Electronic device for diagnosing a user's illness based on biological signals and its control method

Abstract

The present disclosure provides an electronic device and a control method thereof. The electronic device according to an embodiment of the present disclosure includes a communication interface, a memory for storing first and second trained neural network models, and one or more processors for acquiring biometric data of a user via the communication interface, inputting the acquired biometric data into the trained first neural network model to generate first information about a first disease of the user, inputting the acquired biometric data into the trained second neural network model to generate second information about the first disease of the user, and generating a diagnosis result for the user for the first disease of the first type based on the first and second information.

Description

[Technical Field]

[0001] The present disclosure relates to an electronic device for diagnosing a disease of a user based on a biological signal and a control method thereof, and more particularly to an electronic device for diagnosing a disease of a user using a pre-trained neural network model and a control method thereof. [Background technology]

[0002] Myocardial infarction and ischemic heart disease are the most common causes of sudden death, and the number of patients with myocardial infarction and ischemic heart disease has been increasing every year recently. In the case of myocardial infarction and ischemic heart disease, early detection and treatment can prevent death and disability, so much research is being conducted on early diagnosis and prediction of myocardial infarction and ischemic heart disease.

[0003] For early diagnosis of myocardial infarction and ischemic heart disease, a method is widely used in which an electrocardiogram is measured, the measured electrocardiogram signal is displayed in the form of a graph, and the presence or absence of myocardial infarction and ischemic heart disease in a patient's heart is determined based on the graph. In particular, with the recent development of deep learning technology, there has been growing interest in a method of applying deep learning technology to the medical field to obtain an analysis result of an electrocardiogram signal by simply inputting a graph of the electrocardiogram signal into a neural network model. Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure has been devised in view of the above-mentioned background art, and a problem to be solved by the present disclosure is to provide an electronic device and a control method thereof for diagnosing a user's disease based on a pre-trained neural network model and the user's biological signals.

[0005] However, the problems to be solved by the present disclosure are not limited to those mentioned above, and other problems not mentioned will be clearly understood from the following description. [Means for solving the problem]

[0006] To achieve the above object, a method performed by an electronic device according to an embodiment of the present disclosure is disclosed, which includes the steps of acquiring biometric data of a user, inputting the acquired biometric data into a first trained neural network model to generate first information about a first disease of the user, inputting the acquired biometric data into a second trained neural network model to generate second information about the first disease of the user, and generating a diagnosis result for the user for a first type of the first disease based on the first and second information.

[0007] Alternatively, the step of generating the diagnostic result may include a step of adjusting a first probability value included in the first information based on a second probability value included in the second information, and generating a diagnostic result for the user for the first disease of the first type based on the first probability value.

[0008] Alternatively, the step of generating the diagnosis result may include adjusting the first probability value by applying a weighting value corresponding to the second probability value to the first probability value when the second probability value is equal to or greater than a reference value, and maintaining the first probability value when the second probability value is less than the reference value.

[0009] Alternatively, the step of generating the diagnostic result may include the steps of generating a first diagnostic result corresponding to the first disease of the first type when the first probability value is equal to or greater than a first value, generating a second diagnostic result corresponding to the first disease of the first type when the first probability value is less than the first value and equal to or greater than a second value, and generating a third diagnostic result corresponding to the first disease of the first type when the first probability value is less than the second value.

[0010] Alternatively, the first information may be information about the presence or absence of the first disease, and the second information may be information about a first type of the first disease.

[0011] Alternatively, the method may include extracting first and second neural network models from a plurality of neural network models based on the first disease and the diagnosis result type.

[0012] Alternatively, the steps of generating first information about the first disease and generating second information about the first disease may be performed in parallel.

[0013] Alternatively, the first and second neural network models may be trained based on training data to which a plurality of labels, each set according to different criteria, are assigned to the same electrocardiogram signal, and the plurality of labels may include a first type label corresponding to the presence or absence of the first disease and a second type label corresponding to a first type of the first disease.

[0014] Alternatively, the biological data may include an electrocardiogram signal, the first disease may include any one of myocardial infarction and ischemic heart disease, and the first type may include any one of ST elevation myocardial infarction (STEMI) and non-ST elevation myocardial infarction (NSTEMI).

[0015] To achieve the above object, one embodiment of the present disclosure provides an electronic device, including a communication interface, a memory for storing first and second trained neural network models, and one or more processors for acquiring biometric data of a user via the communication interface, inputting the acquired biometric data into the trained first neural network model to generate first information about a first disease of the user, inputting the acquired biometric data into the trained second neural network model to generate second information about the first disease of the user, and generating a diagnosis result for the user for the first disease of the first type based on the first and second information.

[0016] According to one embodiment of the present disclosure, a non-transitory computer-readable recording medium storing computer instructions that, when executed by a processor of an electronic device, perform operations of the electronic device, the operations including: acquiring biometric data of a user; inputting the acquired biometric data into a first neural network model that has already been trained to generate first information about a first disease of the user; inputting the acquired biometric data into a second neural network model that has already been trained to generate second information about the first disease of the user; and generating a diagnosis result for the user for the first disease of the first type based on the first and second information. [Effects of the Invention]

[0017] An electronic device according to an embodiment of the present disclosure can train multiple neural network models to generate information about the presence or absence of a disease and information about the type of disease using the same training data, thereby reducing the cost and time required to obtain and preprocess the training data.

[0018] Furthermore, by providing information on the user's illness and type simply by measuring the user's biosignals, the user's illness can be predicted in advance and guided to early diagnosis. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is an exemplary diagram of an electronic device according to one embodiment of the present disclosure. [Figure 2] FIG. 1 is a block diagram of an electronic device according to an embodiment of the present disclosure. [Figure 3] 1 is a flowchart that schematically illustrates a method for controlling an electronic device according to an embodiment of the present disclosure. [Figure 4] FIG. 2 is an exemplary diagram illustrating first and second neural network models that have already been trained and stored in an electronic device according to an embodiment of the present disclosure. [Figure 5]5a is an exemplary diagram showing a method for training first and second neural network models 20-1 and 20-2 according to an embodiment of the present disclosure, and FIG. 5b is an exemplary diagram showing a conventional method for training multiple neural network models to generate information on the presence or absence of a specific disease and information on a specific type of a specific disease. [Figure 6] FIG. 2 is an exemplary diagram illustrating a method for diagnosing a user's illness using first and second neural network models according to an embodiment of the present disclosure. [Figure 7] FIG. 2 is a detailed block diagram of an electronic device according to an alternative embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present disclosure. The embodiments presented in this disclosure are provided to enable those skilled in the art to use or practice the contents of the present disclosure. Therefore, various modifications to the embodiments of the present disclosure will be apparent to those skilled in the art. That is, the present disclosure may be embodied in various different forms and is not limited to the following embodiments.

[0021] Throughout the specification of the present disclosure, the same or similar reference numerals refer to the same or similar components. In addition, in order to clearly explain the present disclosure, reference numerals of parts that are not relevant to the explanation of the present disclosure may be omitted from the drawings.

[0022] The term "or" as used in this disclosure is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified or otherwise clear from the context in this disclosure, "X uses A or B" should be understood to mean one of the natural inclusive permutations. For example, unless otherwise specified or otherwise clear from the context in this disclosure, "X uses A or B" can be interpreted as either X uses A, X uses B, or X uses both A and B.

[0023] The term "and / or" as used in this disclosure must be understood to indicate and include all possible combinations of one or more of the associated listed concepts.

[0024] The terms "comprises" and / or "comprising" as used in this disclosure should be understood to mean that the specified features and / or components are present. However, the terms "comprises" and / or "comprising" should not be understood to exclude the presence or addition of one or more other features, other components and / or combinations thereof.

[0025] In this disclosure, unless otherwise specified or clear from the context as referring to the singular form, the singular should generally be construed as including "one or more."

[0026] The term "nth (n is a natural number)" used in this disclosure can be understood as an expression used to distinguish components of the present disclosure from one another based on a predetermined criterion, such as functional, structural, or convenience of description. For example, in this disclosure, components that perform different functional roles can be classified as a first component or a second component. However, components that are substantially identical within the technical concept of the present disclosure but must be distinguished for convenience of description can also be classified as a first component or a second component.

[0027] The term "acquire" as used in this disclosure may be understood to mean not only receiving data from an external device or system via a wired or wireless communication network, but also generating data in an on-device form.

[0028] Meanwhile, the terms "module" or "unit" used in this disclosure may be understood to refer to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a portion thereof, hardware or a portion thereof, or a combination of software and hardware. Here, a "module" or "unit" may refer to a unit composed of a single element or a unit expressed as a combination or collection of multiple elements. For example, as a concept of connotation, a "module" or "unit" may refer to a hardware element or a collection of hardware elements of a computing device, an application program that achieves a specific software function, a processing procedure implemented by the execution of software, or a collection of instructions for executing a program. Furthermore, as a broad concept, a "module" or "unit" may refer to a computing device itself that constitutes a system, or an application executed on a computing device. However, the above concepts are merely examples, and the concepts of "module" and "unit" may be defined in various ways within the scope of understanding of those skilled in the art based on the contents of this disclosure.

[0029] The term "model" as used in this disclosure may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a collection of software units for solving a specific problem, or an abstract model of a processing process for solving a specific problem. For example, a neural network "model" may refer to a system implemented as a neural network that has problem-solving capabilities through learning. Here, a neural network may have problem-solving capabilities by optimizing parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network or a neural network ensemble in which multiple neural networks are combined.

[0030] The term "model" as used in this disclosure refers to a system implemented using mathematical concepts and language to solve a specific problem, a collection of software units to solve a specific problem, or an abstract structure of a processing process to solve a specific problem. For example, a neural network "model" refers to the entire system embodied by a neural network that has acquired problem-solving capabilities through learning.

[0031] At this time, the neural network actively optimizes the parameters that connect the nodes or neurons through learning, thereby acquiring the ability to solve problems. A neural network "model" may include a single neural network as a component, or may include a group of neural networks that are intentionally combined together.

[0032] FIG. 1 is an exemplary diagram of an electronic device according to one embodiment of the present disclosure.

[0033] 1 , an electronic device 100 according to an embodiment of the present disclosure acquires a user's biosignal 10 from an external electronic device 200 interfacing with the electronic device 100. Here, the external electronic devices 200-1 and 200-2 (hereinafter referred to as 200) interfacing with the electronic device 100 may be embodied as various devices that are connected to the electronic device 100 via network communication and perform a function of measuring biosignals 10-1 and 10-2 (hereinafter referred to as 10) of users 1-1 and 1-2 (hereinafter referred to as 1). For example, the external electronic device 200 may be embodied as an electrocardiogram measuring device, a smart watch, a display device, etc. Meanwhile, in order to interfacing with the electronic device 100, the external electronic device 200 may register information about the external electronic device 200 (or information about the institution where the external electronic device 200 is located) in advance with the electronic device 100.

[0034] The electronic device 100 may generate a diagnosis result for the user based on the user signal acquired from the external electronic device 200. Specifically, the electronic device 100 may analyze the user signal acquired from the external electronic device 200 to determine the user's health condition or identify the presence or absence of a disease and the type of disease. In particular, to generate a diagnosis result for the user, the electronic device 100 may use a pre-trained neural network model 20. For example, the electronic device 100 may input the acquired biosignal 10 to the pre-trained neural network model 20 and acquire user health condition information or user disease information (e.g., information including the presence or absence of a disease or the type of disease) as an output value of the pre-trained neural network model 20. To this end, the pre-trained neural network model 20 may be pre-trained to output user health condition information or user disease information based on various user biosignals 10.

[0035] Meanwhile, the electronic device 100 can transmit the generated diagnosis result to the external electronic device 200 or to a user terminal device, thereby allowing the user to grasp health condition information by measuring the biosignal 10 alone without the assistance of a specialized facility or expert.

[0036] Hereinafter, an electronic device 100 according to an embodiment of the present disclosure will be described in detail with reference to FIGS.

[0037] FIG. 2 is a block diagram of an electronic device 100 according to one embodiment of the present disclosure.

[0038] The electronic device 100 according to an embodiment of the present disclosure may be a hardware device or part of a hardware device that performs comprehensive data processing and calculations, or may be a software-based computing environment connected via a communication network. For example, the electronic device 100 may be a server (e.g., a platform server) that performs intensive data processing functions and shares resources, or a client that shares resources by interacting with the server. The electronic device 100 may also be a cloud system that enables multiple servers and clients to interact with each other and process data comprehensively.

[0039] The above description is merely an example related to the type of electronic device 100, and the type of electronic device 100 can be configured in various ways within the scope that can be understood by a person skilled in the art based on the contents of this disclosure. For example, the electronic device 100 can be embodied as an electronic device 100 that measures a user's biosignal (e.g., an electrocardiogram signal, etc.). However, the following description will be given assuming that the electronic device 100 of the present disclosure is a server.

[0040] 2, an electronic device 100 according to an embodiment of the present disclosure may include a memory 110, a communication interface (network unit) 120, and one or more processors 130. However, since FIG. 2 is merely an example, the electronic device 100 may include other components for implementing a computer environment. Also, only some of the disclosed components may be included in the electronic device 100.

[0041] 2, an electronic device 100 according to an embodiment of the present disclosure may include a memory 110, a communication interface (network unit) 120, and one or more processors 130. However, since FIG. 1 is merely an example, the electronic device 100 may include other components for implementing a computer environment. Also, the electronic device 100 may include only a portion of the disclosed components.

[0042] The memory 110 according to an embodiment of the present disclosure can be understood as a component including hardware and / or software for storing and managing data processed by the electronic device 100. That is, the memory 110 can store any type of data generated or determined by the processor 130 and any type of data received by the processor 130 via the communication interface 120.

[0043] For example, the memory 110 may store a plurality of pre-trained neural network models. Each neural network model may be matched with at least one other neural network model according to the type of disease to be diagnosed and the type of diagnostic result to be generated, and stored in the memory 110. Here, the matched neural network models may be trained to output different types of information based on the same input (e.g., biosignals). For example, the memory 110 may store first and second neural network models. Here, the first and second neural network models may be pre-trained based on the same training data and then stored in the memory 110. The pre-trained first and second neural network models may be trained to analyze a user's biosignals and generate different types of information from the biosignals. Here, the memory 110 may also store training data used to train the first and second neural network models.

[0044] For example, the memory 110 may include at least one type of storage medium selected from the group consisting of a flash memory 130 type, a hard disk type, a multimedia card micro type, a card-type memory, a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. The memory 110 may also include a database system that manages data in a predetermined manner. The types of memory 110 described above are merely examples, and various configurations of the memory 110 are possible within the scope of what would be understood by one skilled in the art based on the present disclosure.

[0045] The communication interface 120 according to an embodiment of the present disclosure may be understood as a component for transmitting and receiving data via any type of known wired or wireless communication system. The electronic device 100 may transmit and receive various information to and from the external electronic device 100 via the communication interface 120. For example, the electronic device 100 may receive a user's biometric signal measured by the external electronic device 100 from the external electronic device 100 via the communication interface 120, or may transmit user's diagnostic result information generated by the electronic device 100 to the external electronic device 100.

[0046] For this reason, the communication interface 120 can be understood as a component that transmits and receives data via any type of known wired or wireless communication system. For example, the communication interface 120 can transmit and receive data using a wired or wireless communication system such as a local area network (LAN), wideband code division multiple access (WCDMA), long term evolution (LTE), wireless broadband internet (WiBro), 5th generation mobile communication (5G), ultra wide-band, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity (WiFi), near field communication (NFC), or Bluetooth. The above-mentioned communication system is merely an example, and various wired and wireless communication systems for transmitting and receiving data by the network unit 130 can be applied in addition to the above examples.

[0047] The processor 130 according to an embodiment of the present disclosure is electrically connected to the memory 110 and the communication interface 120 and can control the overall operation of the electronic device 100 .

[0048] The processor 130 may be understood as a component including hardware and / or software for executing computer operations. For example, the processor 130 may read a computer program to perform data processing for machine learning. The processor 130 may process operations such as input data processing for machine learning, feature extraction for machine learning, and error calculation based on backpropagation. The processor 130 for performing such data processing may include a central processing unit (CPU), a general-purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), etc. The types of processor 130 described above are merely examples, and various types of processor 130 may be configured within the scope of what one skilled in the art would understand based on the contents of this disclosure.

[0049] FIG. 3 is a flow chart that schematically illustrates a method for controlling the electronic device 100 according to one embodiment of the present disclosure.

[0050] 3, the processor 130 acquires biometric data of the user 1 (S310). Specifically, the processor 130 acquires the biometric data of the user 1 from the external electronic device 200 registered in the electronic device 100 via the communication interface 120. Here, information about the external electronic device 200 may be stored in the memory 110 of the electronic device 100.

[0051] As an example, the biometric data of the user 1 may be an electrocardiogram signal. Specifically, the external electronic device 200 may be an electrocardiogram device including multiple electrodes (or multiple patches including the respective electrodes) attached to different parts of the body of the user 1. The external electronic device 200 may generate an electrocardiogram signal (or electrocardiogram data) corresponding to the user 1 based on the electrocardiogram of the user 1 measured from the respective electrodes, and transmit the generated electrocardiogram signal to the electronic device 100. The processor 130 may then acquire the electrocardiogram signal generated by the external electronic device 200 via the communication interface 120. In the following, for ease of understanding the present disclosure, the biometric data will be described assuming that it is an electrocardiogram signal.

[0052] Referring to FIG. 3, the processor 130 may input the acquired biological data into a first neural network model that has already been trained to generate first information about the disease of the user 1 (S320).

[0053] As an example, the first information may be information about the presence or absence of a disease. Therefore, the first neural network model may be a model trained in advance to generate information about the presence or absence of a disease of user 1. Specifically, the first neural network model may be a model trained in advance to, when an electrocardiogram signal of user 1 is input, determine whether user 1 has a disease based on the electrocardiogram signal and generate information about the presence or absence of the disease of user 1.

[0054] In particular, the first neural network model is pre-trained to determine only the presence or absence of a specific disease. Hereinafter, for the convenience of explanation of this disclosure, the specific disease will be referred to as a first disease.

[0055] As an example, when an electrocardiogram signal of user 1 is input, the first neural network model can determine whether user 1 has a first disease based on the electrocardiogram signal and generate information about the presence or absence of the first disease as an output value.

[0056] Here, the information on the presence or absence of the first disease may include a probability value (or score) of the first disease. Specifically, when the electrocardiogram signal of the user 1 is input to the first neural network model, the processor 130 may acquire the probability value of the first disease as an output value through a sigmoid layer or a softmax layer arranged at an output end among a plurality of layers included in the first neural network model.

[0057] In addition, the processor 130 may input the acquired biological data into the second neural network model that has already been trained, to generate second information about the disease of the user 1 (S330).

[0058] As an example, the second information may be information about the type of disease. Therefore, the second neural network model may be a model that has been trained in advance to generate information about the type of disease of user 1. Specifically, the second neural network model may be a model that has been trained in advance to, when an electrocardiogram signal of user 1 is input, identify the type of disease of user 1 based on the electrocardiogram signal and generate information about the type of disease of user 1.

[0059] In particular, the second neural network model is pre-trained to generate information about a specific disease type, which may be the same disease (i.e., the first disease) that the first neural network model identifies as being present or absent.

[0060] The second neural network model may also be pre-trained to generate information about a specific type of a specific disease. That is, if there are various types associated with a specific disease, the second neural network model may be pre-trained to identify whether the specific disease of user 1 corresponds to a specific type among the various types associated with the specific disease. In particular, the second neural network model may analyze a biosignal to identify characteristics of the biosignal, and then be trained to identify a specific type of the specific disease corresponding to the identified characteristics. Hereinafter, for convenience of explanation of the present disclosure, the specific type will be referred to as a first type.

[0061] As an example, when an electrocardiogram signal of user 1 is input, the second neural network model can determine whether user 1's first disease corresponds to a first type based on the electrocardiogram signal, and generate information about the first type of the first disease as an output value.

[0062] Here, the information on the first type of the first disease may include a probability value (or score) that the disease of user 1 corresponds to the first type of the first disease. Specifically, when the electrocardiogram signal of user 1 is input to the second neural network model, processor 130 may acquire, as an output value, a probability value that the disease of user 1 corresponds to the first type of the first disease through a sigmoid layer or a softmax layer arranged at an output end among a plurality of layers included in the second neural network model.

[0063] For example, the first and second neural network models may be implemented as a Convolutional Neural Network (CNN) model, a Recurrent Neural Network (RNN) model, a Relu model, or the like.

[0064] Meanwhile, the processor 130 may perform a processing process on the acquired biological signal 10 (for example, an electrocardiogram signal) before inputting the acquired biological signal 10 into the first and second neural network models.

[0065] Meanwhile, according to one embodiment of the present disclosure, the memory 110 may store a plurality of neural network models. Here, the plurality of neural network models may be matched with other neural network models according to the type of disease and the type of diagnosis result. Thus, the processor 130 may identify the type of disease to be diagnosed and the type of diagnosis result (or the type of disease and the type of diagnosis result set by the user) and determine a combination of the plurality of neural network models corresponding to the identified type of disease and the type of diagnosis result from the plurality of neural network models. The processor 130 may then extract a plurality of neural network models according to the determined combination, and then generate a diagnosis result for a specific disease for the user 1 using the extracted plurality of neural network models. Meanwhile, information on the combination of the plurality of neural network models corresponding to the type of disease and the type of diagnosis result may be stored in the memory 110 in the form of a table.

[0066] For example, when processor 130 determines to generate a first type of diagnosis result for a first disease, processor 130 may extract a first neural network model and a second neural network model from among the plurality of neural network models stored in memory 110. Then, processor 130 may input the acquired biological signals to the extracted first and second neural network models, respectively, to generate a first type of diagnosis result for the first disease for user 1.

[0067] FIG. 4 is an exemplary diagram showing first and second neural network models 20-1 and 20-2 that have already been trained and stored in the electronic device 100 according to an embodiment of the present disclosure.

[0068] 4, the processor 130 can input the acquired electrocardiogram signal to the first and second neural network models 20-1 and 20-2, respectively. Here, the processor 130 can acquire a plurality of probability values ​​through each neural network model (the first and second neural network models 20-1 and 20-2). Specifically, the processor 130 can acquire a probability value (hereinafter referred to as a first probability value) that the user 1 has a first disease through the first neural network model 20-1, and can acquire a probability value (hereinafter referred to as a second probability value) that the disease of the user 1 corresponds to a first type of the first disease through the second neural network model 20-2.

[0069] 4, the processor 130 can execute the steps of generating information about the presence or absence of a first disease and generating information about a first type of the first disease in parallel. That is, the processor 130 inputs the acquired electrocardiogram signal to the first and second neural network models 20-1 and 20-2 arranged in parallel, respectively, to obtain output values ​​(i.e., first and second probability values) of the first and second neural network models 20-1 and 20-2, respectively, thereby reducing the time required to generate information about the presence or absence of the first disease and information about the first type of the first disease.

[0070] FIG. 5a is an exemplary diagram showing a method for training first and second neural network models 20-1 and 20-2 according to one embodiment of the present disclosure, and FIG. 5b is an exemplary diagram showing a conventional method for training multiple neural network models to generate information about the presence or absence of a specific disease and information about a specific type of a specific disease.

[0071] According to one embodiment of the present disclosure, the first and second neural network models 20-1 and 20-2 can be trained based on training data to which multiple labels, each set according to different criteria, are assigned to the same electrocardiogram signal.

[0072] Specifically, the first and second neural network models 20-1 and 20-2 may have been trained using the same training data. For example, the training data may include a plurality of electrocardiogram signals, which may also be referred to as a training data set. The first and second neural network models 20-1 and 20-2 may each be trained based on the same training data including a plurality of electrocardiogram signals. However, different labels may be assigned to the plurality of electrocardiogram signals included in the training data so as to generate different types of information (information about the presence or absence of a first disease and information about a first type of the first disease).

[0073] Here, the different labels are labels set based on different criteria and may include a first type label corresponding to the presence or absence of a first disease and a second type label corresponding to a first type of the first disease. That is, referring to Fig. 5a, the learning data according to an embodiment of the present disclosure may be assigned a first type label corresponding to the presence or absence of the first disease and a second type label corresponding to the first type of the first disease.

[0074] Thus, the first neural network model 20-1 can be trained based on the labels of the first type of training data to generate information about the presence or absence of the first disease, and the second neural network model 20-2 can be trained based on the labels of the second type of training data to generate information about the first type of the first disease.

[0075] 5b, in the conventional method, in order to generate information on the presence or absence of a specific disease and information on a specific type of a specific disease, separate training data must be prepared for training each neural network model. Therefore, there is a problem that preprocessing each training data takes a lot of time and money, and it is also difficult to prepare sufficient training data for training each neural network model, especially in the case of training data that can be obtained by specialized equipment, such as biological signals 10 (e.g., electrocardiogram signals).

[0076] However, according to one embodiment of the present disclosure, by using the same training data to which multiple labels set according to different standards are assigned in order to generate different types of information (information about the presence or absence of a first disease and information about a first type of the first disease), the training data can be used redundantly to train multiple neural network models (i.e., first and second neural network models 20-1 and 20-2), thereby reducing the cost and time required for preprocessing the training data and ensuring sufficient training data for training each neural network model.

[0077] However, the present disclosure is not limited to the above-described embodiments. That is, the first and second neural network models 20-1 and 20-2 may be trained using different training data. Specifically, the first neural network model 20-1 may be trained to identify the presence or absence of myocardial infarction based on first training data including electrocardiogram data to which a first label (a label for classifying myocardial infarction) has been assigned, and the second neural network model 20-2 may be trained to identify whether the electrocardiogram data corresponds to ST-elevation myocardial infarction based on second training data including electrocardiogram data to which a second label (a label for classifying ST-elevation myocardial infarction) has been assigned.

[0078] Referring again to FIG. 3, the processor 130 can generate a diagnosis result for the user 1 for the first disease of the first type based on information about the presence or absence of the first disease and information about the first type of the first disease (S340).

[0079] Specifically, the processor 130 can generate a diagnosis result for the first disease of the user 1 by combining information on the presence or absence of the first disease and information on the first type. As an example, the processor 130 can generate a diagnosis result for the first disease of the first type by combining information on the presence or absence of the first disease and information on the first type of the first disease. That is, when the processor 130 uses the first neural network model 20-1, it can determine only the presence or absence of the first disease. However, according to one embodiment of the present disclosure, the processor 130 can use the first neural network model 20-1 to combine the result (i.e., the presence or absence of the first disease) with the result of the second neural network model 20-2 (i.e., whether it corresponds to the first type of the first disease) to generate a diagnosis result for the first disease of the first type.

[0080] Meanwhile, the processor 130 may transmit the generated diagnosis result for the first disease to the external electronic device 200 via the communication interface 120 to provide it to the user 1 .

[0081] Meanwhile, according to one embodiment of the present disclosure, the first disease may include one of myocardial infarction and ischemic heart disease, and the first type of the first disease may be a high-risk group for myocardial infarction. That is, the processor 130 may use the first neural network model 20-1 to generate information about whether or not the user 1 has myocardial infarction or ischemic heart disease based on the electrocardiogram signal, and may use the second neural network model 20-2 to generate information about the high-risk group for myocardial infarction for the user 1. Hereinafter, for convenience of explanation of the present disclosure, the first disease will be described as myocardial infarction.

[0082] As an example, the processor 130 may determine that the user 1's myocardial infarction corresponds to a high-risk group if either ST elevation myocardial infarction (STEMI) or non-ST elevation myocardial infarction (NSTEMI) is identified based on the user 1's electrocardiogram signal.

[0083] Therefore, the second neural network model 20-2 can be trained to calculate, as an output value, whether user 1 is suffering from ST-elevation myocardial infarction or non-ST-elevation myocardial infarction based on the input electrocardiogram signal. Here, the second probability value may be a probability value that user 1's acute myocardial infarction corresponds to ST-elevation myocardial infarction or a probability value that user 1's acute myocardial infarction corresponds to non-ST-elevation myocardial infarction. That is, the processor 130 can determine that user 1's disease corresponds to either ST-elevation myocardial infarction or non-ST-elevation myocardial infarction based on the second probability value acquired via the second neural network model 20-2. Meanwhile, since both ST-elevation myocardial infarction and non-ST-elevation myocardial infarction belong to the category of acute myocardial infarction, the information output via the second neural network model 20-2 (e.g., the second probability value) can also be referred to as information about the first type of the first disease of the first type. That is, the second neural network model 20-2 can be trained in advance to output information about the first type of the first disease of the first type.

[0084] For the convenience of explanation of the present disclosure, the following description will be given assuming that the first type of the first disease is ST-segment elevation myocardial infarction.

[0085] Meanwhile, according to one embodiment of the present disclosure, the first disease of the first type may be acute myocardial infarction. Thus, the processor 130 may determine that the user 1 has myocardial infarction based on the first probability value, and when it determines that the user 1 has ST-segment elevation myocardial infarction based on the second probability value, generate a diagnosis result of acute myocardial infarction for the user 1.

[0086] An embodiment of the present disclosure that generates a diagnosis result for user 1 based on the first and second probability values ​​will be described in detail below.

[0087] 6 is a flowchart showing a method for diagnosing a disease of a user 1 using first and second neural network models 20-1 and 20-2 according to an embodiment of the present disclosure. Steps S610 to S630 shown in FIG. 6 may correspond to steps S310 to S330 shown in FIG. 3, respectively, and detailed description thereof will be omitted.

[0088] According to one embodiment of the present disclosure, the processor 130 can adjust the first probability value included in the first information based on the second probability value included in the second information, and generate a diagnosis result for the user 1 for the first disease of the first type based on the first probability value.

[0089] Specifically, when a first type of the first disease is identified based on the output value (i.e., the second probability value) of the second neural network model, the processor 130 can adjust the first probability value for the presence or absence of the first disease based on the second probability value.

[0090] For example, identifying a first type of a first disease through the second neural network model may mean that the user has a high probability of having a first type of the first disease. Thus, processor 130 may adjust the first probability value for the presence or absence of the first disease based on the second probability value for the presence or absence of the first type of the first disease.

[0091] For example, if the second probability value is equal to or greater than a reference value, the processor 130 may adjust the first probability value by applying a weight corresponding to the second probability value to the first probability value, and if the second probability value is less than a reference value, the processor 130 may maintain the first probability value.

[0092] Specifically, the processor 130 may determine whether the second probability value is equal to or greater than a reference value. Here, the reference value may be set differently for each user 1 depending on the age, body shape, medical history, etc. of the user 1.

[0093] As an example, the processor 130 may determine that the user 1 has a first disease of a first type when the second probability value is equal to or greater than the reference value. Specifically, the processor 130 may determine that the second probability value is equal to or greater than the reference value and that ST-elevation myocardial infarction has been identified based on the electrocardiogram signal of the user 1. If the processor 130 determines that ST-elevation myocardial infarction has been identified, the processor 130 may adjust the first probability value by applying a weight corresponding to the second probability value to the first probability value. On the other hand, the processor 130 may determine that the second probability value is less than the reference value and that ST-elevation myocardial infarction has not been identified based on the electrocardiogram signal of the user 1. If the processor 130 determines that ST-elevation myocardial infarction has not been identified, the processor 130 may not apply a weight corresponding to the second probability value to the first probability value and maintain the first probability value.

[0094] More specifically, if the first probability value is 40, the second probability value is 32, and the reference value is 30, the processor may determine that the second probability value is equal to or greater than the reference value and determine that the user has ST-segment elevation myocardial infarction. Here, the processor 130 may determine a weighting value corresponding to the second probability value. For example, if the weighting value is set to 0.5 times the second probability value of 32, the processor 130 may apply the weighting value (16 = 32 × 0.5) to the first probability value of 40 to obtain an adjusted first probability value (56 = 32 + 16). On the other hand, if the first probability value is 40, the second probability value is 24, and the reference value is 30, the processor may determine that the second probability value is less than the reference value and determine that the user does not have ST-segment elevation myocardial infarction. Here, the processor 130 may maintain the obtained first probability value of 40.

[0095] Meanwhile, the processor 130 may adjust the first probability value based on the second probability value in various ways. For example, if the second probability value is equal to or greater than a reference value, the processor 130 may change the first probability value to the second probability value.

[0096] Alternatively, processor 130 may apply different weights depending on the difference between the first probability value and the second probability value. Specifically, if the second probability value is equal to or greater than a reference value and the difference between the first probability value and the second probability value is equal to or greater than a preset difference value, processor 130 may apply the first weight, and if the second probability value is equal to or greater than the reference value and the difference between the first probability value and the second probability value is less than the preset difference value, processor 130 may apply the second weight, which is smaller than the first weight.

[0097] Alternatively, processor 130 may apply different weights depending on which of a plurality of preset intervals includes the first probability value. Specifically, if the second probability value is equal to or greater than a reference value and the first probability value is included in the first interval (0 or greater and less than the first reference value), processor 130 may apply a third weight; if the second probability value is equal to or greater than a reference value and the first probability value is included in the second interval (1 or greater and less than the second reference value), processor 130 may apply a fourth weight that is smaller than the third weight; and if the first probability value is included in the third interval (2 or greater and less than the third reference value), processor 130 may apply a fifth weight that is smaller than the fourth weight.

[0098] Meanwhile, the processor 130 may generate a diagnosis result based on the first probability value, where the first probability value may include a first probability value adjusted based on the second probability value or a first probability value output from the first neural network model (i.e., a maintained first probability value).

[0099] As an example, the processor 130 may generate a first diagnostic result corresponding to a first disease of a first type when the first probability value is greater than or equal to a first value, generate a second diagnostic result corresponding to a first disease of a first type when the first probability value is less than the first value and greater than or equal to a second value, and generate a third diagnostic result corresponding to a first disease of a first type when the first probability value is less than the second value.

[0100] More specifically, when the first value is set to 3 and the second value is set to 48.5, the processor 130 can generate a diagnosis result that the user is normal (or a diagnosis result that there is a low possibility that the condition corresponds to acute myocardial infarction) if the first probability value (adjusted first probability value or maintained first probability value) is less than the first value 3. Furthermore, the processor 130 can generate a diagnosis result that the user belongs to a medium risk group for acute myocardial infarction (or a diagnosis result that there is a possibility that the condition corresponds to acute myocardial infarction) if the first probability value is equal to or greater than the first value 3 and less than the second value 48.5. Finally, the processor 130 can generate a diagnosis result that the user belongs to a high risk group for myocardial infarction (or a diagnosis result that there is a high possibility that the condition corresponds to acute myocardial infarction) if the first probability value is equal to or greater than the second value 48.5.

[0101] Meanwhile, the processor 130 compares the first probability value and the second probability value with the first value and the second value, and if it is determined that the first probability value is less than the first value, it determines that the user 1 does not have myocardial infarction, and if it is determined that the second probability value is equal to or greater than the reference value, it determines that the user 1 belongs to a high-risk group for myocardial infarction. Here, the processor 130 determines that the electrocardiogram signal of the user 1 is abnormal, generates information requesting that the electrocardiogram signal of the user 1 be measured again, and can transmit this information to the external electronic device 200 via the communication interface 120 or output it via an output interface (e.g., a speaker, a display, etc.).

[0102] According to an embodiment of the present disclosure, the electronic device 100 may be implemented as an electrocardiogram measuring device. An electronic device implemented as an electrocardiogram measuring device according to an embodiment of the present disclosure will be described with reference to FIG.

[0103] FIG. 7 is a detailed block diagram of an electronic device according to an alternative embodiment of the present disclosure.

[0104] Here, the electronic device 100 includes a memory 110', a communication interface 120', a measurement unit 140', a display 150', a user interface 160', and a processor 130'. The description of the memory 110, the communication interface 120, and the processor 130 shown in Fig. 2 can be similarly applied to the memory 110', the communication interface 120', and the processor 130' shown in Fig. 7, so detailed description thereof will be omitted.

[0105] The measuring unit 140' may include a plurality of leads attached to different parts of the user 1's body and may generate an electrocardiogram signal of the user 1 based on voltage differences between the body parts of the user 1 measured through the plurality of leads. For example, the measuring unit 140' may include limb leads and chest leads. Here, the limb leads may include four electrodes attached to the limbs (hereinafter referred to as limb electrodes). And the chest leads may include six electrodes attached to the chest (hereinafter referred to as chest electrodes).

[0106] The limb electrodes may include a right arm electrode RA, a left arm electrode LA, a right leg electrode RL, and a left leg electrode LL. The right leg electrode RL may be a common electrode or a ground electrode. The limb electrodes may be attached to positions corresponding to the right arm, left arm, right leg, and left leg, respectively.

[0107] Additionally, the chest electrodes (or precordial electrodes) may include a first chest electrode V1, a second chest electrode V2, a third chest electrode V3, a fourth chest electrode V4, a fifth chest electrode V5, and a sixth chest electrode V6.

[0108] The display 150' can output various types of image information. For example, the processor 130 can output a diagnosis result for the user 1 through the display 150'. Alternatively, the processor 130 can output a first type of warning information for a first disease to the user 1 through the display 150'.

[0109] To this end, the display 150 may be implemented as a display including self-emitting elements or a display including non-self-emitting elements and a backlight, and may be implemented as various types of displays such as a liquid crystal display (LCD), an organic light-emitting diode (OLED) display, a light-emitting diode (LED), a micro light-emitting diode (micro LED), a mini light-emitting diode (Mini LED), a plasma display panel (PDP), a quantum dot (QD) display, or a quantum dot light-emitting diode (QLED).

[0110] The display 150' may also include a driving circuit and a backlight unit, which may be implemented in the form of an a-si thin film transistor (a-si TFT), a low temperature polysilicon (LTPS) TFT, an organic TFT (OTFT), etc. Meanwhile, the display 150' may be implemented as a touch screen combined with a touch sensor, a flexible display, a rollable display, a 3D display, a display in which multiple display modules are physically connected, etc.

[0111] The display 150 can also be used in conjunction with a touch panel to form a touch screen.

[0112] The user interface 160 is a component used for interaction between the electronic device 100 and the user 1, and the processor 130 can output a diagnosis result through the user interface 160'. Meanwhile, the user interface 160' may include at least one of a touch sensor, a motion sensor, a button, a jog dial, a switch, and a microphone, but is not limited thereto.

[0113] The various embodiments of the present disclosure described above can be combined with additional embodiments and can be modified within the scope that can be understood by those skilled in the art based on the above detailed description. It should be understood that the embodiments of the present disclosure are illustrative in all respects and are not limiting. For example, each component described as a single type can also be implemented in a distributed form, and similarly, components described as distributed can be implemented in a combined form. Therefore, all modifications and variations derived from the meaning, scope, and equivalent concepts of the claims of the present disclosure should be construed as being within the scope of the present disclosure.

Claims

1. 1. A method performed by an electronic device including at least one processor, comprising: acquiring biometric data of a user; inputting the acquired biological data into a first neural network model that has already been trained to generate first information about a first disease of the user; inputting the acquired biological data into a second neural network model that has already been trained to generate second information about the first disease of the user; generating a diagnosis of the user for the first type of the first disease based on the first and second information.

2. The step of generating a diagnostic result comprises:

2. The method of claim 1, further comprising: adjusting a first probability value included in the first information based on a second probability value included in the second information; and generating a diagnosis result for the user for the first disease of the first type based on the first probability value.

3. The step of generating a diagnostic result comprises: adjusting the first probability value by applying a weight corresponding to the second probability value to the first probability value if the second probability value is equal to or greater than a reference value; and maintaining the first probability value if the second probability value is less than the reference value.

4. The step of generating a diagnostic result comprises:

3. The method of claim 2, further comprising: generating a first diagnostic result corresponding to a first disease of the first type when the first probability value is greater than or equal to a first value; generating a second diagnostic result corresponding to the first disease of the first type when the first probability value is less than the first value and greater than or equal to a second value; and generating a third diagnostic result corresponding to the first disease of the first type when the first probability value is less than the second value.

5. The method of claim 1 , wherein the first information is information about the presence or absence of the first disease, and the second information is information about a first type of the first disease.

6. The method of claim 1 , further comprising extracting first and second neural network models from a plurality of neural network models based on the type of the first disease and the diagnosis result.

7. the biological data includes an electrocardiogram signal; the first disease includes any one of myocardial infarction and ischemic heart disease; The method of claim 5, wherein the first type includes any one of ST elevation myocardial infarction (STEMI) and non-ST elevation myocardial infarction (NSTEMI).

8. the first and second neural network models have already been trained based on training data to which a plurality of labels set according to different standards are assigned for the same electrocardiogram signal; The method of claim 1 , wherein the plurality of labels includes a first type label corresponding to the presence or absence of the first disease and a second type label corresponding to a first type of the first disease.

9. The method of claim 1 , wherein the steps of generating first information about the first disease and generating second information about the first disease are performed in parallel.

10. 1. An electronic device comprising: a communication interface; a memory for storing the first and second neural network models that have already been trained; and one or more processors that acquire biometric data of a user via the communication interface, input the acquired biometric data into a first neural network model that has already been trained to generate first information about a first disease of the user, input the acquired biometric data into a second neural network model that has already been trained to generate second information about the first disease of the user, and generate a diagnosis result of the user for the first disease of the first type based on the first and second information.

11. 1. A method performed by an electronic device including at least one processor, comprising: acquiring biometric data of a user; inputting the acquired biological data into a first neural network model that has already been trained to generate first information about a first disease of the user; inputting the acquired biological data into a second neural network model that has already been trained to generate second information about the first disease of the user; generating a diagnosis of the user for the first disease of the first type based on the first and second information.

Images