WO2025071243A1 - Method, program, and device for analyzing single-lead electrocardiogram by using artificial intelligence - Google Patents

Abstract

Disclosed are: a prediction method of analyzing a single-lead electrocardiogram by using artificial intelligence, the method being performed by a computing device according to an embodiment of the present disclosure; a program; and a device. The method may comprise the steps of: obtaining single-lead electrocardiogram data; and on the basis of the length of the obtained single-lead electrocardiogram data, analyzing at least one of input conditions of a plurality of neural network models pre-trained to read single-lead electrocardiogram data, so as to process the obtained single-lead electrocardiogram data.

Description

Method, program and device for analyzing single-lead electrocardiogram using artificial intelligence

The present disclosure relates to artificial intelligence technology in the medical field, and more specifically, to a method for diagnosing various diseases and health conditions of a person whose single-lead electrocardiogram has been measured by analyzing a single-lead electrocardiogram using artificial intelligence.

An electrocardiogram is a test that measures electrical signals generated from the heart, and is widely used to diagnose heart disease. The electrocardiogram measured at medical institutions such as hospitals is measured by attaching four electrodes to the arms and legs and six electrodes to the chest, and recording the heart signals measured from different directions, creating a total of 12 leads.

It is difficult to measure 12-lead electrocardiograms for long periods of time because the electrodes are attached to the entire body. 12-lead electrocardiograms are usually recorded for about 10 seconds. Therefore, in the case of diseases with intermittent symptoms such as arrhythmia, it is not easy to detect them by reading a 12-lead electrocardiogram measured for a short period of time.

In order to diagnose diseases with intermittent symptoms such as arrhythmia, a method of measuring electrocardiograms over a long period of time using a patch-type measuring device is used rather than a method of measuring electrocardiograms by attaching electrodes to the arms, legs, and chest. Since this method of measuring usually measures by attaching electrodes only to the chest, only a single-lead electrocardiogram is recorded.

On the other hand, since single-lead electrocardiograms can be measured using portable measuring devices, they have the advantage of being easy to measure in real life. Therefore, attempts to diagnose various diseases and health conditions as well as diseases with intermittent symptoms such as arrhythmia using single-lead electrocardiograms are increasing. In particular, attempts to enable integrated diagnosis of various diseases and health conditions through healthcare platforms are increasing. However, since the form of the electrocardiogram required for diagnosis may vary depending on the disease or health condition to be analyzed, it may be difficult to accurately diagnose various diseases or health conditions by directly using single-lead electrocardiograms measured by devices.

The present disclosure aims to provide a method for integrating diagnosis of various diseases such as arrhythmia, heart failure, and health conditions such as sleep disorders based on a single-lead electrocardiogram using artificial intelligence.

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

According to one embodiment of the present disclosure for realizing the task as described above, a method for analyzing a single-lead electrocardiogram using artificial intelligence, which is performed by a computing device, is disclosed. The method may include the steps of: obtaining single-lead electrocardiogram data; and processing the obtained single-lead electrocardiogram data by analyzing at least one of input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data based on the length of the obtained single-lead electrocardiogram data.

Alternatively, the step of acquiring the single-lead electrocardiogram data may include a step of acquiring the single-lead electrocardiogram data measured by the electrocardiogram measuring device in real time, or a step of acquiring the single-lead electrocardiogram data having a predetermined length after measurement is completed by the electrocardiogram measuring device.

Alternatively, the step of processing the acquired single-lead electrocardiogram data may include, when the single-lead electrocardiogram data measured by the electrocardiogram measuring device is acquired in real time, a step of analyzing whether the length of the time-series accumulated single-lead electrocardiogram data satisfies at least one of the input conditions; and a step of extracting single-lead electrocardiogram data having a length that satisfies at least one of the input conditions from the time-series accumulated single-lead electrocardiogram data.

Alternatively, the step of processing the acquired single-lead electrocardiogram data may include, when single-lead electrocardiogram data having a predetermined length whose measurement has been completed by the electrocardiogram measuring device is acquired, a step of identifying a section satisfying at least one of the input conditions in the single-lead electrocardiogram data having the predetermined length; and a step of extracting single-lead electrocardiogram data of a section satisfying at least one of the input conditions from the single-lead electrocardiogram data having the predetermined length.

Alternatively, the method may further include a step of analyzing a noise level of the processed single-lead electrocardiogram data to determine whether to input the processed single-lead electrocardiogram data to at least one of the plurality of neural network models.

Alternatively, the step of determining whether to input the processed single-lead electrocardiogram data into at least one of the plurality of neural network models may include the step of analyzing whether a noise level of the processed single-lead electrocardiogram data satisfies a noise condition of a model among the plurality of models having an input condition corresponding to a length of the processed single-lead electrocardiogram data; and the step of determining whether to input the processed single-lead electrocardiogram data into a model having an input condition corresponding to a length of the processed single-lead electrocardiogram data, depending on whether the noise condition is satisfied.

Alternatively, the length of the electrocardiogram data included in the above input conditions may vary depending on the analysis tasks of the multiple neural network models.

Alternatively, an input condition of a model analyzing a disease related to electrocardiogram rhythm characteristics among the plurality of neural network models may be whether the length of single-derived electrocardiogram data corresponds to a first time period.

Alternatively, an input condition of a model analyzing a disease related to characteristics other than electrocardiogram rhythm among the plurality of neural network models may be whether the length of single-derived electrocardiogram data corresponds to the second period.

Alternatively, an input condition of a model analyzing sleep among the plurality of neural network models may be whether the length of single-induced electrocardiogram data corresponds to the third period.

Alternatively, an input condition of a model analyzing stress among the plurality of neural network models may be whether the minimum length of single-induced electrocardiogram data corresponds to a fourth period. In addition, the first period, the second period, the third period, and the fourth period may be different.

Alternatively, the step of processing the acquired single-lead electrocardiogram data may include a step of converting the acquired single-lead electrocardiogram data into heart rate variability data when the length of the acquired single-lead electrocardiogram data satisfies an input condition of a model analyzing stress among the plurality of neural network models.

Alternatively, the method may include a step of inputting the processed single-lead electrocardiogram data into a model among the plurality of models having an input condition corresponding to a length of the processed single-lead electrocardiogram data, thereby generating an interpretation result regarding a disease or health condition.

According to one embodiment of the present disclosure for achieving the above-described task, a computer program stored in a computer-readable storage medium is disclosed. When the computer program is executed on one or more processors, it performs operations for analyzing a single-lead electrocardiogram using artificial intelligence. At this time, the operations may include: an operation for obtaining single-lead electrocardiogram data; and an operation for processing the obtained single-lead electrocardiogram data by analyzing at least one of input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data based on the length of the obtained single-lead electrocardiogram data.

According to one embodiment of the present disclosure for achieving the above-described task, a computing device for analyzing a single-lead electrocardiogram using artificial intelligence is disclosed. The device may include a processor including at least one core; a memory including program codes executable by the processor; and a network unit for acquiring single-lead electrocardiogram data. In this case, the processor may analyze at least one of input conditions of a plurality of neural network models pre-learned to read single-lead electrocardiogram data based on the length of the acquired single-lead electrocardiogram data to process the acquired single-lead electrocardiogram data.

The present disclosure processes a single-lead electrocardiogram to various data formats required for the analysis of various diseases and health conditions, thereby constructing an integrated diagnostic system capable of diagnosing not only diseases suitable for the analysis of a single-lead electrocardiogram but also diseases suitable for the analysis of a conventional 12-lead electrocardiogram and even health conditions such as sleep disorders.

FIG. 1 is a block diagram of a computing device according to one embodiment of the present disclosure.

FIG. 2 is a block diagram illustrating a computational process of a computing device according to one embodiment of the present disclosure.

FIG. 3 is a flowchart illustrating a process of analyzing a single-lead electrocardiogram according to one embodiment of the present disclosure.

FIG. 4 is a flowchart illustrating a method for analyzing a single-lead electrocardiogram using artificial intelligence according to one embodiment of the present disclosure.

Hereinafter, embodiments of the present disclosure will be 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 the present disclosure are provided so that those skilled in the art can 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 below.

Throughout the specification of the present disclosure, the same or similar drawing reference numerals refer to the same or similar components. In addition, in order to clearly describe the present disclosure, drawing reference numerals of parts that are not related to the description of the present disclosure may be omitted in the drawings.

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 the context makes clear, "X employs either A or B" should be understood to mean either one of the natural inclusive permutations. For example, unless otherwise specified herein or the context makes clear, "X employs A or B" can be interpreted to mean either X employs A, X employs B, or X employs both A and B.

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

The terms "comprises" and/or "comprising" as used herein should be understood to mean the presence of particular 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.

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.”

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 according to a predetermined standard such as a functional viewpoint, a structural viewpoint, 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 should be distinguished for convenience of explanation may also be distinguished as a first component or a second component.

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

Meanwhile, the term "module" or "unit" used in the present disclosure may 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, a combination of software and hardware, etc. At this time, the "module" or "unit" may be a unit composed of a single element, or may be 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 a computing device itself that constitutes a system, or an application that is executed on a computing device, etc. However, the above-described concept is only an example, and the concept of “module” or “part” may be variously defined within a category understandable to those skilled in the art based on the contents of the present disclosure.

The term "model" used in the present disclosure 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 regarding 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 a problem-solving ability through learning. In this case, the neural network may have a problem-solving ability by optimizing parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network, or may include a neural network set in which multiple neural networks are combined.

The explanation of the terms set forth above is intended to aid in understanding the present disclosure. Therefore, if the terms set forth above are not explicitly stated as matters limiting the contents of the present disclosure, it should be noted that they are not used to limit the technical ideas of the contents of the present disclosure.

FIG. 1 is a block diagram of a computing device according to one embodiment of the present disclosure.

The computing device (100) according to one 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 an intensive data processing function 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 a plurality of servers and clients interact to comprehensively process data. Since the above description is only one example related to the type of the computing device (100), the type of the computing device (100) may be configured in various ways within a category that can be understood by those skilled in the art based on the contents of the present disclosure.

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

The processor (110) according to one 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 of 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 type of the processor (110) described above is only one example, and thus, the type of the 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.

The processor (110) can perform preprocessing on single-induction electrocardiogram data by considering the input conditions of neural network models for electrocardiogram reading. Since the input conditions of neural network models may differ depending on the type of reading, it is necessary to process the data acquired by the processor (110) according to each of the input conditions. Accordingly, the processor (110) can determine whether the single-induction electrocardiogram data satisfies at least one of the input conditions and process the data according to the satisfied condition. For example, when the single-induction electrocardiogram data is accumulated and acquired in real time, the processor (110) can analyze the length of the accumulated data to determine whether it satisfies at least one of the input conditions of the multiple neural network models. If there is a satisfied condition among the input conditions, the processor (110) can extract data according to the satisfied condition from the accumulated data. When the single-induction electrocardiogram data is acquired with a predetermined length, the processor (110) can extract the data according to the input conditions of the multiple neural network models. Since the data is already of a fixed length, the processor (110) can extract data according to the length required by each input condition within a fixed length. Through such preprocessing, the processor (110) can efficiently integrate and manage multiple neural network models that perform different readings.

The processor (110) can input preprocessed single-lead electrocardiogram data into at least one of the neural network models for electrocardiogram interpretation, and analyze the disease or health status of the subject who measured the single-lead electrocardiogram data. The processor (110) can input data processed according to specific input conditions into a neural network model requiring specific input conditions, and generate an interpretation result regarding the disease or health status. If data satisfying all or part of the input conditions of a plurality of neural network models performing different interpretations are generated through preprocessing, the processor (110) can input the preprocessed data into models to which the input conditions correspond, and generate an interpretation result. For example, if single-lead electrocardiogram data is processed according to A and B among input conditions A, B, and C through preprocessing, the processor (110) can input data A processed according to input condition A into model A requiring input condition A, and generate an interpretation result A. And, the processor (110) can input data B processed according to the input condition B into model B that requires the input condition B, and generate a reading result B. Even if single-induction electrocardiogram data is acquired, since it does not have a length that satisfies the input condition C and thus preprocessing is not performed, the processor (110) may not perform reading through model C. If preprocessing is performed by satisfying the input condition C in the process of accumulating and acquiring single-induction electrocardiogram data, the processor (110) can perform reading through model C.

The processor (110) can perform post-processing on the reading results generated from all or part of the plurality of neural network models. When multiple reading data are generated through the above-described reading, the processor (110) can generate an integrated diagnosis result by combining the individual result values included in the plurality of reading data using a statistical technique. The processor (110) can also input the individual result values included in the plurality of reading data into a neural network model for post-processing to generate an integrated diagnosis result regarding a disease or health condition. For example, when outputs of an arrhythmia analysis model, a heart failure analysis model, and a stress analysis model among the plurality of neural network models are generated, the processor (110) can input the probability values corresponding to the outputs of each model into a neural network model for post-processing to generate an integrated diagnosis result of a subject whose single-induced electrocardiogram data was measured. The neural network model for post-processing can comprehensively analyze evaluation indicators such as input values and output reliability of each model to generate an integrated reading result indicating what disease the subject whose electrocardiogram was measured is suffering from or what health condition it is in.

Meanwhile, neural network models for ECG interpretation can be pre-trained based on training data consisting of single-lead ECGs. At this time, neural network models for ECG interpretation can be configured to interpret different types of diseases or health conditions. In addition, neural network models can have the same or different input conditions depending on the change in analysis tasks. For example, a model predicting heart failure can receive training data consisting of a single-lead ECG of 10 seconds in length, extract features from the input training data, and calculate a probability value for the onset of heart failure. A model predicting arrhythmia can receive training data consisting of a single-lead ECG of 24 hours in length, extract features from the input training data, and calculate a probability value for the onset of arrhythmia. The probability value calculated from each model can be compared with the label included in the training data input to each model. In the comparison process, an error can be calculated through a loss function. The parameters constituting the neural network of each model can be updated based on the error calculated through the loss function. The models can be learned through this update process. In addition to the learning methods of the examples described above, the neural network models for electrocardiogram interpretation of the present disclosure can be learned through methods such as unsupervised learning and self-supervised learning depending on the structure of the model.

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 network unit (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 RAM (random access memory), a SRAM (static random access memory), a ROM (read-only memory), an EEPROM (electrically erasable programmable read-only memory), a PROM (programmable read-only memory), 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 an 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.

The memory (120) can manage, by structuring and organizing, data required for the processor (110) to perform operations, combinations of data, and program codes executable by the processor (110). For example, the memory (120) can store electrocardiogram data received through the network unit (130) described below. The memory (120) can store program codes that operate a machine learning model to receive electrocardiogram data as input and perform learning, program codes that operate a machine learning model to receive electrocardiogram data as input and perform inference according to the intended use of the computing device (100), and processed data generated as the program codes are executed.

The network unit (130) according to one embodiment of the present disclosure may be understood as a configuration unit that transmits and receives data via any type of known wired and wireless communication system. For example, the network unit (130) may perform data transmission and reception using a wired and 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), ultra wide-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 network unit (130) may be applied in various ways other than the above-described examples.

The network unit (130) can receive data required for the processor (110) to perform calculations through wired or wireless communication with any system or any client, etc. In addition, the network unit (130) can transmit data generated through the calculation of the processor (110) through wired or wireless communication with any system or any client, etc. For example, the network unit (130) can receive electrocardiogram data through communication with a cloud server that performs tasks such as databases in a hospital environment, standardization of medical data, clients such as smart watches, or medical computing devices. The network unit (130) can transmit output data of a machine learning model, and intermediate data, processed data, etc. derived from the calculation process of the processor (110) through communication with the aforementioned database, server, client, or computing device, etc.

FIG. 2 is a block diagram illustrating a computational process of a computing device according to one embodiment of the present disclosure.

According to one embodiment of the present disclosure, when single-lead electrocardiogram data is acquired, the computing device (100) may analyze the length of the acquired data and the input conditions of neural network models for electrocardiogram interpretation to perform processing on the acquired data. When single-lead electrocardiogram data measured by an electrocardiogram measuring device is acquired in real time, the computing device (100) may continuously store the acquired data and analyze the cumulative length of the acquired data in real time to perform processing. When single-lead electrocardiogram data having a predetermined length whose measurement has been completed by an electrocardiogram measuring device is acquired, the computing device (100) may analyze the predetermined length of the acquired data to perform processing. In this case, the processing may be understood as a task of extracting data that satisfies at least one of the input conditions of neural network models for electrocardiogram interpretation from the acquired data.

For example, as shown in Fig. 2, it is assumed that there are models for interpreting different diseases. At this time, interpretation model A (210) and interpretation model C (230) may be models that input a single-lead electrocardiogram of 10 seconds in length. In addition, interpretation model B (220) and interpretation model D (240) may be models that input a single-lead electrocardiogram of 24 hours in length. In this way, the length included in the input conditions of each model may vary depending on the analysis task determined during the model construction and learning process. Specifically, the length included in the input conditions may vary depending on whether the analysis task of the neural network model corresponds to a disease related to electrocardiogram rhythm characteristics, a disease related to characteristics other than electrocardiogram rhythm, or a health condition. The input conditions that vary depending on the analysis task may be organized as shown in the following [Table 1].

Task Type Detail type Input Conditions Diseases associated with electrocardiogram rhythm characteristics Arrhythmia (atrial fibrillation, flutter, bradycardia, etc.) Period 1: 24 to 72 hours Diseases associated with characteristics other than electrocardiogram rhythm Heart failure, myocardial infarction, etc. 2nd period: 10 seconds sleep quality of sleep Period 3: 7 to 8 hours stress Stress index - Period 4: From 5 minutes to 24 to 72 hours - Format: Heart rate variability (HRV) data

When single-lead electrocardiogram data measured by an electrocardiogram measuring device is acquired in real time, the computing device (100) can continuously check whether the length of the accumulated and stored data corresponds to 10 seconds or 24 hours. When the length of the accumulated and stored data becomes 10 seconds, the computing device (100) can extract a data section (10) corresponding to 10 seconds and generate it as input data of the reading model A (210) and the reading model C (230). In addition, when the length of the accumulated and stored data becomes 24 hours, the computing device (100) can extract a data section (20) corresponding to 24 hours and generate it as input data of the reading model B (220) and the reading model D (240). The data generated through such processing can be input into individual models that meet the input conditions. This process can be repeated until the process of acquiring single-lead electrocardiogram data is completed. When single-lead electrocardiogram data having a predetermined length for which measurement has been completed by an electrocardiogram measuring device is acquired, the computing device (100) can identify 10-second sections and 24-hour sections from the single-lead electrocardiogram data having a predetermined length. The computing device (100) can extract the 10-second section (10) identified from the predetermined length and generate it as input data for reading model A (210) and reading model C (230). In addition, the computing device (100) can extract the 24-hour section (20) identified from the predetermined length and generate it as input data for reading model B (220) and reading model D (240). The data generated through this processing can be input into individual models that meet input conditions.

When 10-second-long input data (10) processed from single-lead electrocardiogram data are input to reading model A (210) and reading model C (230), each of reading model A (210) and reading model C (230) can individually output reading results (31, 33) regarding a disease or health state by analyzing the input data according to a learned task. When 24-hour-long input data (20) processed from single-lead electrocardiogram data are input to reading model B (220) and reading model D (240), each of reading model B (220) and reading model D (240) can individually output reading results (32, 34) regarding a disease or health state by analyzing the input data according to a learned task.

Meanwhile, referring to [Table 1], among the reading models, a model analyzing stress may include receiving heart rate variability data instead of electrocardiogram data as an additional input condition. Accordingly, the computing device (100) may extract single-lead electrocardiogram data having a length that satisfies the input condition of the model analyzing stress from the acquired single-lead electrocardiogram data, and convert the extracted single-lead electrocardiogram data into heart rate variability data. Then, the computing device (100) may input the converted heart rate variability data into the model analyzing stress, and calculate a stress index as a reading result.

Through this processing process, the computing device (100) can manage models with different input conditions by integrating them into a single system. That is, the computing device (100) can build an environment in which diseases or health conditions can be comprehensively and stably diagnosed through data preprocessing that takes into account the input conditions of multiple models.

FIG. 3 is a flowchart illustrating a process of analyzing a single-lead electrocardiogram according to one embodiment of the present disclosure.

Referring to FIG. 3, a computing device (100) according to an embodiment of the present disclosure can obtain single-lead electrocardiogram data (S100). The computing device (100) can obtain single-lead electrocardiogram data through wired or wireless communication with an electrocardiogram measuring device. At this time, the computing device (100) can obtain single-lead electrocardiogram data measured by the electrocardiogram measuring device in real time. When obtaining data in real time, the computing device (100) can store the obtained data in a time series manner. The computing device (100) can also obtain single-lead electrocardiogram data having a predetermined length that has been measured by the electrocardiogram measuring device. That is, the computing device (100) can obtain electrocardiogram data through communication with the electrocardiogram measuring device in a state where electrocardiogram measurement is not completed, or can obtain electrocardiogram data having a specific length in a state where electrocardiogram measurement is completed.

The computing device (100) can analyze at least one of the input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data based on the length of the single-lead electrocardiogram data acquired through step S100 (S120). When there are a plurality of input conditions, the computing device (100) can analyze whether the data acquired through communication with the electrocardiogram measuring device satisfies each of the plurality of input conditions. When the single-lead electrocardiogram data measured by the electrocardiogram measuring device is acquired in real time, the computing device (100) can analyze whether the length of the single-lead electrocardiogram data stored in time series during the process of accumulating the single-lead electrocardiogram data acquired in real time satisfies at least one of the input conditions. At this time, the analysis of the input conditions (S120) can be repeatedly performed until the real-time acquisition is terminated. When single-lead electrocardiogram data having a predetermined length for which measurement has been completed is acquired from an electrocardiogram measuring device, the computing device (100) can identify a section that satisfies at least one of the input conditions in the single-lead electrocardiogram data having the predetermined length.

When the length of the single-lead electrocardiogram data stored in a time series manner satisfies at least one of the input conditions, or when an interval satisfying at least one of the input conditions is identified in the single-lead electrocardiogram data having a predetermined length, the computing device (100) can extract electrocardiogram data satisfying the input conditions from the acquired data (S130). The computing device (100) can extract the single-lead electrocardiogram data having a length satisfying at least one of the input conditions from the single-lead electrocardiogram data stored in a time series manner. The computing device (100) can extract the single-lead electrocardiogram data of an interval satisfying at least one of the input conditions from the single-lead electrocardiogram data having a predetermined length.

Meanwhile, if there is an input condition that is not satisfied by the length of the single-lead electrocardiogram data stored in time series, the computing device (100) may perform step S120 again to re-determine whether the input condition is satisfied. As time passes, the length of the single-lead electrocardiogram data stored in the computing device (100) increases, so the computing device (100) may repeatedly perform step S120 to continuously analyze the input conditions. In addition, if there is no section that satisfies any one of the input conditions in the single-lead electrocardiogram data having a predetermined length, the computing device (100) may perform step S110 again to accumulate and store the single-lead electrocardiogram data. In addition, the computing device (100) may perform step S120 again based on the accumulated and stored single-lead electrocardiogram data to re-identify the section that satisfies the input condition.

The computing device (100) may analyze the noise level of the processed single-lead electrocardiogram data through step S130 to determine whether to input the processed single-lead electrocardiogram data to at least one of the plurality of neural network models (S140). Each of the plurality of neural network models may have an individually analyzable noise level. In this way, the noise level individually set for each of the plurality of neural network models may be understood as a noise condition. The computing device (100) may analyze whether the noise level of the extracted single-lead electrocardiogram data according to the input condition satisfies the noise condition of a model having an input condition corresponding to the length of the processed single-lead electrocardiogram data among the plurality of models. Then, the computing device (100) may determine whether to input the processed single-lead electrocardiogram data to the model depending on whether the noise condition is satisfied. For example, if the noise level of the processed data according to the input condition A does not satisfy the noise condition A, the computing device (100) may determine not to input the corresponding data to model A. And, the computing device (100) can analyze the noise level of other data processed according to the input condition A, or re-perform the process of processing data according to the input condition. If the noise level of the data processed according to the input condition A satisfies the noise condition A, the computing device (100) can determine the corresponding data as input data of model A.

The computing device (100) can perform electrocardiogram analysis by inputting the processed data determined as input data through step S130 into the model (S150). The computing device (100) can generate a reading result regarding a disease or health condition through electrocardiogram analysis. The reading results are individually generated by each of the plurality of models, and the final analysis result can be generated at a time when the electrocardiogram measurement is completed or at a time determined by a user command. At this time, the computing device (10) can process the reading results generated by each of the plurality of models through a statistical technique or a post-processing neural network model to generate the final analysis result. Therefore, the final analysis result can include a comprehensive diagnosis result for the plurality of result values.

FIG. 4 is a flowchart illustrating a method for analyzing a single-lead electrocardiogram using artificial intelligence according to one embodiment of the present disclosure.

Referring to FIG. 4, a computing device (100) according to an embodiment of the present disclosure can obtain single-lead electrocardiogram data (S210). The method and cycle of obtaining single-lead electrocardiogram data may vary depending on a user command. For example, if a user wants real-time diagnosis or monitoring, the computing device (100) can obtain data measured in real time through communication with an electrocardiogram measuring device by reflecting the user command. If a user wants to check the analysis result at a specific cycle or time, the computing device (100) can obtain electrocardiogram data for which measurement has been completed by communicating with an electrocardiogram measuring device at a specific cycle or time.

The computing device (100) can process the single-induced electrocardiogram data by analyzing at least one of the input conditions of a plurality of pre-trained neural network models to read the single-induced electrocardiogram data based on the length of the single-induced electrocardiogram data (S220). The description of step S220 corresponds to the description of steps S120, S130, and S140 of FIGS. 2 and 3, and therefore is omitted below.

The computing device (100) can input the single-induced electrocardiogram data processed through step S220 into a model among multiple models that has an input condition corresponding to the length of the single-induced electrocardiogram data processed through step S220, thereby generating a reading result regarding a disease or health condition (S230).

The various embodiments of the present disclosure described above can be combined with additional embodiments and can be modified within a range that can be understood by those skilled in the art in light of the detailed description set forth 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 concept of the claims of the present disclosure should be interpreted as being included in the scope of the present disclosure.

Claims (12)

A method for analyzing a single-lead electrocardiogram using artificial intelligence, performed by a computing device including at least one processor, Step of acquiring single-lead electrocardiogram data; and A step of processing the acquired single-lead electrocardiogram data by analyzing at least one of input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data based on the length of the acquired single-lead electrocardiogram data; Including, method. In paragraph 1, The step of obtaining the above single-lead electrocardiogram data is: A step of acquiring single-lead electrocardiogram data measured in real time by an electrocardiogram measuring device, or acquiring single-lead electrocardiogram data having a predetermined length after measurement is completed by the electrocardiogram measuring device; Including, method. In the second paragraph, The step of processing the above acquired single-lead electrocardiogram data is: When single-lead electrocardiogram data measured by the above electrocardiogram measuring device is acquired in real time, a step of analyzing whether the length of the single-lead electrocardiogram data accumulated in time series satisfies at least one of the above input conditions; and A step of extracting single-lead electrocardiogram data having a length satisfying at least one of the above input conditions from the time-series accumulated single-lead electrocardiogram data; Including, method. In the second paragraph, The step of processing the above acquired single-lead electrocardiogram data is: When single-lead electrocardiogram data having a predetermined length is acquired by the electrocardiogram measuring device, a step of identifying a section that satisfies at least one of the input conditions in the single-lead electrocardiogram data having the predetermined length; and A step of extracting single-lead electrocardiogram data of a section satisfying at least one of the above input conditions from single-lead electrocardiogram data having the predetermined length; Including, method. In paragraph 1, A step of analyzing a noise level of the processed single-lead electrocardiogram data to determine whether to input the processed single-lead electrocardiogram data into at least one of the plurality of neural network models; Including more, method. In paragraph 5, The step of determining whether to input the processed single-induced electrocardiogram data into at least one of the plurality of neural network models is: A step of analyzing whether the noise level of the processed single-lead electrocardiogram data satisfies the noise condition of a model among the plurality of models having an input condition corresponding to the length of the processed single-lead electrocardiogram data; and A step of determining whether to input the processed single-lead electrocardiogram data into a model having an input condition corresponding to the length of the processed single-lead electrocardiogram data, depending on whether the above noise condition is satisfied; Including, method. In paragraph 1, Regarding the length of the electrocardiogram data included in the above input conditions, Depending on the analysis task of the above multiple neural network models, method. In paragraph 1, Among the above multiple neural network models, the input condition of the model analyzing diseases related to electrocardiogram rhythm characteristics is whether the length of single-induced electrocardiogram data corresponds to the first time period. Among the above multiple neural network models, the input condition of the model analyzing diseases related to characteristics other than electrocardiogram rhythm is whether the length of single-induced electrocardiogram data corresponds to the second period. Among the above multiple neural network models, the input condition of the model analyzing sleep is whether the length of the single-induced electrocardiogram data corresponds to the third period. Among the above multiple neural network models, the input condition of the model analyzing stress is whether the minimum length of single-induced electrocardiogram data corresponds to the fourth period. The first period, the second period, the third period and the fourth period are different, method. In Article 8, The step of processing the above acquired single-lead electrocardiogram data is: A step of extracting single-lead electrocardiogram data having a length that satisfies the input condition of a model analyzing stress among the plurality of neural network models from the acquired single-lead electrocardiogram data and converting the extracted single-lead electrocardiogram data into heart rate variability data; Including, method. In paragraph 1, A step of inputting the processed single-lead electrocardiogram data into a model among the plurality of models having an input condition corresponding to the length of the processed single-lead electrocardiogram data, thereby generating a reading result regarding a disease or health condition; Including, method. A computer program stored in a computer-readable storage medium, wherein the computer program, when executed on one or more processors, performs operations for analyzing a single-lead electrocardiogram using artificial intelligence. The above actions are, The act of acquiring single-lead electrocardiogram data; and An operation of processing the acquired single-lead electrocardiogram data by analyzing at least one of input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data based on the length of the acquired single-lead electrocardiogram data; Including Computer program. A computing device for analyzing single-lead electrocardiograms using artificial intelligence, A processor comprising at least one core; a memory containing program codes executable by the processor; and A network unit for acquiring single-lead electrocardiogram data; Including, The above processor, Based on the length of the single-lead electrocardiogram data obtained above, at least one of the input conditions of a plurality of neural network models pre-trained to read the single-lead electrocardiogram data is analyzed to process the obtained single-lead electrocardiogram data. device.

Images