment. Basic information of a plurality of users who use the diagnosis support system is registered in the distributed database 3. The user wears the wearable terminal 5, and various activity information of the user is uploaded to the diagnosis support device 1 via the user terminal device 4. The uploaded various activity information is associated with the user ID and recorded in the distributed database 3 as health-related information. In addition, the user's electronic medical record information and medical images recorded in the medical institution are also recorded in the distributed database 3 as health-related information after the user's authentication and permission are obtained.

The user can request the diagnosis support device 1 to support the self-diagnosis of the health condition by using the user terminal device 4. When the diagnosis support device 1 receives a request for self-diagnosis support, the diagnosis support device 1 acquires self-physical condition information indicating the user's current physical condition from the user terminal device 4 as an answer to a basic inquiry item. The diagnosis support device 1 reads out the user's health-related information recorded in the distributed database 3 and inputs the health-related information and the self-physical condition information into the disease diagnosis AI (disease diagnosis learning model 14 described later) to make a diagnosis. conduct. The diagnosis support device 1 transmits the diagnosis result to the user terminal device 4. Further, the diagnosis support device 1 performs image diagnosis by inputting a medical image into an image diagnosis AI (image diagnosis learning model 15 described later). The diagnosis support device 1 transmits the diagnosis result by the disease diagnosis AI and the image diagnosis AI to the doctor terminal device 6 as necessary, and acquires the diagnosis result by the doctor. Then, the diagnosis support device 1 transmits the diagnosis result of the doctor to the user terminal device 4 in addition to the diagnosis result of the disease diagnosis AI. The user can confirm the received diagnosis result and self-diagnose his / her health condition.
Further, the diagnosis support device 1 can support the search and reservation of a hospital based on the diagnosis result. The diagnosis support device 1 can also transmit the diagnosis result and the self-physical condition information to the reserved hospital. The doctor can efficiently examine the user by the information provided from the diagnosis support device 1.
Further, when the prescription is provided by the doctor, the diagnosis support device 1 can also transmit the prescription information to the user terminal device 4. The user can purchase a prescription drug based on the prescription information. Further, the diagnosis support device 1 may be configured to directly transmit the prescription information to the terminal device of the pharmacy. The prescription drug may be delivered to the user's home, or the user may receive it at the pharmacy.
Furthermore, the diagnosis support device 1 can transmit the diagnosis result and the self-physical condition information to various other medical institutions and medical personnel such as the healthcare team. The healthcare team or the like can provide various medical care services to efficient and high-quality users based on the information provided by the diagnostic support device 1.

FIG. 3 is a block diagram showing a configuration example of the diagnosis support device 1 according to the first embodiment. The diagnosis support device 1 is a computer including a processing unit 11, a storage unit 12, and a communication unit 13. The diagnostic support device 1 may be a multi-computer composed of a plurality of computers, or may be a virtual machine virtually constructed by software. Even if the various functions of the diagnostic support device 1 are configured to be realized by the cooperation of devices and programs such as a plurality of servers, a personal computer, a software program (including an AI of an external server and an application program to be downloaded to a smartphone). good. The method of configuring the diagnostic support system according to the first embodiment is not particularly limited.

The processing unit 11 includes one or more CPUs (Central Processing Units), MPUs (Micro-Processing Units), GPUs (Graphics Processing Units), GPGPUs (General-purpose computing on graphics processing units), TPUs (Tensor Processing Units), and the like. It has an arithmetic processing unit. The processing unit 11 functions as a diagnostic support device 1 constituting the distributed database 3 by reading and executing the computer program 12a stored in the storage unit 12, and performs processing to support the self-diagnosis of the health condition by the user. Execute.

The communication unit 13 includes a processing circuit, a communication circuit, and the like for performing processing related to communication, and other information constituting the user terminal device 4, the doctor terminal device 6, and the distributed database 3 connected to the diagnosis support device 1. Information is sent and received to and from the processing device 2.

The storage unit 12 is a storage device such as a hard disk, EEPROM (Electrically Erasable Programmable ROM), and a flash memory. The storage unit 12 stores a computer program 12a necessary for the processing unit 11 to execute a process for supporting the self-diagnosis of the health condition by the user. Further, the storage unit 12 constitutes a distributed database 3 and includes a database 12b for storing basic user information and health-related information of the user. Further, the storage unit 12 stores the disease diagnosis learning model 14, the image diagnosis learning model 15, the disease basic information 16, the hospital information 17, the pharmacy information 18, and the cognitive function test question 19.

The computer program 12a is a program for making the computer function as the diagnostic support device 1 of the distributed database 3. The computer program 12a causes the computer to execute a process of assisting the user in self-diagnosis of the health condition.

The computer program 12a may be recorded on the recording medium 10 so that it can be read by a computer. The storage unit 12 stores the computer program 12a read from the recording medium 10 by a reading device (not shown). The recording medium 10 is a semiconductor memory such as a flash memory, an optical disk, a magnetic disk, a magnetic disk disk, or the like. Further, the computer program 12a according to the present embodiment may be downloaded from an external server (not shown) connected to the communication network and stored in the storage unit 12.

FIG. 4 is a conceptual diagram showing an example of basic user information recorded in the distributed database 3. Basic user information includes, for example, user ID, user name, address, date of birth, family information, occupation, hobbies, sports, assets, passport information, medical examination card information, my number information, authentication information, insurance card information, and driver's license. Includes certificate information, payment information, etc. The family information includes, for example, a user ID of the family, relationship information indicating a relationship with the user, and the like. The authentication information includes biometric authentication information such as a user's face, fingerprint, handwriting, voice print, iris, and vein pattern, a password, and the like. The payment information includes information necessary for payment by credit card, debit card, electronic money, and virtual currency. The payment information is used for payment of the usage fee of the diagnosis support device 1 according to the first embodiment.

FIG. 5 is a conceptual diagram showing an example of health-related information recorded in the distributed database 3. The user's health-related information includes, for example, medical information, activity information, body composition information, constitution information, medical history information, drug information, physical condition information, lifestyle information, cognitive function test result information, infectious disease-related information, etc. include. The diagnosis support device 1 collects this information from the user terminal device 4 and the information processing device 2, associates it with the user ID, and records it in the distributed database 3. Since the amount of health-related information is enormous, it is advisable to record the hash value of the health-related information in the blockchain 3a of the distributed database 3. Health-related information is basically collected from the user terminal device 4. However, information such as electronic medical records and medical images may be configured to be collected directly from the information processing device 2 of the hospital. When the diagnosis support device 1 collects information from the information processing device 2, it is preferable to authenticate the information collection authority by the authentication process via the user terminal device 4.

The medical information includes a user's electronic medical record, a medical image, and the like stored in the information processing device 2 of the hospital. Electronic medical records and medical images are user's past medical information. Medical images include X-ray images, CT images, MRI images, PET images, bone scintigraphy images, ultrasonic images and the like. The user terminal device 4 acquires these information from the information processing device 2 of the hospital and provides the information to the diagnosis support device 1.

The activity information includes position information and date and time information indicating an activity state due to the movement of the user, for example, positioning information by GNSS (Global Navigation Satellite System). In addition, the activity information includes acceleration information indicating an activity state due to the movement of the user's limbs, heart rate information, body temperature information, blood pressure information, blood glucose level information, sleep depth information, and the like of the user. These activity information can be collected from the wearable terminal 5 including various sensors such as GPS, acceleration sensor, pressure pressure sensor, azimuth magnetic sensor, heart rate monitor sensor, optical sensor, etc. via the user terminal device 4. As the wearable terminal 5 device, any device such as a wristband type device, a ring type device, smart glasses, and a smart contact lens can be used. The user terminal device 4 acquires this information from the wearable terminal 5 and provides it to the diagnostic support device 1.

Information such as body composition includes information such as body composition such as bone density, body fat percentage, and muscle mass, height, weight, and basal metabolic rate. The user terminal device 4 acquires these information from the body composition measuring device capable of wireless communication with the user terminal device 4 and provides the information to the diagnosis support device 1.

Information such as constitution includes information on allergies, blood types, HLA types, genetic information, races, and the like.
The medical history information includes information such as the name of the disease that has been affected in the past, the hospitalization history, and the surgical history. The user terminal device 4 acquires these information input by the user operation and provides the information to the diagnosis support device 1.

The drug information includes information such as a prescription prepared by a doctor in the past, a history of taking prescription drugs, a history of taking over-the-counter drugs, and side effects when taking the drug. The user terminal device 4 acquires this information from the electronic medicine notebook and provides it to the diagnosis support device 1. The user terminal device 4 may acquire these information input by the user operation and provide the information to the diagnosis support device 1.

The physical condition information includes information such as a user's sleep state, appetite, fever, nausea, diarrhea, and abdominal pain. The user terminal device 4 acquires these information input by the user operation and provides the information to the diagnosis support device 1.

Lifestyle information includes information such as morning, noon, and night eating habits, drinking habits, smoking, and whether or not to drink coffee habitually. The user terminal device 4 acquires these information input by the user operation and provides the information to the diagnosis support device 1.

The cognitive function test result information includes a scoring result and a score history of a predetermined cognitive function test. The user terminal device 4 provides the cognitive function test question 19 to the user at an appropriate timing, scores based on the input answer, and obtains information on the scoring result (score) history of the cognitive function test question 19 as a diagnostic support device. Provide to 1.

Infectious disease-related information and the like include vaccination history, PCR test results, antibody test results, antigen test results, immune function tests and the like related to a predetermined infectious disease. The user terminal device 4 acquires these information input by the user operation and provides the information to the diagnosis support device 1.
The infectious disease-related information and the like may be configured to include a document indicating that the various test results are correct information, for example, a document issued by a testing institution. Although this information is not input to the disease diagnosis learning model 14, it is recorded in the distributed database 3 so that the user can appropriately download it from the diagnosis support device 1 as a certificate regarding infectious diseases as needed. It should be configured.

Further, although the health-related information shown in FIG. 5 is recorded in the distributed database 3, the retention period may be appropriately set in the health-related information. For example, the body temperature or the like may be set to a storage period of about one week or several weeks, and the processing unit 11 may be configured to be deleted from the distributed database 3 after the storage period has elapsed. At the discretion of the doctor, the storage period of the health-related information set in the diagnostic support device 1 may be extended. The diagnosis support device 1 stores each piece of health-related information in association with the storage period. The doctor can request the diagnosis support device 1 to change the storage period via the doctor terminal device 6. When the diagnosis support device 1 receives the change of the storage period from the doctor terminal device 6, the storage period is changed. If the storage period is erroneously shortened, the diagnostic support may be hindered. Therefore, it is preferable that the diagnostic support device 1 is configured to extend the storage period only. Of course, in some cases, the diagnostic support device 1 may be configured so that the storage period can be shortened, or the health-related information recorded in the distributed database 3 can be deleted at the discretion and operation of the doctor. You may.
Further, the diagnostic support device 1 may be configured to record the health-related information whose storage period has expired as archive information instead of completely deleting it.

<Disease diagnosis learning model 14>
FIG. 6 is a conceptual diagram showing a configuration example of the disease diagnosis learning model 14. It is a conceptual diagram which shows the structural example of the disease diagnosis learning model 14 which concerns on Embodiment 1. FIG. The disease diagnosis learning model 14 outputs diagnosis result information related to the user's health condition from the user's health-related information and the self-physical condition information indicating the user's current physical condition by machine learning or deep learning using teacher data in advance. It is a trained model trained in this way. The disease diagnosis learning model 14 performs a predetermined calculation on an input value and outputs the calculation result, and the storage unit 12 contains data such as a coefficient and a threshold of a function that defines this calculation in the disease diagnosis learning model. It is stored as 14. By reading the data stored as the disease diagnosis learning model 14, the processing unit 11 can execute arithmetic processing for diagnosing the health condition of the user.
In the first embodiment, the learning process of the disease diagnosis learning model 14 is performed by the diagnosis support device 1. However, the learning process may be performed by a device other than the diagnosis support device 1. The data related to the disease diagnosis learning model 14 learned in this case may be provided in the form of distribution via the communication network as in the computer program 12a, or may be provided in the form recorded on the recording medium 10. May be good.

In the first embodiment, the disease diagnosis learning model 14 includes, for example, an input layer 14a that receives user's health-related information and self-physical condition information, an intermediate layer 14b that extracts features of health-related information and self-physical condition information, and feature extraction. It is a neural network having an output layer 14c that outputs the diagnosis result information related to the health condition of the user derived by.

The input layer 14a of the neural network includes electronic charts, activity information, body composition information, constitutional information, history information, drug information, physical condition information, lifestyle information, cognitive function test result information, and infectious diseases, which are health-related information. It has a plurality of neurons to which related information and the like are input, and each input data is passed to the intermediate layer 14b.
The information input to the input layer 14a is non-image data. The health-related information is information related to the user's health recorded in the distributed database 3 from the past to the present, but the influence of the past state on the present health state varies. Therefore, it is advisable to perform preprocessing for giving weights of less than 1 to the past information according to the time difference between the present and the time when the information is collected. In addition, time-series information such as user activity information is converted into numerical values representing user activity information such as daily average walking distance, number of steps, degree of exercise (weak, medium, strong), hypertension, and hypotension. It is advisable to perform preprocessing to reduce the amount of information.

The intermediate layer 14b has a plurality of layers composed of a plurality of neurons. Each layer extracts the feature amount related to the user's health condition from the input data and passes it from the front stage to the rear stage in order, and the last layer transfers the output feature amount to the disease classification code and the point of examination. Data relating to no, urgency, and necessity of medicine are passed to the output layer 14c. The disease classification code is, for example, the International Classification of Diseases (ICD).

The output layer 14c includes a plurality of neurons corresponding to the plurality of disease classification codes, and each neuron outputs a certainty level corresponding to each disease.
Further, the output layer 14c includes a neuron indicating the necessity of a medical examination, and the neuron outputs a certainty degree that the medical examination is necessary.
Further, the output layer 14c includes a neuron indicating an urgent degree, and the neuron outputs an urgent degree of certainty.
Furthermore, the output layer 14c comprises a neuron indicating the need for the drug, which outputs the certainty that the drug should be taken.
Each of the above-mentioned convictions is represented by, for example, a decimal number of 0 or more and 1 or less, and the larger the value, the more certain the output result. Further, the contents of each neuron and output data described above are examples, and are not particularly limited.

Although the example in which the disease diagnosis learning model 14 is configured by one learning model has been described in the first embodiment, the disease diagnosis learning model 14 may be configured by a plurality of learning models.

Further, when the disease diagnosis learning model 14 is composed of a plurality of learning models, some learning models may be provided in an external computer.

Although the example in which the disease diagnosis learning model 14 is a neural network has been described in the first embodiment, it may be a model having a configuration such as an SVM (Support Vector Machine), a Bayesian network, or a regression tree.

<Learning method of disease diagnosis learning model 14>
First, the processing unit 11 collects health-related information, self-physical condition information, and a doctor's diagnosis result, which are the sources of teacher data, from the distributed database 3. The doctor's diagnosis result is obtained from the medical record. Then, the processing unit 11 assigns the disease classification code based on the diagnosis result of the doctor, the necessity of medical examination, the degree of urgency, and the teacher data indicating the necessity of medicine to a plurality of health-related information and self-physical condition information. Generate training data by. Next, the processing unit 11 generates the disease diagnosis learning model 14 by machine learning or deep learning the pre-learning neural network model using the generated learning data.
Specifically, the processing unit 11 inputs health-related information and self-physical condition information included in the learning data into the neural network model before learning, undergoes arithmetic processing in the intermediate layer 14b, and is output from the output layer 14c. Acquire diagnosis result information. Then, the processing unit 11 compares the diagnosis result information output from the output layer 14c with the diagnosis result information indicated by the teacher data, and in the intermediate layer 14b so that the data output from the output layer 14c approaches the correct answer value. Optimize the parameters used for the arithmetic processing of. The parameter is, for example, a weight (coupling coefficient) between neurons. The method of optimizing the parameters is not particularly limited, but for example, the processing unit 11 optimizes various parameters by using the steepest descent method or the like.

The information processing apparatus 2 repeatedly performs the above processing based on the teacher data of a large number of patients included in the teacher data, thereby obtaining the learned disease diagnosis learning model 14.

When the disease diagnosis learning model 14 is configured by a plurality of learning models, the plurality of learning models may be generated as follows. The learning data generated as described above is classified into a plurality of classes by a clustering process. For example, it is classified into learning data on circulatory system diseases, learning data on respiratory diseases, learning data on food poisoning, and the like. The clustering process may be performed entirely or partially manually. Then, the neural network is trained using the learning data classified into the first class, and the first learning model is generated. Using the training data classified in the second class, the neural network is trained and the second learning model is generated. In the same way, generate multiple learning models. Then, the disease diagnosis learning model 14 is generated by connecting a plurality of learning models. A plurality of learning models may be simply connected in parallel, and the same health-related information and self-physical condition information may be input to each learning model. When health-related information and self-physical condition information are input to multiple learning models, diagnosis result information is output from each learning model, but diagnosis result information with high certainty is output from the learning model that is suitable for the user's condition. Therefore, diagnosis result information with low certainty is output from an inappropriate learning model. Therefore, the diagnosis support device 1 naturally selects a learning model suitable for the user's condition by selecting the diagnosis result information having a high degree of certainty, and the diagnosis result information can be obtained.

<Image diagnosis learning model 15>
FIG. 7 is a conceptual diagram showing a configuration example of the diagnostic imaging learning model 15. In the first embodiment, the diagnostic imaging learning model 15 has, for example, an input layer 15a for receiving input of a medical image, an intermediate layer 15b for extracting features of a medical image, and an output layer 15c for outputting a diagnostic imaging result. It is a neural network. The diagnostic imaging learning model 15 of the first embodiment is, for example, a CNN (Convolution Neural Network). The diagnosis support device 1 generates an image diagnosis learning model 15 by performing deep learning on the CNN model to learn the relationship between the feature amount of the medical image and the disease classification code.

The input layer 15a of the neural network has a plurality of neurons that receive the input of the pixel value of each pixel related to the medical image, and passes the input pixel value to the intermediate layer 15b. The intermediate layer 15b has a plurality of neurons for extracting the feature amount of the medical image, and the extracted feature amount is passed to the output layer 15c. For example, when the image diagnosis learning model 15 is CNN, the intermediate layer 15b is a convolution layer that convolves the pixel values of each pixel input from the input layer 15a and a pooling layer that maps the pixel values convoluted by the convolution layer. And are alternately connected to each other, and the feature amount of the medical image is finally extracted while compressing the pixel information of the medical image. The output layer 15c has a plurality of neurons that output disease confidence for each of the plurality of disease classification codes.

In the first embodiment, it is assumed that the diagnostic imaging learning model 15 is a CNN, but the configuration of the model is not limited to the CNN. The diagnostic imaging learning model 15 may be, for example, a diagnostic imaging learning model 15 having a configuration such as a neural network other than CNN, an SVM, a Bayesian network, or a regression tree.

In the learning process of the image diagnosis learning model 15, the diagnosis support device 1 uses learning data in which the medical image stored in the distributed database 3 and the diagnosis result of the doctor for the medical image are associated with each other.

The diagnosis support device 1 inputs a plurality of medical images, which are learning data, to the input layer 15a of the CNN, performs arithmetic processing on the intermediate layer 15b, and acquires a diagnosis result output from the output layer 15c.

The diagnostic support device 1 compares the diagnostic result output from the output layer 15c with the diagnostic result labeled on the medical image of the learning data, and is intermediate so that the output value from the output layer 15c approaches the correct answer value. The parameters used for the arithmetic processing on the layer 15b are optimized. The parameter is, for example, a weight (coupling coefficient) between neurons. The method of optimizing the parameters is not particularly limited, but for example, the diagnostic support device 1 optimizes various parameters by using the steepest descent method or the like.

The diagnosis support device 1 obtains a trained image diagnosis learning model 15 by repeating the above processing for each medical image included in the learning data.

Although an example in which the image diagnosis learning model 15 is configured by one learning model has been described, a plurality of learning models corresponding to the types of medical images such as a user's image pickup site, CT image, PET image, etc. included in the medical image may be used. In combination, the diagnostic imaging learning model 15 may be configured. The structure and generation method of each learning model are the same as the structure and generation method of the diagnostic imaging learning model 15 described above.

FIG. 8 is a conceptual diagram showing an example of basic disease information 16. The storage unit 12 has, for each of the plurality of disease classification codes, a disease classification code, a heading of the classification to which the code belongs, a disease name corresponding to the code, an inquiry item for diagnosing the disease, and medical treatment suitable for the treatment of the disease. The department and the explanation document for the user are associated and stored. The user explanation document is a document that explains information such as the cause, symptom, and treatment method of the disease so that the general public can easily understand it. In addition, the user explanatory document may include typical medical information and advice to the user. For example, medical information on diabetes may include a list of stages of diabetes, subjective symptoms, test contents, and treatment methods.
The diagnosis support device 1 can ask the user a question as needed by referring to the question items of the basic disease information 16, and can improve the diagnosis accuracy. Further, the diagnosis support device 1 can provide hospital information according to the content of the user's illness by referring to the clinical department of the basic illness information 16. Further, the diagnosis support device 1 can provide the user with a document explaining the disease related to the diagnosis result by the user explanation document of the disease basic information 16.

FIG. 9 is a conceptual diagram showing an example of hospital information 17. The storage unit 12 supports passwords, hospital names, addresses, clinical departments, doctor profile information, consultation hours, contact information, and communication addresses for transmitting information to hospital terminal devices for each of a plurality of hospital IDs. I remember it. By referring to the hospital information 17, the diagnosis support device 1 can introduce a hospital suitable for the diagnosis result to the user and execute the reservation process. Further, the diagnosis support device 1 can transmit information such as a diagnosis result to the hospital by using the communication address.

FIG. 10 is a conceptual diagram showing an example of pharmacy information 18. The storage unit 12 stores the password, the address, the profile information of the pharmacist, the business hours, the contact information, and the communication address for transmitting the information to the terminal device of the pharmacy in association with each other for each of the plurality of pharmacy IDs. The diagnosis support device 1 can introduce a pharmacy to a user by referring to the pharmacy information 18. Further, the diagnosis support device 1 can transmit the prescription data to the pharmacy by using the communication address.

FIG. 11 is a flowchart showing a procedure for collecting and processing health-related information according to the first embodiment. The processing unit 11 of the diagnosis support device 1 collects medical information from the user terminal device 4 or the information processing device 2 of the medical institution (step S111).

The user terminal device 4 periodically transmits activity information and body composition information to the diagnosis support device 1. The processing unit 11 of the diagnosis support device 1 collects the activity information and the body composition information transmitted from the user terminal device 4 (step S112, step S113).

Further, the user terminal device 4 periodically or each time the information is updated, transmits information such as constitution, medical history information, drug information, physical condition information, lifestyle information, and infectious disease information to the diagnosis support device 1. The processing unit 11 of the diagnosis support device 1 collects these information transmitted from the user terminal device 4 (steps S114 to S119).
The physical condition information collected here is similar to the self-physical condition information, but is information indicating the physical condition of the user at the time of diagnosis by the disease diagnosis learning model 14, whereas the physical condition information is the disease. It differs in that it is past information collected before the time of diagnosis by the diagnostic learning model 14.
Further, when the diagnosis is made by the disease diagnosis learning model 14 and then the treatment is performed by the doctor, the information on the follow-up such as the physical condition of the user and the side effects due to the ingestion of the prescription drug should be collected as the physical condition information in step S117. It may be configured in. As information related to follow-up, examinations and treatments at the time of initial or re-examination, examinations by hospitalization, surgery, other treatments, etc. are recorded as medical history.

Further, the user terminal device 4 periodically executes a cognitive function test based on the cognitive function test question 19, scores the user's answer, and transmits the cognitive function test result to the diagnostic support device 1. The processing unit 11 of the diagnosis support device 1 collects the cognitive function test results transmitted from the user terminal device 4 (step S120).

Next, the processing unit 11 of the diagnosis support device 1 that has collected the health-related information records each collected information in the distributed database 3 in association with the user ID (step S121). When recording each information in the distributed database 3, the processing unit 11 records the hash value of each information in the blockchain 3a.

The timing of collecting the above-mentioned various health-related information may be appropriately determined according to the type of information, and is not particularly limited.

12 and 13 are flowcharts showing the processing procedure of the diagnostic support according to the first embodiment. When the diagnosis support device 1 receives a request for self-diagnosis support from the user terminal device 4, the diagnosis support device 1 executes the following processing. First, the processing unit 11 of the diagnosis support device 1 authenticates the user using the biometric authentication information of the user (step S131).

If the authentication is successful, the processing unit 11 acquires health-related information from the distributed database 3 using the user ID of the authenticated user as a key (step S132). The processing unit 11 confirms whether or not the health-related information has been falsified or changed as necessary.

Next, the processing unit 11 transmits a predetermined basic inquiry item to the user terminal device 4, and acquires self-physical condition information indicating the user's current physical condition via the user terminal device 4 (step S133). The question is standard, and the user terminal device 4 displays, for example, the content of asking basic information such as body temperature, appetite, nausea, headache, and the presence or absence of dizziness. The user inputs the answer to the medical examination item into the user terminal device 4 as self-physical condition information. The user terminal device 4 transmits the input self-physical condition information to the diagnostic support device 1, and the diagnostic support device 1 acquires the self-physical condition information transmitted from the user terminal device 4.

Then, the processing unit 11 executes the diagnosis regarding the health condition of the user by inputting the acquired health-related information and the self-physical condition information into the disease diagnosis learning model 14 (step S134).

Next, the processing unit 11 determines whether or not an additional inquiry is necessary based on the diagnosis result (step S135). For example, when there are a plurality of diseases whose conviction is less than a predetermined value and there are additional medical examination items that can improve the conviction, the diagnostic support device 1 determines that an additional medical examination is necessary. Specifically, the diagnosis support device 1 may refer to the disease basic information 16 to read out the inquiry items related to each disease and determine the presence or absence of unacquired information.

When it is determined that an additional medical examination is required (step S135: YES), the processing unit 11 transmits additional medical examination items to the user terminal device 4, acquires additional self-physical condition information (step S136), and processes the process to step S134. return.

Even if the additional interviews are repeated, the conviction may not exceed the predetermined value. Therefore, it is advisable to limit the number of additional interviews to the predetermined number or less. Even if it is determined in step S135 that an additional medical examination is required, if the number of additional medical examinations has reached a predetermined number, the process may proceed to step S137.

When it is determined that the additional inquiry is unnecessary (step S135: NO), the processing unit 11 executes the diagnosis regarding the health condition of the user by inputting the medical image into the image diagnosis learning model 15 (step S137).

Next, the processing unit 11 receives the user's instruction and determines whether or not a diagnosis by a doctor is necessary (step S138). The user can select whether or not to request a diagnosis by a doctor via the user terminal device 4, and the user terminal device 4 transmits the selection result to the diagnosis support device 1. Further, if the certainty of the diagnosis result executed in step S134 is less than a predetermined value, it may be configured to automatically determine that a diagnosis by a doctor is necessary. It can also be configured to perform a doctor's diagnosis at all times.

When it is determined that a diagnosis by a doctor is necessary (step S138: YES), the processing unit 11 transfers the diagnosis results of steps S134 and S137, the self-physical condition information acquired in steps S133 and S136, and the like to the doctor terminal device 6. Transmit (step S139). The doctor receives the diagnosis result, the self-physical condition information corresponding to the interview result, etc. at the doctor terminal device 6, makes a diagnosis, and transmits the diagnosis result to the diagnosis support device 1 via the doctor terminal device 6. The doctor can also create a prescription and send the prescription information together with the diagnosis result to the diagnosis support device 1.
The doctor basically makes a diagnosis based on the diagnosis result by the disease diagnosis learning model 14, the diagnosis result by the image diagnosis learning model 15, and the self-physical condition information of the user. , Zoom and other video conferencing systems, telephone conferencing systems, real-time communication systems such as chat, e-mail, and telephone may be used to appropriately communicate and make a diagnosis. The communication between the doctor and the user may be performed by direct communication between the user terminal device 4 and the doctor terminal device 6, or may be performed by communication via the diagnosis support device 1.
Further, the doctor may access the diagnosis support device 1 via the doctor terminal device 6 to acquire necessary user health-related information and perform a diagnosis.
Further, the doctor can change the health-related information and the self-physical condition information input to the disease diagnosis learning model 14 via the doctor terminal device 6 so as to request the implementation of the re-diagnosis by the disease diagnosis learning model 14. May be good. The diagnosis support device 1 receives the re-diagnosis execution request transmitted from the doctor terminal device 6. The implementation request includes changes in health-related information and self-physical condition information to be input to the disease diagnosis learning model 14. The changes include, for example, information that is not used for rediagnosis and information that specifies information to be used. The diagnosis support device 1 changes the health-related information and the self-physical condition information to be input to the disease diagnosis learning model 14 according to the changed contents, inputs the changed information to the disease diagnosis learning model 14, and inputs the changed information to the disease diagnosis learning model 14. The diagnosis result information output from is transmitted to the doctor terminal device 6.
Of the health-related information and self-physical condition information, regarding binary information such as smoking or drinking, if it is not used for re-diagnosis, it is advisable to change this information to "none". Of the health-related information and self-physical condition information, non-binary information such as body temperature and blood pressure may be changed to a predetermined average normal value when not used for re-diagnosis. If the body temperature is 36.5 degrees of normal temperature, if the blood pressure is systolic blood pressure 125 mmHg, diastolic blood pressure 82 mmHg, etc., the average normal values are changed and input to the disease diagnosis learning model 14.

The diagnosis support device 1 acquires the diagnosis result transmitted from the doctor terminal device 6 (step S140). When the process of step S140 is completed, or when it is determined that the diagnosis by a doctor is unnecessary (step S138: NO), the diagnosis support device 1 transmits the diagnosis result to the user terminal device 4 (step S141). Specifically, the diagnosis support device 1 transmits data for displaying the diagnosis display screen 90 to the user terminal device 4.
When the diagnosis by the doctor is not made, the diagnosis support device 1 transmits the diagnosis result by the disease diagnosis learning model 14 to the user terminal device 4. When the diagnosis is made by the doctor, the diagnosis support device 1 transmits the diagnosis result by the doctor together with the diagnosis result by the disease diagnosis learning model 14 to the user terminal device 4. Further, when there is prescription information, the diagnosis support device 1 transmits the prescription information together with the diagnosis result of the doctor to the user terminal device 4.
The diagnosis support device 1 may be configured to transmit the diagnosis result and the self-physical condition information to various medical institutions related to the health management of the user, medical personnel such as the healthcare team, and the like. Medical personnel such as various medical institutions and healthcare teams receive diagnosis results and self-physical condition information, and based on the received diagnosis results and self-physical condition information, various medical care services that are efficient and high-quality for users. Can be provided.

FIG. 14 is a schematic diagram showing an example of the diagnostic display screen 90 according to the first embodiment. The diagnosis display screen 90 includes an authentication unit 90a, a user basic information display unit 90b, an action history display unit 90c, a lifestyle display unit 90d, a physical condition / cognitive function / infectious disease display unit 90e, a medical history display unit 90f, and an electronic medical record display unit. 90 g, AI diagnosis result display unit 90h, doctor diagnosis result display unit 90i, prescription display unit 90j are included.
By operating the authentication unit 90a, the user can authenticate the diagnostic support device 1 with face authentication, fingerprint authentication, voiceprint authentication, handwriting authentication, and other passwords.
The user can confirm the diagnosis result by the disease diagnosis learning model 14 by the AI diagnosis result display unit 90h.
The user can confirm the diagnosis result by the doctor by the doctor diagnosis result display unit 90i.
The user can confirm the presence or absence of a prescription by the prescription display unit 90j. By operating the prescription display unit 90j, the user can execute a process of transmitting prescription information to the pharmacy.

Returning to FIG. 13, the processing after step S142 will be described. The user can select whether or not to make a medical examination reservation via the user terminal device 4, and the user terminal device 4 transmits the selection result to the diagnosis support device 1. The processing unit 11 that has completed the processing in step S141 accepts the selection result of the user via the user terminal device 4 and determines whether or not to make a medical examination reservation (step S142).

When it is determined that the medical examination appointment is to be made (step S142: YES), the processing unit 11 executes the search process of the medical institution (step S143) and executes the appointment process (step S144). The diagnosis support device 1 transmits data for displaying the hospital search reservation screen 91 to the user terminal device 4, and supports the hospital search and reservation.

FIG. 15 is a schematic diagram showing an example of the hospital search reservation screen 91 according to the first embodiment. The hospital search reservation screen 91 includes a name search unit 91a, a nearest hospital search unit 91b, a clinical department search unit 91c, a recommended medical institution display unit 91d, and a manual search result display unit 91e.
The user can search for a hospital by inputting a keyword in the name search unit 91a. Further, by operating the icon displayed on the search unit 91b such as the nearest hospital, it is possible to search for the nearest hospital, for example, or to search for a hospital by address or station name. You can also search for hospitals that are registered as favorites and your family's hospital.
The user terminal device 4 transmits the search operation content of the user to the diagnosis support device 1. The diagnosis support device 1 searches for hospital information 17 according to the inspection operation content and transmits it to the terminal device of the hit hospital. The user terminal device 4 displays a list of searched hospitals on the manual search result display unit 91e.
Further, the diagnosis support device 1 searches the hospital information 17 for a recommended hospital according to the disease diagnosis learning model 14 or the diagnosis result by the doctor. For example, the diagnosis support device 1 refers to the basic disease information 16, identifies the clinical department associated with the diagnosed disease classification code, and searches for the hospital including the clinical department from the hospital information 17. Further, the diagnosis support device 1 acquires the position information of the user terminal device 4, searches for the hospital information 17 based on the clinical department and the location information, and obtains the information of the nearest hospital having the corresponding clinical department in the user terminal device 4. Send to. The user terminal device 4 displays a list of searched hospitals on the recommended medical institution display unit 91d.

A medical appointment icon 91f is displayed on the side of each of the plurality of hospital names displayed in the list. The user can make a medical appointment for the corresponding hospital by operating the medical appointment icon 91f. The user terminal device 4 transmits an instruction to start the diagnosis reservation process to the diagnosis support device 1. When the diagnosis support device 1 receives the instruction to start the diagnosis reservation process from the user terminal device 4, the diagnosis support device 1 executes the diagnosis reservation process. Specifically, the diagnosis support device 1 accepts selection of outpatient or online medical examination, selection of initial medical examination and re-examination, desired date and time of medical examination, selection of clinical department, etc. via the user terminal device 4. The diagnosis support device 1 transmits the received hospital reservation contents to the terminal device of the reserved hospital, and confirms the reservation.

Further, the processing unit 11 that has executed the reservation process in the process of step S144 specifies the communication address of the hospital for which the medical examination is reserved with reference to the hospital information 17, and provides the diagnosis result and the self-physical condition information (step S145).

Even if the user's self-physical condition information is received here, a certain amount of time may have passed by the time the user actually visits the medical examination. Therefore, the doctor may be configured to access the diagnosis support device 1 via the doctor terminal device 6 immediately before the medical examination and acquire the latest health-related information of the user.
Further, as in step S139, the doctor can change the health-related information and the self-physical condition information input to the disease diagnosis learning model 14 via the doctor terminal device 6 and request the execution of re-diagnosis by the disease diagnosis learning model 14. It may be configured as follows. Since the processing content of the diagnosis support device 1 is the same as that in step S139, the details will be omitted.

The diagnosis support device 1 may provide an online medical care service between the hospital terminal and the user terminal device 4. Further, the diagnosis support device 1 may be configured to provide the diagnosis result and the self-physical condition information obtained by the diagnosis support method according to the first embodiment to an external application that provides a hospital reservation and an online medical treatment service. ..

When it is determined that the appointment for medical examination is unnecessary (step S142: NO), or when the process of step S145 is completed, the processing unit 11 determines whether or not the prescription information is provided by the doctor (step S146). When it is determined that there is prescription information (step S146: YES), the processing unit 11 receives the user's instruction and determines whether or not to transmit the prescription information to the pharmacy (step S147). When it is determined to transmit the prescription information (step S147: YES), the processing unit 11 executes the pharmacy search process (step S148) and transmits the prescription information to the terminal device of the selected pharmacy (step S149). If the transmission to the pharmacy is not specified, the diagnosis support device 1 transmits the prescription information to the user terminal device 4 in association with the user ID.

When it is determined that there is no prescription information (step S146: NO), when it is determined that the prescription information is not transmitted (step S147: NO), or when the processing of step S149 is completed, the processing unit 11 is a disease diagnosis learning model. Information such as the diagnosis result by the image diagnosis learning model 15 and the diagnosis result by the doctor and the prescription information is recorded in the distributed database 3 (step S150), and the process is completed.

The cost related to the use of the diagnostic support device 1 according to the first embodiment may be configured so that it can be settled using a virtual currency such as a credit card, a debit card, electronic money, or bit coin. Further, the setting or monetization method of the diagnosis fee using the disease diagnosis learning model 14, the diagnosis fee by the doctor, the hospital reservation fee, the prescription preparation fee, the prescription drug cost, and the prescription drug delivery cost is not particularly limited. ..

As described above, according to the diagnosis support method according to the first embodiment, the user's health condition is diagnosed based on the past to present health-related information and the current self-physical condition information collected from the user, and the diagnosis result is obtained by the user. It can be transmitted to the terminal device 4. The user can self-diagnose his / her health condition by referring to the diagnosis result. In addition, the user can determine whether or not to receive a medical examination at the hospital by referring to the diagnosis result.

Further, the diagnosis support device 1 provides the diagnosis result and the self-physical condition information to the doctor, acquires the diagnosis result of the doctor, and obtains the diagnosis result based on the disease diagnosis learning model 14 and the diagnosis result of the doctor in the user terminal device 4. Can be sent to. The user can self-diagnose his / her health condition by referring to the diagnosis result by the arithmetic processing and the diagnosis result of the doctor based on the health-related information. In addition, the user can determine whether or not to receive a medical examination at the hospital by referring to the diagnosis result.

Further, when the diagnosis support device 1 acquires the prescription created by the doctor together with the diagnosis result of the doctor, the diagnosis result and the prescription information can be transmitted to the user terminal device 4. The user can purchase a prescription drug by using the prescription information. Specifically, the user can send the prescription information to the pharmacy and receive the prescription drug at the pharmacy. You can also receive prescription drugs by delivery. In addition, when delivering the prescription drug, when delivering from the pharmacy to the user, the delivery state of the prescription drug may be recorded in the distributed database 3.

Furthermore, the diagnosis support device 1 can improve the diagnosis accuracy by providing the doctor with the diagnosis result by the image diagnosis learning model 15 based on the medical image. In addition, the efficiency of diagnosis by a doctor can be improved.

Furthermore, additional interviews can be made to the user according to the diagnosis result by the disease diagnosis learning model 14, and the diagnosis accuracy can be improved.

In the first embodiment, an example in which the diagnosis is performed by inputting the user's health-related information and the self-physical condition information into the disease diagnosis learning model 14 has been described, but the user's health is also described without using the self-physical condition information. Diagnosis may be made by inputting relevant information into the disease diagnosis learning model 14. Both the diagnosis result in which the self-physical condition information mainly based on the user's subjectivity is taken into consideration and the diagnostic result in which the self-physical condition information is not taken into consideration are transmitted to the user terminal device 4. The user can judge his / her health condition more objectively and from various angles based on the diagnosis results of both.

(Embodiment 2)
The diagnostic support system according to the second embodiment is different from the first embodiment in that it can diagnose a predetermined infectious disease risk and aggravation risk. Since the other configurations of the diagnosis support system are the same as those of the diagnosis support system according to the first embodiment, the same reference numerals are given to the same parts, and detailed description thereof will be omitted.

FIG. 16 is a block diagram showing a configuration example of the diagnosis support device 1 according to the second embodiment. The diagnosis support device 1 according to the second embodiment includes the same processing unit 11, storage unit 12, and communication unit 13 as in the first embodiment. The storage unit 12 further stores the infectious disease diagnosis learning model 20.

The diagnostic support device 1 according to the second embodiment has behavior history information including the positions, dates and times of a plurality of users, and an infection contact indicating whether or not each user has or may have a predetermined infectious disease. The information is recorded in the distributed database 3. The predetermined infectious disease is, for example, a new type coronavirus infectious disease.

Infectious contact information includes close contact information indicating that the user has made close contact with a predetermined infectious disease patient. The user terminal device 4 includes an application for detecting close contact with a predetermined infectious disease patient, for example, COCOA. When the user is notified that there is a dense contact, the user terminal device 4 transmits the dense contact information indicating that the dense contact has occurred to the diagnosis support device 1. The diagnosis support device 1 records the rich contact information transmitted from the user terminal device 4 in the distributed database 3 in association with the user ID.

The infected contact information includes follow-up information indicating that the user used the same facility at the same date and time as the predetermined infectious disease patient. When the user receives the warning information that the facility in which the cluster occurred is used at the date and time when the cluster of the predetermined infectious disease patient is presumed to have occurred, the user terminal device 4 has the same date and time as the predetermined infectious disease patient. The tracking information indicating that the same facility was used is transmitted to the diagnosis support device 1. The diagnosis support device 1 records the tracking information transmitted from the user terminal device 4 in the distributed database 3 in association with the user ID.

The infectious disease information includes infection declaration information indicating that the user has a predetermined infectious disease. When the user suffers from a predetermined infectious disease, the user terminal device 4 transmits the infection report information indicating that the user has suffered from the predetermined infectious disease to the diagnosis support device 1. The diagnosis support device 1 records the infection report information transmitted from the user terminal device 4 in the distributed database 3 in association with the user ID.

The infectious disease information includes quasi-dense contact information indicating that the diagnostic support device 1 has determined that the patient has made close contact with a predetermined infectious disease patient or a person at risk of the predetermined infectious disease. The diagnosis support device 1 searches for a user who has been within a predetermined distance from a predetermined infectious disease patient for a predetermined time or longer based on the position information of each of a plurality of users, and searches for a user who has been within a predetermined distance from the predetermined infectious disease patient for a predetermined time or longer. The semi-rich contact information is recorded in the distributed database 3 in association with the user ID of the user who was within the range.

FIG. 17 is a flowchart showing a processing procedure for collecting infectious disease-related information and warning notification according to the second embodiment. The processing unit 11 of the diagnosis support device 1 periodically collects the position information and the date and time information of each of a plurality of users and records them in the distributed database 3 (step S211). Further, the processing unit 11 of the diagnosis support device 1 periodically receives the rich contact information from the user (step S212). The processing unit 11 records the received rich contact information as infectious disease information in the distributed database 3 in association with the user ID of the user who is the source of the rich contact information (step S213).

In addition, the processing unit 11 periodically receives the alert information from the user (step S214). The processing unit 11 records the received alert information as infectious disease information in the distributed database 3 in association with the user ID of the user who is the sender of the alert information (step S215).

Further, the processing unit 11 periodically receives the infection report information from the user (step S216). The processing unit 11 records the received infection report information as infectious disease information in the distributed database 3 in association with the user ID of the user who is the source of the infectious disease information (step S217).

Further, the processing unit 11 acquires and collects infection information of a person (non-user) suffering from a predetermined infectious disease, specifically, location information and date and time information of the non-user from a communication terminal device such as a health center (step S218). ..

The processing unit 11 determines the presence or absence of quasi-dense contact based on the position information and date and time information of each user and non-user (step S219). Specifically, the processing unit 11 is a user of one of the diagnostic support systems according to the first embodiment, and has location information and date and time information of a user who has received rich contact information or alert information, and other information. Based on the user's location information and date and time information, it is determined whether or not the user has been with the one user for a predetermined time or more and within a predetermined distance during the past predetermined period. The user determines that there is a risk of a predetermined infectious disease.
Similarly, the processing unit 11 is one user of the diagnosis support system according to the first embodiment, and includes the location information and the date and time information of the user who has transmitted the infection report information, and the location information and the date and time information of another user. Based on, it is determined whether or not the user has been within a predetermined distance for a predetermined time or more during the past predetermined period, and if the condition is satisfied, the other user is at risk of a predetermined infectious disease. It is determined that there is.
Similarly, the processing unit 11 has been notified by the health center or the like for the past predetermined period based on the location information and date and time information of the non-user suffering from the predetermined infectious disease and the location information and date and time information of other users. Then, it is determined whether or not the user has been within a predetermined distance for a predetermined time or more with the one user, and if the condition is satisfied, the other user is determined to be at risk of a predetermined infectious disease. When the predetermined infection is a new type coronavirus infection, the predetermined period is 14 days, the predetermined time is 15 minutes, and the predetermined distance is, for example, 1 m. When the position accuracy is 1 m or more, it may be determined whether or not another user has approached one user suffering from a predetermined infectious disease with the minimum resolution accuracy.
According to the above determination process, it is possible to detect an infection risk that cannot be grasped by the application. The risk of infection due to contact with a user who has received rich contact information or alert information is different from the risk of infection due to contact with the user who sent the infection report information or a non-user who has suffered from a predetermined infectious disease. The dense contact may be leveled.

When it is determined that there is a quasi-dense contact in the above process, the processing unit 11 records the quasi-dense contact information as infectious disease information in the distributed database 3 in association with the user ID of another user (step S220). ..

Next, when the user corresponding to the semi-dense contact is identified, the processing unit 11 notifies the health center and other medical institutions of the information of the user at risk of infectious disease as warning information (step S221). It is possible to notify the health center and other medical institutions of the information of the user who is at risk of infectious disease detected by the diagnostic support device 1 according to the second embodiment, which is not detected by a specific application. The content of the information to be notified may be set as appropriate.

FIG. 18 is a block diagram showing a configuration example of the infectious disease diagnosis learning model 20. The basic configuration of the infectious disease diagnosis learning model 20 is the same as that of the disease diagnosis learning model 14, and includes an input layer 20a, an intermediate layer 20b, and an output layer 20c. Infection contact information is further input as input information in the input layer 20a of the infectious disease diagnosis learning model 20. The infectious disease diagnosis learning model 20 is learned to output information indicating the infection risk for a predetermined infectious disease and the aggravation risk due to infection when health-related information including infection contact information and self-physical condition information are input. It is a completed model.

FIG. 19 is a flowchart showing a processing procedure of diagnostic support according to the second embodiment. The processing unit 11 of the diagnosis support device 1 authenticates the user (step S231) and acquires health-related information from the distributed database 3 (step S232) in the same manner as in steps S131 to S133 of the first embodiment. Acquire self-physical condition information indicating the current physical condition (step S233).

Next, the processing unit 11 refers to the infection contact information associated with the user ID, and determines whether or not there is a risk of a predetermined infectious disease (step S234).

The processing unit 11 determines that there has been a quasi-dense contact when the user has received a notification that there has been a dense contact and when the user has received a warning information that the facility where the cluster has occurred has been used. If so, it is determined that there is a risk of a predetermined infectious disease. That is, when the infected contact information includes rich contact information, alert information, or semi-rich contact information, it is determined that there is a risk of a predetermined infectious disease.

If a predetermined number of days have passed from the day when it is determined that there is a risk of a predetermined infectious disease, the processing unit 11 determines that there is no risk. For example, when the predetermined infectious disease is a new type coronavirus infection, the processing unit 11 determines that there is no risk of the predetermined infectious disease when 14 days have passed.

When it is determined that there is an infectious disease risk (step S234: YES), the processing unit 11 transmits an additional inquiry item regarding the infectious disease to the user terminal device 4 and acquires additional self-physical condition information (step S235). Then, the infectious disease risk and the aggravation risk are diagnosed by inputting the user's health-related information and self-physical condition information into the infectious disease diagnosis learning model 20 (step S236).

When it is determined that there is no risk of infectious disease (step S234: NO), or when the process of step S236 is completed, the processing unit 11 executes the process of step S134 or less described in the first embodiment.

FIG. 20 is a schematic diagram showing an example of the diagnostic display screen 290 according to the second embodiment. The diagnostic display screen 290 according to the second embodiment has the same configuration as the diagnostic display screen 90 of the first embodiment, and the display items in the AI diagnosis result display unit 290h are mainly different. The AI diagnosis result display unit 290h according to the second embodiment displays the name of a predetermined infectious disease, the risk of infectious disease, and the risk of aggravation in addition to the diagnosis result by the disease diagnosis learning model 14.

As described above, according to the diagnosis support method according to the second embodiment, the user terminal device 4 can be notified of the infection risk for a predetermined infectious disease and the aggravation risk due to the infection.

(Embodiment 3)
The diagnostic support system according to the third embodiment is capable of selecting an appropriate medical examination template from a plurality of standard medical examination templates based on health-related information and acquiring the optimum user's self-physical condition information. It is different from the first embodiment and the second embodiment. Since the other configurations of the diagnosis support system are the same as those of the diagnosis support system according to the first embodiment, the same reference numerals are given to the same parts, and detailed description thereof will be omitted.

FIG. 21 is a block diagram showing a configuration example of the diagnosis support device 1 according to the third embodiment. The diagnosis support device 1 according to the third embodiment includes the same processing unit 11, storage unit 12, and communication unit 13 as in the first embodiment. The storage unit 12 further stores the medical examination learning model 21 and the medical examination information 22.

The medical examination information 22 stores a plurality of typical standard medical examination templates. Each interview template contains multiple interview items. The interview template includes, for example, a standard interview template, a template for hypertension users, a template for hypertension users, a template for users suspected of having an infectious disease, and the like.

FIG. 22 is a block diagram showing a configuration example of the interview learning model 21 according to the third embodiment. The basic configuration of the interview learning model 21 is the same as that of the disease diagnosis learning model 14, and includes an input layer 21a, an intermediate layer 21b, and an output layer 21c. The interview learning model 21 is a trained model trained to output information indicating the appropriateness of each of the plurality of interview templates when health-related information is input. The output layer 21c includes a plurality of neurons corresponding to each of the plurality of inquiry templates, and each neuron outputs the degree of certainty that each inquiry template is appropriate.
As in the second embodiment, the input layer 21a of the interview learning model 21 may be configured to further input infection contact information.

The method of generating the interview learning model 21 is not particularly limited, but as in the first embodiment, machine learning may be performed by supervised learning. Specifically, for the user's health-related information, learning data labeled with teacher data indicating the type of interview template that the doctor deems appropriate for the user is generated. Then, the processing unit 11 of the diagnosis support device 1 may generate the interview learning model 21 by machine learning or deep learning the neural network model before learning using the generated learning data.

When the processing unit 11 of the diagnostic support device 1 acquires the self-physical condition information in step S133 shown in FIG. 12, the processing unit 11 identifies an appropriate medical examination pattern by inputting the health-related information acquired in step S132 into the medical examination learning model 21. , The medical examination item based on the specified medical examination pattern is transmitted to the user terminal device 4. The user terminal device 4 displays the inquiry items transmitted from the diagnosis support device 1, and transmits the self-physical condition information input by the user to the diagnosis support device 1. In this way, the diagnosis support device 1 can acquire an answer to an appropriate inquiry item based on the user's health-related information as self-physical condition information.

As described above, according to the diagnosis support method according to the third embodiment, it is possible to collect more appropriate self-physical condition information of the user and diagnose the health condition of the user.

(Embodiment 4)
The diagnostic support system according to the fourth embodiment is different from the first to third embodiments in that the inquiry item is generated by using the inquiry item generation learning model 23. Since the other configurations of the diagnosis support system are the same as those of the diagnosis support system according to the first embodiment, the same reference numerals are given to the same parts, and detailed description thereof will be omitted.

The storage unit 12 of the diagnosis support device 1 according to the fourth embodiment stores the inquiry item generation learning model 23. The question-and-answer generation learning model 23 is, for example, an LSTM (Long Short-Term Memory) which is a kind of recurrent neural network (RNN), and interactively asks the following question-and-answer items according to the user's answer. It is a model to generate.

FIG. 23 is a block diagram showing a configuration example of the inquiry item generation learning model 23 according to the fourth embodiment. The inquiry item generation learning model 23 has an input layer 23a, an intermediate layer 23b, and an output layer 23c. The input layer 23a includes a plurality of neurons that receive user answers to a series of questions. The user's answer is converted into a numerical code and input to each neuron. The intermediate layer 23b extracts the feature amount related to the user's health condition based on the feature amount of the user's answer input to each neuron of the input layer 23a and the feature amount of the user's answer input and held in the past, and the feature is said. Execute the arithmetic processing to convert to the numerical code indicating the inquiry item according to the quantity. The output layer 23c has a plurality of neurons that output a numerical code indicating the next question to be asked in response to the user's answer.

Questionnaire generation The method of generating the learning model 23 is not particularly limited, but for example, the content of an interactive examination between a doctor and a patient, that is, for learning including a series of a plurality of inquiry items and answers. It is advisable to prepare the data and generate the interview pattern by machine learning the recursive neural network so that the interview conducted by the doctor in the actual examination is reproduced. Questionnaire generation The generation of the learning model 23 may be executed by the diagnostic support device 1 or by another device.

FIG. 24 is an explanatory diagram showing a method of generating a medical examination item according to the fourth embodiment. In FIG. 24, the upper and lower center circles indicate the intermediate layer 23b, and each of the plurality of circles indicates the intermediate layer 23b at different time points. S0 to S4 indicate the state of the intermediate layer 23b of the inquiry item generation learning model 23, in other words, the information held in the intermediate layer 23b. In FIG. 24, the lower circle indicates the information input to the inquiry item generation learning model 23, specifically, the user's answer, and the upper circle indicates the information output from the inquiry item generation learning model 23. Specifically, it shows the items to be interviewed by the user.

When the processing unit 11 of the diagnosis support device 1 acquires self-physical condition information in step S133 shown in FIG. 12, the diagnosis support device 1 generates a question item using the question item generation learning model 23, and the user terminal device 4 The user's self-physical condition information is acquired interactively with the user via.

Specifically, the processing unit 11 of the diagnostic support device 1 first transmits the medical examination items based on the standard medical examination template to the user terminal device 4. The user terminal device 4 displays the inquiry items transmitted from the diagnosis support device 1, and transmits the answer input by the user to the diagnosis support device 1. The processing unit 11 acquires the answer transmitted from the user terminal device 4, inputs the acquired answer (answer 1) into the input layer 23a of the inquiry item generation learning model 23, and thereby receives the next inquiry item (question item 1). ) Is generated. The processing unit 11 transmits the generated inquiry items to the user terminal device 4. The user terminal device 4 displays the inquiry items transmitted from the diagnosis support device 1, and transmits the answer input by the user to the diagnosis support device 1. The processing unit 11 acquires the answer transmitted from the user terminal device 4, inputs the acquired answer (answer 2) into the input layer 23a of the inquiry item generation learning model 23, and thereby receives the next inquiry item (question item 2). ) Is generated. In the same manner below, the processing unit 11 can effectively acquire the user's self-physical condition information by transmitting and receiving the medical examination items and the answers via the user terminal device 4.

As described above, according to the diagnosis support method according to the fourth embodiment, it is possible to collect more appropriate self-physical condition information of the user and diagnose the health condition of the user.
By using the medical examination item generation learning model 23 in combination with the medical examination learning model 21 of the third embodiment, the user's self-physical condition information may be collected more effectively. That is, by generating the first medical examination item by the medical examination learning model 21 and acquiring the user's answer, and then using the medical examination item generation learning model 23 to generate the medical examination item and acquiring the user's answer, it is more appropriate. The user's self-physical condition information can be acquired.

(Embodiment 5)
The diagnostic support system according to the fifth embodiment is different from the first embodiment in that it can monitor and notify the cognitive function and the user's behavior. Since the other configurations of the diagnosis support system are the same as those of the diagnosis support system according to the first embodiment, the same reference numerals are given to the same parts, and detailed description thereof will be omitted.

FIG. 25 is a flowchart showing a processing procedure of diagnostic support according to the fifth embodiment. The processing unit 11 of the diagnosis support device 1 acquires the user's health-related information from the distributed database 3 (step S551). Then, the processing unit 11 determines the user's risk of dementia based on the cognitive function test result information included in the health-related information (step S552). The processing unit 11 may determine the risk of dementia by using the information included in other health-related information in addition to the cognitive function test result information.

When it is determined that there is no risk of dementia (step S552: NO), the processing unit 11 ends the processing. When it is determined that there is a risk of dementia (step S552: YES), the processing unit 11 transmits the history of the cognitive function test results to related persons such as relatives of the user (step S553).

Next, the processing unit 11 determines the presence or absence of abnormal behavior based on the user's activity information, particularly the position and date and time information (step S554). When it is determined that there is no abnormal behavior (step S554: NO), the processing unit 11 ends the processing. When it is determined that there is an abnormal behavior (step S554: YES), the warning information is notified to the relatives of the user, the caregiver, and the like (step S555).

In addition, the processing unit 11 transmits tracking information indicating the user's movement route based on the user's position and date and time to the user's related parties (step S556).

(Embodiment 6)
The diagnostic support system according to the sixth embodiment is different from the first embodiment in that the diagnosis and treatment by medical treatment can be optimized. Since the other configurations of the diagnosis support system are the same as those of the diagnosis support system according to the first embodiment, the same reference numerals are given to the same parts, and detailed description thereof will be omitted.

FIG. 26 is a conceptual diagram of the disease diagnosis learning model 614 according to the sixth embodiment. The disease diagnosis learning model 614 according to the sixth embodiment performs machine learning using basic medical information, external medical information, and user's health-related information as learning data, and is related to medical care together with the same diagnosis results as in the first embodiment. Auxiliary information can be output. Auxiliary information includes appropriate treatment method, examination method, latest treatment method, medical paper information, new drug information, etc. in relation to the diagnosis result, and medical optimization is planned.

The basic medical information includes a plurality of medical record information. The medical record information includes the user's self-physical condition information and information such as a doctor's diagnosis result, a test method, a test result, and a treatment method. As the use of the diagnosis support system according to the present embodiment progresses, the medical chart information includes the user's health-related information and self-physical condition information, the doctor's diagnosis result, test method, test result, treatment method, and the like, and the doctor. The medical information used and the treatment results will be included. Of the external medical information, the medical information used by the doctor is information that the doctor has determined to be useful and recorded in an electronic medical record.

External medical information includes domestic and foreign treatment methods, clinical trial data, research and development information, medical paper information, new drug information, side effect information, medical examination machine information, surgical methods, and information on approval of medicines and medical devices.

The user's health-related information is basically the same as the health-related information described in the first embodiment, and includes feedback information regarding follow-up after treatment.

FIG. 27 is a block diagram showing a configuration example of the disease diagnosis learning model 614 according to the sixth embodiment. The disease diagnosis learning model 614 according to the sixth embodiment includes, for example, a first learning model 614a, a second learning model 614b, and a third learning model 614c.

The first learning model 614a is, for example, a model in which diagnosis result information is output when user's health-related information and self-physical condition information are input. The configuration of the first learning model 614a is the same as that of the first embodiment.

In the second learning model 614b, in addition to the user's health-related information and self-physical condition information, the diagnosis result information output from the first learning model 614a is input. The second learning model 614b is a model that outputs the treatment method information and medical information suitable for the diagnosis result as auxiliary information when the user's health-related information, self-physical condition information and diagnosis result information are input.
By collecting medical record information, which is basic medical information, it is possible to generate learning data in which the user's health-related information, self-physical condition information, and diagnosis results are associated with treatment methods and useful medical information as teacher data. .. The second learning model 614b can be obtained by training the neural network using the training data.

The diagnosis support device 1 stores a table in which the diagnosis result is associated with the treatment method and medical information in the storage unit 12, and refers to the table using the diagnosis result as a key to obtain the treatment method and medical information. It may be configured to determine. If the information necessary to generate the second learning model 614b has not been collected, operate using the table, and if sufficient training data has been collected, generate and replace the second learning model 614b. You may do it.

In the third learning model 614c, in addition to the user's health-related information and self-physical condition information, the diagnosis result information output from the first learning model 614a, the treatment method information output from the second learning model 614b, and the like are input. .. The third learning model 614c is a model that outputs the treatment result prediction information when the treatment method is selected when the user's health-related information, self-physical condition information, diagnosis result information and treatment method information are input as auxiliary information. Is.
By collecting medical record information, which is basic medical information, it is possible to generate learning data in which the user's health-related information, self-physical condition information, diagnosis result and treatment method information are associated with treatment result prediction as teacher data. .. Treatment outcome prediction is information about the user's follow-up after treatment, in other words health-related information collected after treatment. By training the neural network using the training data, the third learning model 614c can be obtained.

As described above, according to the diagnosis support method according to the sixth embodiment, the diagnosis and treatment by the doctor can be optimized.

The disease diagnosis learning model 614 according to the sixth embodiment may constantly collect basic medical information, which is medical record information, and relearn the diagnosis support device 1 at an appropriate timing.

1 Diagnosis support device 2 Information processing device 3 Distributed database 3a Blockchain 4 User terminal device 5 Wearable terminal 6 Doctor terminal device 10 Recording medium 11 Processing unit 12 Storage unit 12a Computer program 12b Database 13 Communication unit 14 Disease diagnosis learning model 15 Image Diagnostic learning model 16 Disease basic information 17 Hospital information 18 Pharmacy information 19 Cognitive function test questions 20 Infectious disease diagnosis learning model 21 Questionnaire learning model 22 Questionnaire information 23 Questionnaire generation learning model

Claims (14)

Get user health related information recorded in a decentralized database including blockchain,
Get the user's current self-physical condition information and
When the health-related information and the self-physical condition information are input, the health-related information acquired from the distributed database and the self-physical condition information acquired from the user are added to a learning model that outputs a diagnosis result related to the user's health condition. Diagnose the user's health by entering
Send the diagnosis result of the user's health condition to the user's terminal device,
An information processing method for transmitting a diagnosis result of a user's health condition and the self-physical condition information to an external communication device. The diagnosis result and the self-physical condition information are transmitted to an external communication device which is a terminal device of a doctor.
The information processing method according to claim 1, wherein a doctor's diagnosis result regarding a user's health condition is acquired. The information processing method according to claim 2, wherein both or one of the diagnosis result and the diagnosis result of the doctor based on the learning model is transmitted to the terminal device of the user. Obtain prescription information from the doctor and
The information processing method according to claim 2 or 3, wherein the acquired prescription information is transmitted to the terminal device of the user or the terminal device of the pharmacy. The information processing method according to claim 4, wherein the prescription information is recorded in the distributed database as the health-related information. The health-related information includes medical images and includes medical images.
The learning model includes an image diagnosis learning model that outputs a diagnosis result related to a user's health condition when the medical image is input.
The information processing method according to any one of claims 1 to 5, wherein the medical image acquired from the distributed database is input to the diagnostic imaging learning model to diagnose the health condition of the user. The health-related information is
Claims 1 to 6 including behavior history information including the positions, dates and times of a plurality of users, and infection contact information indicating whether or not each user has or may have a predetermined infectious disease. The information processing method according to any one of the above items. The infection contact information is
The information processing method according to claim 7, wherein the information processing method includes dense contact information indicating that the user has made close contact with a predetermined infectious disease patient. The infection contact information is
The information processing method according to claim 7, wherein the information processing method includes tracking information indicating that the user used the same facility at the same date and time as the predetermined infectious disease patient. According to any one of claims 7 to 9, the user is requested to additionally input the self-physical condition information according to the behavior history information and the infected contact information of the user, and the additional self-physical condition information is acquired from the user. The information processing method described. The information processing according to any one of claims 1 to 10, which requests the user to additionally input the self-physical condition information according to the diagnosis result based on the learning model and acquires the additional self-physical condition information from the user. Method. The health-related information is
User's past medical information provided by the hospital, information detected by a wearable terminal that measures the user's activity status, user activity information, user's body composition information, user's constitution information, medical history, medication history The information processing method according to any one of claims 1 to 11, which includes information on side effects to a drug, information on the physical condition of the user, information on the lifestyle of the user, information on cognitive function test results, or information related to infectious diseases. The first acquisition unit that acquires the user's health-related information recorded in the distributed database including the blockchain,
The user's current self-physical condition information is obtained from the second acquisition unit,
When the health-related information and the self-physical condition information are input, a learning model that outputs a diagnosis result related to the user's health condition and a learning model.
A processing unit that diagnoses the health condition of the user by inputting the health-related information acquired from the distributed database and the self-physical condition information acquired from the user.
A diagnostic support device including a communication unit that transmits the diagnosis result of the user's health condition to the user's terminal device, and transmits the diagnosis result of the user's health condition and the self-physical condition information to an external communication device. Get user health related information recorded in a decentralized database including blockchain,
Get the user's current self-physical condition information and
When the health-related information and the self-physical condition information are input, the health-related information acquired from the distributed database and the self-physical condition information acquired from the user are used in a learning model that outputs a diagnosis result related to the user's health condition. Diagnose the user's health condition by entering
Send the diagnosis result of the user's health condition to the user's terminal device,
A computer program for causing a computer to execute a process of transmitting a diagnosis result of a user's health condition and the self-physical condition information to an external communication device.

Images