Method, program and device for predicting health status using electrocardiogram segments

Abstract

According to an embodiment of the present disclosure, a method, program, and apparatus for predicting a health condition using electrocardiogram segments, performed by a computing device, are disclosed, which include the steps of: dividing electrocardiogram data to generate a plurality of segments; inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the probability of disease corresponding to each of the plurality of segments; and combining the probability of disease from which the outliers have been removed to generate a result value for predicting the health condition.

Description

[Technical Field]

[0001] The present disclosure relates to data processing technology in the medical field, and more particularly to a method for filtering electrocardiogram data using artificial intelligence for stable and accurate analysis of electrocardiograms and predicting health conditions based on the filtered data. [Background technology]

[0002] An electrocardiogram (ECG) is a test that records the electrical activity of the heart. An ECG is a relatively simple and cost-effective test that can confirm the health of the heart, which plays an important role in the early diagnosis and management of heart disease. For example, ECG signals measured through an ECG can be used to determine whether each part of the heart is functioning normally, whether the size and position of the heart are normal, and whether there is any damage to the myocardium. Based on these determinations, ECG signals can be used to diagnose various heart-related problems and predict a person's health status.

[0003] Meanwhile, with the development of artificial intelligence technology, attempts to analyze electrocardiograms using artificial intelligence technology are increasing. One representative example is a service that inputs an electrocardiogram into a neural network-based model to predict various heart diseases. Most of these conventional technologies and services process all or part of an electrocardiogram at once through an artificial intelligence model. However, these conventional methods are likely to fail to output accurate results if there is a problem with the electrocardiogram signal itself. For example, if the electrocardiogram signal itself contains a lot of noise or if missing intervals occur during the electrocardiogram signal measurement process, the conventional technologies and services cannot provide accurate analysis values ​​through the artificial intelligence model and are forced to measure the electrocardiogram signal again. Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure aims to provide a method for dividing an electrocardiogram signal into multiple parts and filtering the signals appropriate for analysis, thereby improving the accuracy and convenience of analysis.

[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 based on the following description. [Means for solving the problem]

[0006] To achieve the above object, an embodiment of the present disclosure provides a method for predicting a health condition using electrocardiogram segments, the method being performed by a computing device and including the steps of: dividing electrocardiogram data to generate a plurality of segments; inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the probability of disease corresponding to each of the plurality of segments; and combining the probability of disease from which the outliers have been removed to generate a result value for predicting a health condition.

[0007] Alternatively, dividing the electrocardiogram data to generate a plurality of segments may include dividing the electrocardiogram data into signals of a predetermined length without overlapping regions to generate the plurality of segments.

[0008] Alternatively, the step of dividing the electrocardiogram data to generate a plurality of segments may include the step of dividing the electrocardiogram data while moving a window of a fixed size over a predetermined area of ​​the electrocardiogram data to generate the plurality of segments.

[0009] Alternatively, the method may further include inputting the plurality of segments into a pre-trained machine learning model to calculate output uncertainty corresponding to each of the plurality of segments.

[0010] Alternatively, the machine learning model may include a neural network with probability distributions over neural network weights to quantify the uncertainty in the output.

[0011] Alternatively, the step of removing outliers from the disease possibilities corresponding to each of the plurality of segments based on the output uncertainty may include, when the output uncertainty is equal to or greater than a predetermined first reference value, treating the disease possibilities corresponding to the output uncertainty that is equal to or greater than the first reference value as outliers and removing them.

[0012] Alternatively, the step of combining the disease probabilities from which the outliers have been removed to generate a result value for predicting the health state may include the step of comparing an average value for the disease probabilities from which the outliers have been removed with a second preset reference value to generate a result value for predicting the health state.

[0013] Alternatively, the step of combining the disease probabilities from which the outliers have been removed to generate a result value for predicting the health state may include the step of comparing each of the disease probabilities from which the outliers have been removed with a third preset reference value to generate a result value for predicting the health state.

[0014] Alternatively, the step of comparing each of the possibilities of disease after the abnormal values ​​have been removed with a third predetermined reference value to generate a result value for the prediction of the health state may include the steps of converting the possibilities of disease after the abnormal values ​​have been removed into a binary value depending on whether the possibilities of disease after the abnormal values ​​have been removed are equal to or greater than the third predetermined reference value; and generating a result value for the prediction of the health state based on the value that accounts for the highest proportion of the converted binary values.

[0015] To achieve the above object, one embodiment of the present disclosure provides a computer program stored in a computer-readable storage medium. When executed by one or more processors, the computer program performs operations for predicting a health state using electrocardiogram segments. The operations include inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the probability of disease corresponding to each of the plurality of segments; and generating a result value for predicting the health state by combining the probability of disease from which the outliers have been removed.

[0016] According to one embodiment of the present disclosure, there is disclosed a computing device for predicting a health condition using electrocardiogram segments. The device may include a processor including at least one core, a memory including program code executable by the processor, and a network unit for acquiring electrocardiogram data. In this case, the processor may divide the electrocardiogram data to generate a plurality of segments, input the plurality of segments into a pre-trained machine learning model to calculate a disease probability corresponding to each of the plurality of segments, remove outliers from the disease probability corresponding to each of the plurality of segments, and combine the disease probabilities from which the outliers have been removed to generate a result value for predicting the health condition. [Effects of the Invention]

[0017] According to the method of the present disclosure, the accuracy of the analysis can be increased by dividing one electrocardiogram signal into multiple parts and filtering the signals appropriate for analysis.

[0018] In addition, by dividing one electrocardiogram into multiple parts and filtering the signals appropriate for analysis, the hassle of having to remeasure the signals required for analysis can be minimized, thereby improving the convenience of analysis. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 1 is a block diagram of a computing device according to one embodiment of the present disclosure. [Figure 2] FIG. 1 is a block diagram illustrating a process for predicting health status according to one embodiment of the present disclosure. [Figure 3] FIG. 1 is a conceptual diagram illustrating a neural network structure included in a machine learning model according to an embodiment of the present disclosure. [Figure 4] 1 is a flowchart illustrating a method for predicting health status using electrocardiogram segments according to one embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, with reference to the accompanying drawings, embodiments of the present disclosure will be described in detail 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 in the drawings that are not related to the explanation of the present disclosure may be omitted.

[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 in this disclosure or the context makes the meaning unclear, it should be understood to mean one of the natural inclusive permutations of "X utilizes A or B." For example, unless otherwise specified in this disclosure or the context makes the meaning unclear, "X utilizes A or B" can be interpreted as either X utilizes A, X utilizes B, or X utilizes both A and B.

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

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

[0025] Unless otherwise specified in this disclosure or clear from the context as indicating 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 the present disclosure may 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, components performing different functional roles in the present disclosure may be classified as a first component or a second component. However, components that are substantially the same within the technical concept of the present disclosure but must be distinguished for convenience of description may 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 generating data in an on-device manner as well as receiving data over a wired or wireless communication network with an external device or system.

[0028] Meanwhile, the terms "module" or "unit" used in this disclosure may 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. In this regard, a "module" or "unit" may refer to a unit composed of a single element or a unit expressed as a combination or set of multiple elements. For example, in a narrow sense, a "module" or "unit" may refer to a hardware element or a set of elements of a computing device, an application program that performs a specific software function, a processing procedure implemented through the execution of software, or a set of instructions for program execution. In a broad sense, 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 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 set 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. In this case, a neural network can acquire 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 explanations of the above terms are intended to aid in understanding the present disclosure. Therefore, unless the above terms are explicitly stated as matters that limit the contents of the present disclosure, please note that they are not used to limit the technical ideas of the contents of the present disclosure.

[0031] FIG. 1 is a block diagram of a computing device according to an embodiment of the present disclosure.

[0032] The computing 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 computing device 100 may be a server that performs intensive data processing functions and shares resources, or a client that shares resources through interaction with the server. The computing device 100 may also be a cloud system in which multiple servers and clients interact to process data comprehensively. The above description is merely an example of a type of computing device 100, and various types of computing devices 100 may be configured within a scope that would be understandable to one skilled in the art based on the contents of this disclosure.

[0033] 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, since FIG. 1 is merely an example, the computing device 100 may include other components for implementing a computing environment. Also, only some of the disclosed components may be included in the computing device 100.

[0034] The processor 110 according to an embodiment of the present disclosure may be understood as a component 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 operations such as input data processing 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), a field programmable gate array (FPGA), etc. The types of processor 110 described above are merely examples, and various types of processor 110 may be configured within the scope of what would be understood by one skilled in the art based on the present disclosure.

[0035] The processor 110 can divide electrocardiogram data including an electrocardiogram signal to generate a plurality of segments. The processor 110 can divide one piece of electrocardiogram data into a preset number of pieces of any size or a predetermined size to generate various segments including portions of the electrocardiogram signal included in the electrocardiogram data. For example, when electrocardiogram data including an electrocardiogram signal measured for t minutes is acquired, the processor 1100 can divide the electrocardiogram data including the electrocardiogram signal measured for t minutes so as to generate segments including an electrocardiogram signal having a length of i minutes (i is a value smaller than t).

[0036] The processor 110 can input multiple segments to a pre-trained neural network model and calculate a probability of disease corresponding to each of the multiple segments. The disease probability can include a disease occurrence probability value derived by analyzing a portion of the electrocardiogram signal included in the segment. The processor 110 can input multiple segments to the pre-trained neural network model and calculate an output uncertainty corresponding to each of the multiple segments. The output uncertainty can be understood as an indicator of the reliability of the disease probability calculated by the neural network model. That is, the processor 110 can input the segments to the pre-trained neural network model and derive a probability value that the subject of the electrocardiogram signal has a disease based on each segment. The processor 110 can then confirm the reliability of the derived probability value through the output of the neural network model.

[0037] Meanwhile, neural network models that calculate the probability of disease and the uncertainty of the output can perform learning by comparing each output and label through a loss function and adjusting the neural network parameters. In this case, functions such as cross entropy can be used as the loss function. The neural network model can perform learning by adjusting the neural network parameters in a direction that minimizes the error between the output and the label. The neural network model can perform learning in the form of supervised learning, but can also perform learning in the form of unsupervised learning or self-supervised learning depending on the structure of the neural network.

[0038] The processor 110 can filter the disease possibilities corresponding to each of the multiple segments based on the uncertainty of the output calculated through the neural network model. The processor 110 can compare the uncertainty of the output calculated through the neural network model with a preset reference value to select values ​​to be analyzed from the disease possibilities corresponding to each of the multiple segments. Even in a single electrocardiogram signal, noise or missing values ​​can occur in some sections during the measurement process, and sections where noise or missing values ​​occur can cause problems that actually reduce the accuracy of the analysis. Therefore, to improve the accuracy of the analysis, the processor 110 can analyze the reliability of the values ​​calculated for each of the multiple segments and select values ​​with a higher reliability compared to the reference value to be analyzed.

[0039] The processor 110 can generate a result value for predicting a health state by combining the disease probabilities filtered based on the output uncertainty. The processor 110 can combine the disease probabilities corresponding to certain sections of the ECG signal to derive the disease probabilities detected in the entire ECG signal. The processor 110 can then generate an analysis result of the health state of the subject whose ECG signal was measured based on the disease probabilities detected in the entire ECG signal. The analysis result of the health state can include a diagnosis or prediction result for the disease, such as which disease the subject currently has or the probability of developing the disease in the future. In this way, the processor 110 can obtain a more accurate result value by dividing the ECG signal into multiple parts, performing an in-depth analysis, and then selecting the analysis values, compared to when the entire ECG signal is not analyzed at once.

[0040] The memory 120 according to an embodiment of the present disclosure may be understood as a component including hardware and / or software for storing and managing data processed by 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 network unit 130. For example, the memory 120 may include at least one type of storage medium selected from the group consisting of flash memory, hard disk, multimedia card micro, card-type memory, random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, and optical disk. The memory 120 may also include a database system for managing data in a predetermined manner. The types of memory 120 described above are merely examples, and the types of memory 120 may be configured in various ways within the scope of what would be understood by a person skilled in the art based on the contents of this disclosure.

[0041] The memory 120 may structure, organize, and manage data, combinations of data, and program code executable by the processor 110 required for the processor 110 to perform calculations. For example, the memory 120 may store medical data received through the network unit 130 (described below). The memory 120 may store program code for operating a neural network model to receive input medical data and perform learning, program code for operating a neural network model to receive input medical data and perform inference according to the intended use of the computing device 100, and processed data generated by executing the program code.

[0042] The network unit 130 according to an embodiment of the present disclosure may be understood as a component that transmits and receives data through any type of publicly-known wired / wireless communication system. For example, the network unit 130 may transmit and receive data using a wired / 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 wideband, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity (WiFi), near field communication (NFC), or Bluetooth. The above-mentioned communication systems are merely examples, and various wired / wireless communication systems for transmitting and receiving data by the network unit 130 may be applied other than the above-mentioned examples.

[0043] The network unit 130 may receive data necessary for the processor 110 to perform calculations via wired / wireless communication with any system or any client. The network unit 130 may also transmit data generated through calculations by the processor 110 via wired / wireless communication with any system or any client. For example, the network unit 130 may receive medical data through communication with a database in a hospital environment, a cloud server that performs tasks such as medical data standardization, a client such as a smart watch, or a medical computing device. The network unit 130 may transmit output data of the neural network model, and intermediate data and processed data derived during the calculation process of the processor 110 via communication with the aforementioned database, server, client, or computing device.

[0044] FIG. 2 is a block diagram illustrating a process for predicting a health condition according to one embodiment of the present disclosure.

[0045] 2, a computing device 100 according to an embodiment of the present disclosure may generate electrocardiogram segments 20 including a portion of an electrocardiogram signal based on electrocardiogram data 10 including an electrocardiogram signal. The computing device 100 may generate a plurality of electrocardiogram segments 20 by dividing the electrocardiogram data 10 into predetermined lengths without overlapping regions. The computing device 100 may generate a plurality of electrocardiogram segments 20 by dividing the electrocardiogram data 10 while moving a window of a fixed size over a predetermined region in the electrocardiogram data 10.

[0046] For example, assume that computing device 100 acquires electrocardiogram data including an electrocardiogram signal consisting of [A1, A2, A3, A4, A5, A6, A7, A8, A9]. To generate three segments, computing device 100 can divide the electrocardiogram data into [A1, A2, A3], [A4, A5, A6], and [A7, A8, A9] so that there are no overlapping regions in the entire signal. Alternatively, to generate multiple segments by moving a window of size three over the electrocardiogram signal one region at a time, computing device 100 can divide the electrocardiogram signal into segments such as [A1, A2, A3], [A2, A3, A4], [A3, A4, A5], ..., [A7, A8, A9] so that some of the electrocardiogram signal overlaps. This division method can be preset and selectively performed depending on the intended use.

[0047] The computing device 100 can input a plurality of electrocardiogram segments 20 into the pre-trained machine learning model 200 to derive a disease likelihood 30 and an output uncertainty 40 for each of the plurality of segments 20. Specifically, the computing device 100 can input each of the plurality of segments 20 into the machine learning model 200 to calculate a disease likelihood 30 including a disease invention probability value corresponding to each of the plurality of segments. The computing device 100 can then obtain an output uncertainty 40 corresponding to each of the disease likelihoods 30 from the output of the machine learning model 200.

[0048] For example, assume that multiple segments are generated as S1, S2, S3, ..., Sn. Each of the multiple segments has information for a certain size (or time). The computing device 100 can input S1 into a machine learning model and calculate P1, which corresponds to the probability of the disease being analyzed from S1. The computing device 100 can also obtain U1, which is uncertainty indicating the degree to which the value P1 can be trusted, from the output of the machine learning model. The computing device 100 can input S2 into a machine learning model and calculate P2, which corresponds to the probability of the disease being analyzed from S2. The computing device 100 can also obtain U1, which is uncertainty indicating the degree to which the value P1 can be trusted, from the output of the machine learning model. The computing device 100 can sequentially repeat this process to calculate Pn, which corresponds to the probability of the disease being analyzed from Sn, and Un, which indicates the degree of confidence in Pn.

[0049] The computing device 100 can remove outliers present in the disease probability 30 corresponding to each of the multiple segments based on the output uncertainty 40. The computing device 100 can compare the uncertainty 40 paired with the disease probability 30 with a preset reference value and extract outliers within the disease probability 30 that are determined to be unreliable as analytical values. Through this outlier removal (or extraction) process, the computing device 100 can select good quality data as the basis for deriving the health condition prediction result 50.

[0050] For example, the computing device 100 may compare the uncertainty U1, which indicates the degree of confidence in the disease probability P1, with a preset first reference value to determine whether U1 is equal to or greater than the first reference value. If U1 is equal to or greater than the first reference value, the computing device 100 may determine that P1 corresponding to U1 is an abnormal value and remove it. If U1 is equal to or less than the first reference value, the computing device 100 may determine that P1 corresponding to U1 is a normal value and reflect it in the final analysis result. The computing device 100 may sequentially repeat the above process to determine whether Pn is an abnormal value and filter it.

[0051] The computing device 100 can combine the disease probabilities from which the outliers have been removed to generate a result for predicting the health condition 50. The computing device 100 can compare the average value for the disease probabilities from which the outliers have been removed with a second preset reference value to generate a result for predicting the health condition 50. The computing device 100 may also compare each of the disease probabilities from which the outliers have been removed with a third preset reference value to generate a result for predicting the health condition 50.

[0052] For example, assume that the remaining disease probability values ​​after the abnormal values ​​have been removed through the above process are three values, i.e., 0.5, 0.7, and 0.4. The computing device 100 can calculate the average of the three values, (0.5 + 0.7 + 0.4) / 3 = 0.53. In this case, if the second reference value is set to 0.5, the average of the three values, 0.53, is greater than the second reference value, 0.51, and therefore the computing device 100 can finally determine that a specific disease has occurred.

[0053] Furthermore, if the third reference value is set to 0.51, the computing device 100 may compare each of the three values ​​with the third reference value 0.51. Because the first value 0.5 is smaller than the third reference value 0.51, the computing device 100 may tag the corresponding value as 0, meaning that the specific disease does not occur. Because the second value 0.7 is larger than the third reference value 0.51, the computing device 100 may tag the corresponding value as 1, meaning that the specific disease has occurred. Because the third value 0.4 is smaller than the third reference value 0.51, the computing device 100 may tag the corresponding value as 0. Then, the computing device 100 may make a final determination that the specific disease does not occur based on the value 0, which is the value that accounts for the largest proportion of the values ​​tagged as 0, 1, and 0. This final determination result may be reflected in the health condition prediction result.

[0054] In the above example, it can be seen that the results of the method of comparing the second reference value with the average value and the method of comparing the third reference value with the individual value may differ. Therefore, which method to use may be determined according to the user's selection depending on the intended use. That is, the computing device 100 may determine which method to use to make the final decision based on user input.

[0055] Meanwhile, the first, second, or third reference value may be preset by a user depending on the intended use. For example, if the disease analyzed through the machine learning model is a rare disease, the possibility of misdiagnosis may be higher than for other diseases if the uncertainty of the analysis results is high. Therefore, in such a case, the first, second, or third reference value may be set higher than for other diseases. In this way, the reference values ​​may be set in various ways depending on the type of disease, data attributes, etc.

[0056] FIG. 3 is a conceptual diagram illustrating a neural network structure included in a machine learning model according to one embodiment of the present disclosure.

[0057] Referring to FIG. 3, a machine learning model according to an embodiment of the present disclosure may include a neural network 300 having a probability distribution for neural network weights to quantify output uncertainty. The neural network 300 of FIG. 3 assumes that the neural network weights do not exist as fixed values ​​but as a probability distribution. That is, the trained neural network weights of the neural network 300 of FIG. 3 may follow a Gaussian distribution. Furthermore, during the training process of the neural network 300 of FIG. 3, the mean and variance of the Gaussian distribution may be adjusted rather than the values ​​of the neural network weights themselves. Through this neural network 300, the machine learning model may derive uncertainty, which serves as an indicator of how reliable the output value, i.e., the likelihood of disease, is.

[0058] FIG. 4 is a flowchart illustrating a method for predicting health status using electrocardiogram segments according to one embodiment of the present disclosure.

[0059] Referring to FIG. 4, a computing device 100 according to an embodiment of the present disclosure may divide electrocardiogram data to generate a plurality of segments (S100). If the computing device 100 is a server providing an electrocardiogram interpretation service, the computing device 100 may acquire electrocardiogram data through communication with equipment for measuring electrocardiogram signals. If the computing device 100 is equipment for measuring and analyzing electrocardiogram signals, the computing device 100 may measure the electrocardiogram signals through a measurement unit to generate electrocardiogram data. Once the electrocardiogram data is acquired, the computing device 100 may divide the electrocardiogram data at a predetermined length without overlapping regions to generate a plurality of segments. The computing device 100 may divide the electrocardiogram data by moving a window of a fixed size over a predetermined region in the electrocardiogram data to generate a plurality of segments.

[0060] The computing device 100 can input multiple segments into a pre-trained machine learning model to calculate the probability of disease and the output uncertainty corresponding to each of the multiple segments (S200).

[0061] The computing device 100 can remove outliers from the disease possibilities corresponding to each of the plurality of segments based on the output uncertainty (S300). The computing device 100 can compare the output uncertainty with a predetermined first reference value. If the output uncertainty is equal to or greater than the predetermined first reference value, the computing device 100 can consider the disease possibilities corresponding to the output uncertainty equal to or greater than the first reference value as outliers and remove them. If the output uncertainty is less than the predetermined first reference value, the computing device 100 can use the disease possibilities corresponding to the output uncertainty less than the first reference value in the final analysis performed in step S400.

[0062] The computing device 100 may combine the disease probabilities from which outliers have been removed to generate a result value for predicting a health state (S400). The computing device 100 may compare an average value for the disease probabilities from which outliers have been removed with a preset second reference value to generate a result value for predicting a health state. The computing device 100 may compare each of the disease probabilities from which outliers have been removed with a preset third reference value to generate a result value for predicting a health state. Specifically, the computing device 100 may convert the disease probabilities from which outliers have been removed into a binary value depending on whether the disease probabilities from which outliers have been removed are equal to or greater than the preset third reference value. The computing device 100 may then generate a result value for predicting a health state based on the value that occupies the highest proportion of the converted binary values.

[0063] The various embodiments of the present disclosure described above may be combined with additional embodiments and may be modified within the scope that would be understandable to one skilled in the art in light of the above detailed description. The embodiments of the present disclosure are illustrative in all respects and should not be considered limiting. For example, components described as a single type may be implemented in a distributed form, and similarly, components described as distributed may 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 for predicting a health condition using electrocardiogram segments, performed by a computing device including at least one processor, comprising: segmenting the electrocardiogram data to generate a plurality of segments; inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the disease probabilities corresponding to each of the plurality of segments; and combining the disease probabilities with the outliers removed to generate an outcome value for predicting health status.

2. The step of dividing the electrocardiogram data to generate a plurality of segments comprises:

2. The method of claim 1, further comprising: dividing the electrocardiogram data signals into predetermined lengths without overlapping regions to generate the plurality of segments.

3. The step of dividing the electrocardiogram data to generate a plurality of segments comprises:

2. The method of claim 1, further comprising dividing the electrocardiogram data to generate the plurality of segments while moving a fixed-size window through a predetermined region of the electrocardiogram data.

4. The method of claim 1 , further comprising: inputting the plurality of segments into the pre-trained machine learning model to calculate output uncertainty corresponding to each of the plurality of segments.

5. The machine learning model 5. The method of claim 4, comprising: a neural network having probability distributions over neural network weights to quantify the uncertainty of the output.

6. The step of removing outliers from the disease possibilities corresponding to each of the plurality of segments includes:

5. The method of claim 4, further comprising: if the uncertainty of the output is equal to or greater than a preset first reference value, regarding a possible disease corresponding to the uncertainty of the output equal to or greater than the first reference value as an outlier and removing it.

7. The step of combining the disease probabilities from which the outliers have been removed to generate a result value for predicting the health state includes:

2. The method of claim 1, further comprising: comparing the mean value for the probability of disease from which the outliers have been removed with a second preset reference value to generate a result value for the prediction of the health state.

8. The step of combining the disease probabilities from which the outliers have been removed to generate a result value for predicting the health state includes:

2. The method of claim 1, further comprising: comparing each of the disease probabilities from which the outliers have been removed to a third preset reference value to generate a result value for the prediction of the health state.

9. The step of comparing each of the disease possibilities from which the abnormal values ​​have been removed with a preset third reference value to generate a result value for predicting the health state includes: converting the possibility of a disease after the outlier has been removed into a binary value depending on whether the possibility of a disease after the outlier has been removed is equal to or greater than a third predetermined reference value; and 9. The method of claim 8, further comprising: generating a result value for predicting the health state based on the value that occupies the highest proportion of the converted binary values.

10. A computer program stored in a computer-readable storage medium, the computer program, when executed by one or more processors, performs an operation for predicting a health condition using electrocardiogram segments; The operation is segmenting the electrocardiogram data to generate a plurality of segments; inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the disease probabilities corresponding to each of the plurality of segments; and an operation of combining the disease probabilities from which the outliers have been removed to generate a result value for predicting a health condition.

11. 1. A computing device for predicting a health condition using electrocardiogram segments, comprising: a processor including at least one core; a memory containing program code executable by the processor; and a network unit for acquiring electrocardiogram data; Including, The processor: Segmenting the electrocardiogram data to generate multiple segments; inputting the plurality of segments into a pre-trained machine learning model to calculate a probability of disease corresponding to each of the plurality of segments; removing outliers from the disease probabilities corresponding to each of the plurality of segments; The apparatus combines the disease probabilities with the outliers removed to generate a result value for predicting a health condition.

Images