Method, device and program for determining patient severity from biometric data on basis of deep learning model

Abstract

The present disclosure provides a method, device and computer program for determining patient severity on the basis of a deep learning model. The method according to one embodiment of the disclosure comprises the steps of: acquiring biometric data of a patient; and, on the basis of a pre-trained first neural network model, analyzing the biometric data so as to identify patient severity, wherein the pre-trained first neural network model is trained on the basis of training data labeled according to patient conditions and treatment methods classified by patient severity.

Description

Method, device, and program for determining patient severity from biometric data based on a deep learning model

[0001] The present disclosure relates to deep learning technology in the medical field, and more specifically, to a device and method for diagnosing the severity of a patient's condition by analyzing the patient's biometric data using artificial intelligence technology.

[0002] Electrocardiogram (ECG) data is a time-lapse recording of the heart's electrical activity, accurately reflecting physiological conditions such as heart rate and rhythm. ECG data can be obtained noninvasively, and its high accessibility and reliability make it useful for real-time monitoring of patient status. In particular, ECG data can be easily acquired using portable devices or in-hospital equipment, making it highly practical for collecting data for diagnosis and treatment.

[0003] In healthcare settings, the need for technologies that support precise and rapid decision-making in assessing patient severity is growing. Existing severity assessment methods rely heavily on the experience of medical professionals or involve complex processes requiring the comprehensive analysis of numerous diagnostic data. This process can hinder effective response in emergency situations or environments with limited medical resources. Therefore, the development of a system capable of monitoring patient severity in real time and quantitatively assessing it based on electrocardiogram data is required.

[0004] The present disclosure is conceived in response to the aforementioned background technology, and aims to provide a method, device, and program for determining the severity of a patient's condition from biometric data based on a deep learning model.

[0005] However, the problems to be solved in this disclosure are not limited to the problems mentioned above, and other problems not mentioned can be clearly understood based on the description below.

[0006] A method for determining the severity of a patient based on a deep learning model, performed by a computing device including at least one processor according to an embodiment of the present disclosure for realizing the task as described above, comprises the steps of: acquiring biometric data of a patient; and analyzing the biometric data based on a pre-learned first neural network model to determine the severity of the patient, wherein the pre-learned first neural network model is characterized in that it is learned based on learning data to which labels corresponding to the condition and treatment method of the patient are assigned, which are distinguished according to the severity of the patient.

[0007] Alternatively, the labels corresponding to the patient's condition and treatment method, which are distinguished according to the severity of the patient, are characterized by being distinguished as the patient's normal condition, the patient's outpatient treatment, the patient's admission to a general ward, the patient's admission to an intensive care unit, and the patient's condition within a preset time from cardiac arrest.

[0008] Alternatively, the step of determining the severity of the patient includes the step of inputting the biometric data into the first neural network model that has been previously learned to extract feature information corresponding to the condition of the patient, obtaining a first score corresponding to the severity of the patient based on the extracted feature information, and determining the severity of the patient based on the obtained first score.

[0009] Alternatively, the step of determining the severity of the patient based on the acquired first score includes a step of determining the severity of the patient as normal if the acquired first score is less than a first value, determining the severity of the patient as a severity that can be treated as an outpatient if the acquired first score is greater than or equal to the first value and less than a second value, determining the severity of the patient as a severity that requires hospitalization if the acquired first score is greater than or equal to the second value and less than a third value, determining the severity of the patient as a severity that requires intensive care if the acquired first score is greater than or equal to the third value, and determining the severity of the patient as a state within a preset time before cardiac arrest if the acquired first score is greater than or equal to the fourth value.

[0010] Alternatively, the labels corresponding to the patient's condition and treatment method, which are distinguished according to the severity of the patient, include a fifth value corresponding to the patient's normal condition, a sixth value corresponding to outpatient treatment for the patient, a seventh value corresponding to the patient's admission to a general ward, an eighth value corresponding to the patient's admission to an intensive care unit, and a ninth value corresponding to the patient's condition within a preset time from cardiac arrest.

[0011] Alternatively, the first to fourth values ​​are characterized in that they are set by adjusting the fifth to ninth values ​​to an average value of the first score corresponding to the normal state of the patient determined based on biometric data acquired from the patient in the past.

[0012] Alternatively, the method includes analyzing the biometric data based on a pre-trained second neural network model to obtain a second score for the possibility of the patient's disease, and determining the severity of the patient's disease based on the first and second scores.

[0013] Alternatively, the step of determining the severity of the patient according to the disease includes a step of determining the severity of the patient according to the disease by determining a correlation between the disease and the severity of the patient based on the first score, the second score, and the disease type information of the patient.

[0014] Alternatively, the first and second neural network models learned above are characterized by sharing a network that extracts feature information corresponding to the patient's condition from the biometric data.

[0015] In order to realize the task as described above, a computing device for determining the severity of a patient based on a deep learning model according to an embodiment of the present disclosure comprises a processor including at least one core, a memory including a program code executable by the processor, and a communication interface, wherein the processor obtains biometric data of the patient and analyzes the biometric data based on a pre-learned first neural network model to determine the severity of the patient, and the pre-learned first neural network model is learned based on learning data to which labels corresponding to the condition and treatment method of the patient are assigned, which are distinguished according to the severity of the patient.

[0016] A computer program stored in a computer-readable storage medium according to one embodiment of the present disclosure for realizing the task as described above, wherein the computer program, when executed on one or more processors, performs an operation for determining the severity of a patient based on a deep learning model, the operation including an operation for obtaining biometric data of the patient, and an operation for analyzing the biometric data based on a first neural network model that has been previously learned to determine the severity of the patient, wherein the first neural network model that has been previously learned is learned based on learning data that has been assigned labels corresponding to the condition and treatment method of the patient that are classified according to the severity of the patient.

[0017] A method for assessing a patient's severity from biometric data based on a deep learning model, according to one embodiment of the present disclosure, provides a technique for effectively analyzing a patient's biometric data to quickly and accurately assess severity. This allows for real-time monitoring of the patient's condition and helps determine appropriate treatment options based on severity.

[0018] Additionally, it overcomes the limitations of existing severity assessment methods that rely on medical staff experience or qualitative judgment, and provides more objective and consistent results by quantitatively analyzing biometric data using a deep learning model.

[0019] FIG. 1 is a block diagram of a computing device according to an embodiment of the present disclosure.

[0020] FIG. 2 is a flowchart of a method for determining the severity of a patient based on a deep learning model according to an embodiment of the present disclosure.

[0021] FIG. 3 is an exemplary diagram showing learning data used to train a first neural network model for determining severity according to one embodiment of the present disclosure.

[0022] FIG. 4 is a block diagram of a second neural network model according to an embodiment of the present disclosure.

[0023] FIG. 5 is an example diagram of a first neural network model and a second neural network model according to an embodiment of the present disclosure.

[0024] FIG. 6 is a detailed configuration diagram of a computing device according to another embodiment of the present disclosure.

[0025] Below, embodiments of the present disclosure are described in detail with reference to the attached 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 utilize or implement the contents of the present disclosure. Accordingly, 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 implemented in various different forms and is not limited to the embodiments described below.

[0026] Throughout the specification of this disclosure, identical or similar drawing numbers refer to identical or similar components. Furthermore, for clarity in the description of this disclosure, drawing numbers for parts unrelated to the description of this disclosure may be omitted in the drawings.

[0027] The term "or" as used herein is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified herein or clear from context, "X employs A or B" should be understood to mean either of the natural inclusive permutations. For example, unless otherwise specified herein or clear from context, "X employs A or B" can be interpreted to mean either X employs A, X employs B, or X employs both A and B.

[0028] The term "and / or" as used herein should be understood to refer to and include all possible combinations of one or more of the related concepts listed.

[0029] The terms "comprises" and / or "comprising" as used herein should be understood to mean the presence of certain features and / or components. However, it should be understood that the terms "comprises" and / or "comprising" do not exclude the presence or addition of one or more other features, other components, and / or combinations thereof.

[0030] Unless otherwise specified in this disclosure or unless the context makes it clear that the singular form is intended to be referred to, the singular should generally be construed to include “one or more.”

[0031] The term "Nth (N is a natural number)" used in the present disclosure can be understood as an expression used to mutually distinguish components of the present disclosure based on a predetermined standard such as a functional perspective, a structural perspective, or convenience of explanation. For example, components performing different functional roles in the present disclosure can be distinguished as a first component or a second component. However, components that are substantially the same within the technical spirit of the present disclosure but must be distinguished for convenience of explanation may also be distinguished as a first component or a second component.

[0032] The term "acquisition" as used in this disclosure may be understood to mean not only receiving data through a wired or wireless communication network with an external device or system, but also generating data in an on-device form.

[0033] Meanwhile, the term "module" or "unit" used in the present disclosure can be understood as a term referring to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a part thereof, hardware or a part thereof, or a combination of software and hardware. At this time, the "module" or "unit" may be a unit composed of a single element, or a unit expressed as a combination or set of multiple elements. For example, as a narrow concept, a "module" or "unit" may refer to a hardware element of a computing device or a set thereof, an application program that performs a specific function of software, a processing process implemented through software execution, or a set of instructions for program execution, etc. In addition, as a broad concept, a "module" or "unit" may refer to the computing device itself that constitutes the system, or an application running on the computing device, etc. However, since the above-described concept is only an example, the concept of “module” or “part” may be defined in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0034] The term "model" as used herein may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a set of software units to solve a specific problem, or an abstract model of a processing process to solve a specific problem. For example, a neural network "model" may refer to the entire system implemented as a neural network that has problem-solving capabilities through learning. In this case, the neural network can have problem-solving capabilities by optimizing the parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network or a set of neural networks that are a combination of multiple neural networks.

[0035] The term "data" used in this disclosure may include "images," signals, and the like. The term "image" used in this disclosure may refer to multidimensional data composed of discrete image elements. In other words, "image" may be understood as a term referring to a digital representation of an object visible to the human eye. For example, "image" may refer to multidimensional data composed of elements corresponding to pixels in a two-dimensional image. "Image" may refer to multidimensional data composed of elements corresponding to voxels in a three-dimensional image.

[0036] The explanation of the above terms is intended to aid understanding of the present disclosure. Therefore, unless explicitly stated as limiting the contents of the present disclosure, it should be noted that the above terms are not intended to limit the technical ideas of the contents of the present disclosure.

[0037] FIG. 1 is a block diagram of a computing device according to an embodiment of the present disclosure.

[0038] A computing device (100) according to an embodiment of the present disclosure may be a hardware device or a part of a hardware device that performs comprehensive processing and calculation of data, or may be a software-based computing environment connected to a communication network. For example, the computing device (100) may be a server that performs intensive data processing functions and shares resources, or may be a client that shares resources through interaction with a server. In addition, the computing device (100) may be a cloud system in which multiple servers and clients interact to comprehensively process data. Since the above description is only one example related to the type of computing device (100), the type of computing device (100) may be configured in various ways within a category understandable to those skilled in the art based on the contents of the present disclosure. Meanwhile, the computing device (100) may be implemented as various electronic devices such as a desktop, a laptop, a smartphone, a smart watch, a smart ring, and a server.

[0039] Referring to FIG. 1, a computing device (100) according to an embodiment of the present disclosure may include a processor (110), a memory (120), and a network unit (130). However, FIG. 1 is merely an example, and the computing device (100) may include other components for implementing a computing environment. In addition, only some of the disclosed components may be included in the computing device (100).

[0040] A processor (110) according to an embodiment of the present disclosure may be understood as a configuration unit including hardware and / or software for performing computing operations. For example, the processor (110) may read a computer program to perform data processing for machine learning. The processor (110) may process computational processes such as processing input data for machine learning, feature extraction for machine learning, and error calculation based on backpropagation. The processor (110) 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), or a field programmable gate array (FPGA). The above-described type of processor (110) is only one example, and thus, the type of processor (110) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0041] The processor (110) is connected to other components of the computing device (100) (i.e., memory (120) and communication interface (130)) and controls the overall operation of the computing device (100).

[0042] The memory (120) according to one embodiment of the present disclosure may be understood as a configuration unit including hardware and / or software for storing and managing data processed in the computing device (100). That is, the memory (120) may store any type of data generated or determined by the processor (110) and any type of data received by the communication interface (130). For example, the memory (120) may include at least one type of storage medium among a flash memory 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. In addition, the memory (120) may also include a database system that controls and manages data in a predetermined system. The type of memory (120) described above is only one example, and thus the type of memory (120) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0043] The memory (120) can structure and organize and manage data, combinations of data, and program codes executable by the processor (110) required for the processor (110) to perform operations. For example, the memory (120) can store bio-data (aortic blood pressure data, brain wave data, and electrocardiogram data) received through a communication interface (130) to be described later. In addition, the memory (120) can store a neural network model learned to determine the severity of a patient or a neural network model learned to determine the responsiveness of a patient to a possibility of a disease, and can store program codes that operate to perform learning of each neural network model, program codes that operate the neural network model to receive bio-data (aortic blood pressure data, brain wave data, and electrocardiogram data, etc.) and perform inference according to the purpose of use of the computing device (100), and processed data generated as the program codes are executed.

[0044] A communication interface (130) according to an embodiment of the present disclosure may be understood as a component that transmits and receives data through any known wired or wireless communication system. For example, the communication interface (130) may perform data transmission and reception 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), fifth generation mobile communication (5G), ultrawide-band, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity, near field communication (NFC), or Bluetooth. Since the above-described communication systems are only examples, the wired and wireless communication system for data transmission and reception of the communication interface (130) may be applied in various ways other than the above-described examples.

[0045] The communication interface (130) can receive data necessary for the processor (110) to perform calculations through wired or wireless communication with any system or any client, etc. In addition, the communication interface (130) can transmit data generated through calculations of the processor (110) through wired or wireless communication with any system or any client, etc. For example, the communication interface (130) can receive a patient's biometric data through communication with a cloud server that performs tasks such as standardization of databases and medical data in a hospital environment, or a computing device, etc. The communication interface (130) can transmit output data of a neural network model, intermediate data derived from the calculation process of the processor (110), processed data, etc. through communication with the aforementioned database, server, or computing device, etc.

[0046] FIG. 2 is a flowchart of a method for determining the severity of a patient based on a deep learning model according to an embodiment of the present disclosure.

[0047] Referring to FIG. 2, the processor (110) acquires the patient's biometric data (S210). The biometric data may include electrocardiogram data. Specifically, the processor (110) may acquire the patient's electrocardiogram data acquired by an external biometric signal measuring device connected to the computing device (100). The processor (110) may receive the electrocardiogram data from the external biometric signal measuring device through the communication interface (130). In addition, the processor (110) may directly detect an electrical signal generated from the patient's heart through a sensing unit of the computing device (100) to acquire the patient's electrocardiogram data. To this end, the sensing unit includes one or more electrodes, and the processor (110) may measure the patient's electrocardiogram by attaching one or more electrodes to the patient's body, thereby acquiring the electrocardiogram data from the patient. At this time, the processor (110) can obtain a 1-lead electrocardiogram for the patient using one electrode, or can obtain various forms of electrocardiogram data such as 3-lead, 6-lead, and 12-lead using multiple electrodes.

[0048] Meanwhile, the present invention is not limited thereto, and the processor (110) may acquire various bio-data that can be obtained from the patient in addition to electrocardiogram data. For example, the processor (110) may acquire various bio-data such as the patient's body temperature, photoreceptor blood flow, heart rate, electroencephalogram data, etc. This may be determined based on the type of bio-data that constitutes the learning data used to train the neural network model to identify the severity of the patient. The sensing unit may include a temperature sensor, an image sensor, or a light sensor to acquire various bio-data of the patient.

[0049] The processor (110) analyzes the acquired biometric data using a pre-trained neural network model and can determine the patient's severity (S220). The severity includes arbitrary information that quantifies the patient's critical condition. The patient's severity may be categorized into multiple classes or grades. For convenience of explanation of the present disclosure, the neural network model trained to determine the patient's severity will be referred to as the first neural network model, and the biometric data will be described as electrocardiogram data.

[0050] The processor (110) can analyze electrocardiogram data using a first neural network model trained based on training data including multiple electrocardiogram data. The processor (110) can extract feature information related to the patient's severity from the electrocardiogram data using the first neural network model, and determine the patient's severity based on the extracted feature information. To this end, the trained first neural network model can be trained based on training data labeled with patient conditions and treatment methods, which are distinguished according to the patient's severity.

[0051] That is, patient conditions and treatment methods appropriate for each patient can be differentiated based on the severity of the condition. Labeled training data related to these patient conditions and treatment methods can be used to train the first neural network model. The labels include labels related to treatment methods and labels indicating the condition of a specific patient.

[0052] The processor (110) may obtain training data including input data composed of multiple electrocardiogram data obtained from different patients and label data in which labels indicating the severity level (or class) of each patient are assigned, and may train a first neural network model in advance using the obtained training data. The training data may include feature information (e.g., P wave, QRS complex, T wave, etc.) extracted from an electrocardiogram signal (more specifically, an electrocardiogram signal corresponding to the electrocardiogram data). The processor (110) may train the first neural network model to determine the severity level of the patient based on the feature information. The processor (110) may input the training data into the first neural network model, and may calculate a loss function based on the difference between the output value of the first neural network model and the label data during the training process. The loss function may be defined as a cross-entropy loss or an objective function that optimizes the balance between precision and recall. Based on the calculated loss function, the processor (110) can adjust the weights of the neural network model through backpropagation. By repeating this process, the processor (110) can improve the classification performance of the first neural network model with respect to the severity level, and ultimately obtain a first neural network model trained to identify the severity level of a patient based on electrocardiogram data.

[0053] Meanwhile, the electrocardiogram signal corresponding to the electrocardiogram data included in the training data used to train the first neural network model may have a commonly preset length. Accordingly, when the processor (110) acquires electrocardiogram data from a patient, the processor (110) may divide the electrocardiogram signal corresponding to the acquired electrocardiogram data into preset lengths or divide the electrocardiogram signal into multiple segments by applying a window of preset length along the time axis. For example, the electrocardiogram signal may be measured in 12 leads of 10 seconds in length and may be a signal measured at 500 points per second.

[0054] The first neural network model can be implemented as a multi-layer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), or generative adversarial network.

[0055] FIG. 3 is an exemplary diagram showing learning data used to train a first neural network model for determining severity according to one embodiment of the present disclosure.

[0056] According to one embodiment of the present disclosure, labels corresponding to the patient's condition and treatment method, which are distinguished according to the severity of the patient, may be distinguished as follows: 1) normal condition of the patient, 2) outpatient treatment for the patient, 3) admission of the patient to a general ward, 4) admission of the patient to an intensive care unit, and 5) condition of the patient within a preset time from cardiac arrest. Specifically, referring to FIG. 3, the processor (110) can provide input data composed of a plurality of electrocardiogram data (20-1) acquired from a plurality of patients (10-1) in a normal state, a plurality of electrocardiogram data (20-2) acquired from a plurality of patients (10-2) receiving outpatient treatment, a plurality of electrocardiogram data (20-3) acquired from a plurality of patients (10-3) hospitalized and receiving treatment in a general ward, a plurality of electrocardiogram data (20-4) acquired from a plurality of patients (40-1) hospitalized and receiving treatment in an intensive care unit, and a plurality of electrocardiogram data (20-5) acquired from a plurality of patients (10-5) who are within a preset time (e.g., 30 minutes) from cardiac arrest. In addition, the processor (110) can prepare label data to which each label (i.e., a normal state of the patient, an outpatient treatment label for the patient, a general ward admission label for the patient, an intensive care unit admission label for the patient, and a patient status label within a preset time from cardiac arrest) is assigned in response to the input data. In addition, the processor (110) can train the first neural network model (30) using training data composed of the input data and the label data.

[0057] Meanwhile, labels corresponding to the patient's condition and treatment method, which are distinguished according to the severity of the patient according to one embodiment of the present disclosure, may be set to specific values. The labels may include a fifth value corresponding to the patient's normal condition, a sixth value corresponding to outpatient treatment for the patient, a seventh value corresponding to the patient's admission to a general ward, an eighth value corresponding to the patient's admission to an intensive care unit, and a ninth value corresponding to the patient's condition within a preset time from cardiac arrest. For example, as illustrated in FIG. 3, an electrocardiogram acquired from a patient in a normal condition may be assigned a label of "0," an electrocardiogram acquired from a patient receiving outpatient treatment may be assigned a label of "0.1," an electrocardiogram acquired from a patient receiving inpatient treatment in a general ward may be assigned a label of "0.4," an electrocardiogram acquired from a patient receiving treatment in an intensive care unit may be assigned a label of "0.8," and an electrocardiogram acquired from a patient within a preset time from cardiac arrest may be assigned a label of "1." However, this is just an example and the specific values ​​set as labels can be set in various ways.

[0058] According to one embodiment of the present disclosure, the processor (110) may input the acquired electrocardiogram data into a pre-trained first neural network model (30) to obtain a score regarding the severity of the patient. For example, the processor (110) may input the acquired electrocardiogram data from the patient into the pre-trained first neural network model (30) to obtain a score regarding the severity of the patient. In this case, the score may be a probability value regarding the severity output from the first neural network model (30). Hereinafter, the score acquired from the first neural network model (30) is referred to as the first score.

[0059] And, the processor (110) determines the patient's severity as normal if the acquired first score is less than the first value, determines the patient's severity as severe enough to be treated as an outpatient, determines the patient's severity as severe enough to be treated as an outpatient, determines the patient's severity as severe enough to require hospitalization if the acquired first score is greater than or equal to the second value, determines the patient's severity as severe enough to require intensive care if the acquired first score is greater than or equal to the third value, determines the patient's severity as severe enough to require intensive care, and determines the patient's severity as severe enough to be treated as an emergency within a pre-set time before cardiac arrest if the acquired first score is greater than or equal to the fourth value. Here, the first to fourth values ​​may be reference values ​​for determining the patient's severity level (and the patient's condition set according to the severity level). The first to fourth values ​​may be determined according to specific values ​​set as labels.

[0060] In addition, the first to fourth values ​​may be set as the average first score value of a plurality of other patients having similar biometric information to the patient. In particular, the processor (110) may adjust and set the fifth to ninth values ​​as the average value of the first score corresponding to the normal state of the patient identified based on biometric data acquired from the patient in the past. That is, the processor (110) may repeatedly obtain the first score of the patient in the normal state from the first neural network model (30) based on biometric data repeatedly acquired from the patient in the past, and adjust the fifth to ninth values ​​as the average value of the repeatedly acquired first scores to set the first to fourth values ​​as reference values. For example, the processor (110) may first adjust the fifth to ninth values ​​as the average first score value of a plurality of other patients having similar biometric information to the patient, and then secondarily adjust the fifth to ninth values ​​adjusted by the average value of the acquired first scores to set reference values ​​(the first to fourth values) suitable for the user. And, the processor (110) can set the first to fourth values ​​to an appropriate range including the adjusted fifth to ninth values.

[0061] Additionally, according to one embodiment of the present disclosure, the processor (110) analyzes biometric data based on a neural network model trained to determine whether a patient has a disease, obtains a score for the likelihood of the patient having a disease, and may predict the patient's disease based on the obtained score. The higher the obtained score, the more likely the processor (110) is to determine that the patient has a higher likelihood of developing a disease or to predict that the disease has occurred. Hereinafter, the neural network model trained to determine whether a patient has a disease is referred to as a second neural network model, and the score obtained from the second neural network model is referred to as a second score.

[0062] The patient's disease includes various diseases such as heart failure, hypertension, and left ventricular systolic dysfunction, and the second neural network model can be trained to identify a specific disease based on the disease type labeled in the training data.

[0063] The processor (110) can analyze electrocardiogram data using a second neural network model trained based on learning data including a plurality of biometric data (i.e., electrocardiogram data). The processor (110) can extract feature information related to a specific disease from the electrocardiogram data using the second neural network model, and determine whether the patient has a disease based on the extracted feature information.

[0064] To this end, the processor (110) may obtain training data including input data composed of multiple electrocardiogram data obtained from different patients and label data in which labels indicating whether each patient has a specific disease (e.g., left ventricular systolic dysfunction), and may train a second neural network model in advance using the obtained training data. The training data may include feature information (e.g., P wave, QRS complex, T wave, etc.) extracted from an electrocardiogram signal (more specifically, an electrocardiogram signal corresponding to the electrocardiogram data). The processor (110) may train the second neural network model to determine the possibility of a specific disease in the patient based on such feature information. The processor (110) may input the training data into the second neural network model, and may calculate a loss function based on the difference between the output value of the second neural network model and the label data during the training process. The loss function may be defined as a cross-entropy loss or an objective function that optimizes the balance between precision and recall. Based on the calculated loss function, the processor (110) can adjust the weights of the model through backpropagation. By repeating this process, the processor (110) can improve the classification performance of the second neural network model for a specific disease, and finally obtain a second neural network model trained to determine whether a patient has a specific disease based on electrocardiogram data and to produce a score (or probability value) for the disease.

[0065] Meanwhile, the electrocardiogram signals corresponding to the electrocardiogram data included in the training data used to train the second neural network model may have a commonly preset length. In this regard, the aforementioned embodiments of the present disclosure may be commonly applied.

[0066] Meanwhile, the second neural network model can be implemented as a multi-layer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), or generative adversarial network.

[0067] According to one embodiment of the present disclosure, the processor (110) may input acquired electrocardiogram data into a pre-trained second neural network model to obtain a score regarding a specific disease of the patient. For example, the processor (110) may input acquired electrocardiogram data from the patient into the pre-trained neural network model to obtain a score regarding the possibility of the patient having heart failure (hereinafter referred to as a second score). Then, the processor (110) may determine the possibility of the patient having a specific disease based on the acquired score. In particular, if the processor (110) determines that the acquired second score is equal to or greater than a reference value, the processor (110) may determine that the patient has heart failure. On the other hand, if the acquired score is less than the reference value, the processor (110) may determine that the patient does not have heart failure, i.e., the patient is in a normal condition.

[0068] A reference value, which serves as a criterion for determining whether a specific disease, for example, heart failure, exists, can be set differently for each patient. To this end, the processor (110) can repeatedly obtain electrocardiogram data for the patient, analyze the obtained electrocardiogram data, and set a reference value by reflecting the patient's constitution, health status, age, gender, etc. In particular, the processor (110) can input electrocardiogram data obtained from a patient in a normal state into a pre-trained second neural network model to obtain a second score, and set a reference value for determining whether heart failure exists based on the obtained second score. For example, the processor (110) can repeatedly obtain electrocardiogram data from a wearable device (e.g., a smart watch, a smart ring, etc.) worn by the patient, analyze the patient's electrocardiogram in a normal state based on the obtained electrocardiogram data, and set a reference value that takes into account the patient's health status, etc. Meanwhile, it goes without saying that the computing device (100) can be implemented as the above-described wearable device.

[0069] According to one embodiment of the present disclosure, the processor (110) may input acquired electrocardiogram data and the patient's biometric information into a pre-trained second neural network model to obtain a score regarding a specific disease of the patient. Here, the biometric information may be at least one of the patient's age, gender, height, and weight. To this end, the second neural network model may include a fully connected layer into which the patient's biometric information is input. To this end, the training data may include the patient's biometric information from which each electrocardiogram data was acquired.

[0070] FIG. 4 is a block diagram of a second neural network model according to an embodiment of the present disclosure.

[0071] Referring to FIG. 4, the second neural network model may include a neural network composed of at least one residual block into which electrocardiogram data is input. The residual block may include a plurality of sub-modules, each of which may include a plurality of convolutional neural network (CNN), batch normalization, and ReLU activation function layers, and may include a dropout layer. The first sub-module illustrated in FIG. 4 may include a Maxpooling layer that directly inputs normalized electrocardiogram data to the last ReLU activation function layer. In addition, the second neural network model may include a fully connected layer into which auxiliary information such as age, gender, height, and weight is input. The output of the fully connected layer and the output of the residual block may be concatenated to derive the probability of onset of cardiomyopathy before and after delivery. The electrocardiogram data input to the first neural network model (30) can be processed prior to downsampling and noise application augmentation and then input to the residual block. For example, the sampling rate of the electrocardiogram data was downsampled from 500 Hz to 250 Hz, and data augmentation using various noises was used.

[0072] Meanwhile, the second neural network model may include multiple neural networks corresponding to the number of leads used to acquire electrocardiogram data. That is, the second neural network model may include individual neural networks into which electrocardiogram data measured by individual leads are respectively input. For example, the second neural network model may include a 2-1 sub-neural network model trained based on electrocardiogram data measured by 12 multiple leads. In addition, the first neural network model (30) may further include a 2-2 sub-neural network model trained based on electrocardiogram data measured by six limb leads or six precordial leads. In addition, the first neural network model (30) may further include a 2-3 sub-neural network model trained based on electrocardiogram data measured by a single lead.

[0073] The second neural network model may selectively use at least one of the 2-1 sub-neural network model, the 2-2 sub-neural network model, or the 2-3 sub-neural network model depending on the number of leads used to acquire electrocardiogram data. In this case, the 2-1 to 2-3 sub-neural network models may be trained using training data composed of electrocardiogram data corresponding to the measurement method (i.e., the number of leads) used to acquire each electrocardiogram data.

[0074] Additionally, the second neural network model may include multiple sub-neural network models depending on the disease type. For example, the second neural network model may include at least one of a sub-neural network model (sub-neural network model 2-4) that determines whether a patient has heart failure based on electrocardiogram data, a sub-neural network model (sub-neural network model 2-5) that determines whether a patient has left ventricular systolic dysfunction, and a sub-neural network model (sub-neural network model 2-6) that determines whether a patient has high blood pressure.

[0075] At this time, each sub-neural network model (sub-neural network models 2-4 to 2-6) can be pre-trained with training data consisting of label data corresponding to each disease type. The above explanation applies equally to this, so a detailed explanation will be omitted.

[0076] Meanwhile, the processor (110) can determine the severity of a patient's disease based on the first and second scores. That is, the processor (110) can determine the severity as being manifested according to the disease identified based on the second score. In particular, the processor (110) can obtain a second score from a sub-neural network model corresponding to each disease type, and can identify the disease causing the patient's severity based on the obtained plurality of second scores.

[0077] At this time, the processor (110) can determine the correlation between the disease and the disease type based on the first score, the second score, and the disease type corresponding to the second score. In particular, the processor (110) can determine that the higher the second score, the greater the disease influence on the patient's severity, and thus determine that the degree of correlation is high. At this time, the processor can determine the correlation between the second score and the first score by taking the disease type into consideration. That is, even if the second score is the same, the correlation with the severity may be determined differently depending on the disease type. To this end, the processor can input the first and second scores and the disease type information into a pre-trained neural network model (hereinafter, referred to as a third neural network model) to produce an index indicating the correlation. However, the present invention is not limited thereto, and the processor (110) can also obtain the correlation index based on a machine learning model (e.g., a logistic regression model, etc.), a pre-generated table (e.g., a look-up table, etc.) regarding the correlation between the first and second scores and the disease type information.

[0078] FIG. 5 is an example diagram of a first neural network model (30) and a second neural network model according to one embodiment of the present disclosure.

[0079] To this end, a network that extracts feature information corresponding to a patient's condition from biometric data according to an embodiment of the present disclosure may be shared. For example, the first neural network model (30) and the second neural network model may share a portion (partial neural network) of a neural network model that extracts feature information, and may each include a classifier that classifies the patient's severity and disease likelihood based on the extracted feature information. For example, referring to FIG. 5, the first neural network model (30) and the second neural network model (40) may share a neural network (50) that extracts feature information from electrocardiogram data. In this case, the cause of the severity identified by the first score output by the first neural network model (30) may be determined to be a disease identified based on the second score output by the second neural network model (30). Meanwhile, the first neural network model (30) may include a plurality of sub-neural network models trained to identify the severity associated with each disease type, and at this time, each sub-neural network model may share a neural network that extracts feature information with a sub-neural network model (e.g., the 2nd-4th to 2nd-6th sub-neural network models) corresponding to each disease type. At this time, the processor may determine a disease type having the highest score among a plurality of second scores obtained through the plurality of second sub-neural network models, and may identify a disease affecting the severity and a severity corresponding to the determined disease type as the corresponding disease type. However, a weight according to the disease type may be applied to each of the plurality of second scores, and then a disease type having a high score may be determined.

[0080] In addition, according to one embodiment of the present disclosure, the processor (110) can identify the patient's condition by dividing the time before cardiac arrest into multiple time units along with the patient's severity. For example, the processor (110) can identify the patient's condition as within the first hour from cardiac arrest, within the second hour from cardiac arrest, within the third hour from cardiac arrest, within the fourth hour from cardiac arrest, and within the fifth hour from cardiac arrest using the fourth neural network model. To this end, the processor (110) can train the fourth neural network model using training data composed of electrocardiogram data acquired from multiple patients corresponding to each time unit. The description of the present disclosure described above can be equally applied with respect to the training method of the fourth neural network model. The fourth neural network model can be trained by extracting only feature information related to cardiac arrest from the electrocardiogram data. This is to focus on and predict only the patient's cardiac arrest, i.e., death, unlike the first neural network model (30) related to the patient's severity.

[0081] The remaining time until cardiac arrest and whether or not a patient has cardiac arrest based on the fourth neural network model can be performed on a patient whose severity is higher than that of an intensive care unit admission. That is, if the processor (110) determines that the patient's severity is higher than that of an intensive care unit admission based on the first score, the processor (110) can monitor the patient's condition using the fourth neural network model. Meanwhile, the processor (110) can determine and manage the possibility of the patient's cardiac arrest and the number of hours from cardiac arrest based on the score (third score) obtained from the fourth neural network model. The reference value applied to the third score can be applied in the same way as that for the first score, so a detailed description thereof will be omitted.

[0082] Meanwhile, the fourth neural network model can be implemented as a multi-layer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), or generative adversarial network.

[0083] FIG. 6 is a detailed configuration diagram of a computing device (600) according to another embodiment of the present disclosure.

[0084] Referring to FIG. 6, a computing device (600) according to an embodiment of the present disclosure includes a processor (610), a memory (620), a communication interface (630), a display (640), a user interface (650), a sensing unit (660), and a speaker (670). Among the configurations illustrated in FIG. 6, a detailed description of configurations that overlap with those illustrated in FIG. 1 will be omitted.

[0085] The display (640) can display various images. Here, the images include both still images and moving images. The display (640) can output an electrocardiogram (ECG) signal, and can also output a noise score, noise level, etc. of the ECG signal. The display (640) can be implemented as a display in various forms, such as an LCD (Liquid Crystal Display Panel), an OLED (Organic Light Emitting Diodes), an LCoS (Liquid Crystal on Silicon), a DLP (Digital Light Processing), etc. In addition, the display (640) can also include a driving circuit, a backlight unit, etc., which can be implemented in a form, such as an a-si TFT, an LTPS (low temperature poly silicon) TFT, an OTFT (organic TFT), etc. Meanwhile, the display (640) can be implemented as a touch screen by being combined with a touch panel. In this case, the display (640) can perform the function of not only an output interface that outputs an image through the touch screen, but also an input interface that receives a user's touch input. The display (640) can display the patient's severity level, first score, second score, etc.

[0086] The user interface (650) is a configuration used by the computing device (600) to perform interaction with the user, and may include, but is not limited to, at least one of a touch sensor, a motion sensor, a button, a jog dial, and a switch.

[0087] The sensing unit (660) senses the patient's bio-signals to obtain bio-data. For example, the patient's electrocardiogram data can be obtained by detecting the electrical signal of the patient's heartbeat through at least one electrode included in the sensing unit (660).

[0088] The speaker (680) is a component that outputs various audio data that have undergone various processing operations, such as decoding, amplification, and noise filtering, by an audio processing unit (not shown). The speaker (680) can output various notification sounds or voice messages. According to one embodiment of the present disclosure, the processor (910) can convert an electrical signal into a user's (1) voice and output it through the speaker (680). As an example, the speaker (680) can output a voice message that informs of the patient's severity, the time remaining until cardiac arrest, and the patient's emergency condition.

[0089] Meanwhile, according to one embodiment of the present disclosure, a non-transitory computer-readable medium storing a program for performing the method for assessing the severity of a patient based on artificial intelligence as described above may be provided. Here, the non-transitory computer-readable medium refers to a medium that permanently stores data and can be read by a device, rather than a medium that stores data for a short period of time, such as a register, cache, or memory. Specifically, the various applications or programs described above may be stored and provided on a non-transitory computer-readable medium, such as a CD, DVD, hard disk, Blu-ray disc, USB, memory card, or ROM.

[0090] The various embodiments of the present disclosure described above can be combined with additional embodiments and modified within the scope understood by those skilled in the art in light of the detailed description above. It should be understood that the embodiments of the present disclosure are illustrative in all respects and not restrictive. For example, each component described as a single component may be implemented in a distributed manner, and likewise, components described as distributed may be implemented in a combined manner. Accordingly, all changes or modifications derived from the meaning, scope, and equivalent concepts of the claims of the present disclosure should be construed as being included within the scope of the present disclosure.

Claims

1. A method for determining the severity of a patient based on a deep learning model, performed by a computing device including at least one processor, A step of acquiring the patient's biometric data; and A step of analyzing the biometric data to determine the severity of the patient based on the first neural network model that has been learned; The above-mentioned first neural network model is characterized in that it is learned based on learning data that is assigned labels corresponding to the patient's condition and treatment method classified according to the patient's severity. method.

2. In paragraph 1, Labels corresponding to the patient's condition and treatment method, which are classified according to the severity of the patient's condition, are as follows: Distinguished by the patient's normal condition, outpatient treatment for the patient, admission of the patient to the general ward, admission of the patient to the intensive care unit, and the patient's condition within a preset time from cardiac arrest. method.

3. In paragraph 2, The steps to determine the severity of the above patient are: A step of inputting the biometric data into the first neural network model that has been previously learned to extract feature information corresponding to the patient's condition, obtaining a first score corresponding to the severity of the patient based on the extracted feature information, and determining the severity of the patient based on the obtained first score; method.

4. In paragraph 3, The step of determining the severity of the patient based on the first score obtained above is: If the acquired first score is less than the first value, the patient's severity is determined to be normal, if the acquired first score is greater than or equal to the first value and less than the second value, the patient's severity is determined to be a severity that can be treated as an outpatient, if the acquired first score is greater than or equal to the second value and less than the third value, the patient's severity is determined to be a severity that requires hospitalization, if the acquired first score is greater than or equal to the third value and less than the fourth value, the patient's severity is determined to be a severity that requires intensive care, and if the acquired first score is greater than or equal to the fourth value, the patient's severity is determined to be within a pre-set time before cardiac arrest. method.

5. In paragraph 4, Labels corresponding to the patient's condition and treatment method, which are classified according to the severity of the patient's condition, are as follows: Including a fifth value corresponding to the patient's normal state, a sixth value corresponding to outpatient treatment for the patient, a seventh value corresponding to the patient's admission to a general ward, an eighth value corresponding to the patient's admission to an intensive care unit, and a ninth value corresponding to the patient's state within a preset time from cardiac arrest. method.

6. In paragraph 5, The first to fourth values are, The fifth to ninth values are set by adjusting the average value of the first score corresponding to the normal state of the patient, which is determined based on biometric data obtained from the patient in the past. method.

7. In paragraph 3, A step of analyzing the biometric data based on the learned second neural network model to obtain a second score for the possibility of the patient's disease, and determining the severity of the patient's disease based on the first and second scores; method.

8. In paragraph 7, The step of determining the severity of the patient according to the above disease is: A step of determining the severity of the patient according to the disease by determining the correlation between the disease and the severity of the patient based on the first score, the second score, and the disease type information of the patient; including; method.

9. In paragraph 7, The first and second neural network models learned above are, Sharing a network that extracts feature information corresponding to the patient's condition from the above biometric data, method.

10. In a computing device that determines the severity of a patient's condition based on a deep learning model, A processor comprising at least one core; A memory containing program codes executable by the processor; and including a communication interface (communication unit); The above processor, Obtaining the patient's biometric data and analyzing the biometric data based on the first neural network model that has been trained to determine the severity of the patient's condition, The above-mentioned first neural network model is characterized in that it is learned based on learning data that is assigned labels corresponding to the patient's condition and treatment method classified according to the patient's severity. Computing device.

11. A computer program stored in a computer-readable storage medium, wherein the computer program, when executed on one or more processors, performs an operation of determining the severity of a patient based on a deep learning model. The above action is, The act of obtaining the patient's biometric data; and An operation of analyzing the biometric data to determine the severity of the patient based on the first neural network model that has been learned; The above-mentioned first neural network model is characterized in that it is learned based on learning data that is assigned labels corresponding to the patient's condition and treatment method classified according to the patient's severity. Computer program.

Images