Method, program and apparatus for classifying data in electrocardiogram reading center

Abstract

A method by which a computing device classifies data in an electrocardiogram reading center, a program and an apparatus are disclosed according to one embodiment of the present disclosure. The method may comprise the steps of: inputting electrocardiogram data into a pre-trained first model so as to perform a first classification that distinguishes the electrocardiogram data as normal or abnormal; and determining, on the basis of the first classification results, whether to perform a first reading process of analyzing, in multiple stages, the likelihood of disease or the health condition of a subject about whom the electrocardiogram data was measured. The normal or abnormal classification can be related to the likelihood of disease or the health condition of the subject about whom the electrocardiogram data was measured.

Description

Data classification method, program, and device of electrocardiogram interpretation center

[0001] The present disclosure relates to data processing technology in the medical field, and more specifically, to a method for efficiently improving the existing interpretation process of an electrocardiogram interpretation center.

[0002]

[0003] With the recent advancement of artificial intelligence technology, electrocardiogram interpretation centers are operating a process that interprets electrocardiograms through multiple steps, utilizing a combination of artificial intelligence and human electrocardiogram interpretation experts. For example, referring to Fig. 1, when a reading request is received from an electrocardiogram interpretation requester (200), in a conventional electrocardiogram interpretation center (300), a primary interpretation expert (310), such as an artificial intelligence model, performs the initial reading. Once the initial reading by the primary interpretation expert (310) is completed, a secondary interpretation expert (320), including an electrocardiogram interpretation expert, automatically or at the request of the electrocardiogram interpretation requester (200), performs a precise reading to complete the reading report. The reading report completed by the secondary interpretation expert (320) is then transmitted to the electrocardiogram interpretation requester (200). The conventional electrocardiogram interpretation center (300) aims to increase the reliability and accuracy of the reading through this two-step process.

[0004] However, the existing reading process, such as the example described above, has the disadvantage of requiring a large amount of time and cost because it performs readings through the same steps for all requested ECGs. For example, if an ECG reading is requested for a healthy person who does not need to perform a two-step reading, it may be more efficient to perform readings only for the primary reader (310) or not even for the primary reader (310). In other words, since the level of precision required for ECG reading may vary depending on the condition of the subject whose ECG is measured, it is necessary to perform readings by applying an appropriate reading process according to the condition of the subject whose ECG is measured. The reading process of the existing reading center cannot meet these needs.

[0005]

[0006] The present disclosure aims to provide a method for efficiently improving the existing electrocardiogram reading process, which was performed mechanically, through pre-screening that classifies electrocardiogram data requested for reading prior to reading.

[0007] However, the problems to be solved in this disclosure are not limited to the problems mentioned above, and other problems not mentioned can be clearly understood based on the description below.

[0008]

[0009] According to one embodiment of the present disclosure for achieving the aforementioned task, a data classification method of an electrocardiogram interpretation center, performed by a computing device, is disclosed. The method may include the steps of inputting electrocardiogram data into a pre-trained first model and performing a first classification to classify the electrocardiogram data as normal or abnormal; and, based on the result of the first classification, determining whether to perform a first interpretation process for analyzing the possibility of a disease or the health status of a subject whose electrocardiogram data has been measured in multiple stages. The normality or abnormality may be related to the possibility of a disease or the health status of the subject whose electrocardiogram data has been measured.

[0010] Alternatively, the first model may be trained based on training data constructed from electrocardiogram data containing abnormalities in the readings.

[0011] Alternatively, the first reading process may perform a first reading through a neural network-based reading model or a first reading expert terminal based on the electrocardiogram data, and perform a second reading through a second reading expert terminal based on the electrocardiogram data and the first reading result.

[0012] Alternatively, the step of determining whether to perform a first interpretation process for analyzing the possibility of a disease or the health status of the subject whose electrocardiogram data was measured in multiple stages based on the result of the first classification may include the step of determining, if the electrocardiogram data is classified as abnormal, to perform the first interpretation process based on the electrocardiogram data classified as abnormal at the electrocardiogram interpretation center; and the step of performing a second classification for distinguishing data corresponding to the boundary between normal and abnormal and data not corresponding to the boundary by inputting the electrocardiogram data classified as normal into a pre-trained second model, if the electrocardiogram data is classified as normal.

[0013] Alternatively, the second model may be trained based on training data constructed from electrocardiogram data in which the readings include normal sinus rhythm and the electrocardiogram feature values ​​fall within the normal range.

[0014] Alternatively, the method may further include a step of determining whether to perform a second reading process for analyzing the disease possibility or health status of the subject whose electrocardiogram data was measured as part of the multi-step included in the first reading process, based on the results of the second classification.

[0015] Alternatively, the second reading process may be a process of performing a reading through a second reading expert terminal based on the electrocardiogram data classified as normal.

[0016] Alternatively, the step of determining whether to perform a second reading process for analyzing the possibility of a disease or the health status of the subject whose electrocardiogram data was measured based on the result of the second classification as part of the multi-step included in the first reading process may include the step of determining, in the case where the electrocardiogram data is classified as data corresponding to the boundary between normal and abnormal, that the second reading process is performed based on the electrocardiogram data classified as data corresponding to the boundary at the electrocardiogram reading center; and the step of determining, in the case where the electrocardiogram data is classified as data not corresponding to the boundary, that the electrocardiogram data is normal.

[0017] According to one embodiment of the present disclosure for achieving the aforementioned 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 causes an electrocardiogram interpretation center to perform operations for classifying data. At this time, the operations may include an operation of inputting electrocardiogram data into a pre-trained first model and performing a first classification to classify the electrocardiogram data as normal or abnormal; and an operation of determining whether to perform a first interpretation process for analyzing the possibility of a disease or the health status of a subject whose electrocardiogram data was measured in multiple stages based on the result of the first classification.

[0018] According to one embodiment of the present disclosure for realizing the task described above, a computing device for classifying data in an electrocardiogram interpretation center 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 electrocardiogram data. In this case, the processor may input the acquired electrocardiogram data into a pre-learned first model, perform a first classification for classifying the electrocardiogram data as normal or abnormal, and determine whether to perform a first interpretation process for analyzing the possibility of a disease or the health status of a subject whose electrocardiogram data was measured in multiple stages based on the result of the first classification.

[0019]

[0020] According to the method of the present disclosure, by classifying electrocardiograms that do not require reading or precise reading prior to reading, the reading process that requires multiple steps at an electrocardiogram reading center can be efficiently performed, and the cost and time required for the operation of the reading center can be optimized.

[0021]

[0022] Figure 1 is a conceptual diagram showing the existing interpretation process of an electrocardiogram interpretation center.

[0023] FIG. 2 is a block diagram of a computing device according to one embodiment of the present disclosure.

[0024] FIG. 3 is a conceptual diagram illustrating an electrocardiogram reading process according to one embodiment of the present disclosure.

[0025] FIG. 4 is a flowchart illustrating a data classification process for electrocardiogram reading according to one embodiment of the present disclosure.

[0026] FIG. 5 is a flowchart illustrating a data classification method of an electrocardiogram reading center according to one embodiment of the present disclosure.

[0027]

[0028] Below, embodiments of the present disclosure are described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present disclosure. The embodiments presented in this disclosure are provided to enable those skilled in the art to utilize or implement the contents of the present disclosure. Accordingly, various modifications to the embodiments of the present disclosure will be apparent to those skilled in the art. That is, the present disclosure may be implemented in various different forms and is not limited to the embodiments described below.

[0029] Throughout the specification of this disclosure, identical or similar drawing numbers refer to identical or similar components. Furthermore, for the purpose of clearly describing the disclosure, drawing numbers for parts in the drawings that are not relevant to the description of the disclosure may be omitted.

[0030] The term "or" as used herein is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified herein or clear from context, "X employs A or B" should be understood to mean either of the natural inclusive permutations. For example, unless otherwise specified herein or clear from context, "X employs A or B" can be interpreted to mean either X employs A, X employs B, or X employs both A and B.

[0031] The term "and / or" as used herein should be understood to refer to and include all possible combinations of one or more of the related concepts listed.

[0032] The terms "comprises" and / or "comprising" as used herein should be understood to mean the presence of certain features and / or components. However, it should be understood that the terms "comprises" and / or "comprising" do not exclude the presence or addition of one or more other features, other components, and / or combinations thereof.

[0033] 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.”

[0034] The term "Nth (N is a natural number)" used in the present disclosure can be understood as an expression used to mutually distinguish components of the present disclosure based on a predetermined standard such as a functional perspective, a structural perspective, or convenience of explanation. For example, components performing different functional roles in the present disclosure can be distinguished as a first component or a second component. However, components that are substantially the same within the technical spirit of the present disclosure but must be distinguished for convenience of explanation may also be distinguished as a first component or a second component.

[0035] The term "acquisition" as used in this disclosure may be understood to mean not only receiving data through a wired or wireless communication network with an external device or system, but also generating data in an on-device form.

[0036] Meanwhile, the term "module" or "unit" used in the present disclosure can be understood as a term referring to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a part thereof, hardware or a part thereof, or a combination of software and hardware. At this time, the "module" or "unit" may be a unit composed of a single element, or a unit expressed as a combination or set of multiple elements. For example, as a narrow concept, a "module" or "unit" may refer to a hardware element of a computing device or a set thereof, an application program that performs a specific function of software, a processing process implemented through software execution, or a set of instructions for program execution, etc. In addition, as a broad concept, a "module" or "unit" may refer to the computing device itself that constitutes the system, or an application running on the computing device, etc. However, since the above-described concept is only an example, the concept of “module” or “part” may be defined in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0037] The term "model" as used herein may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a set of software units to solve a specific problem, or an abstract model of a processing process to solve a specific problem. For example, a neural network "model" may refer to the entire system implemented as a neural network that has problem-solving capabilities through learning. In this case, the neural network can have problem-solving capabilities by optimizing the parameters connecting nodes or neurons through learning. A neural network "model" may include a single neural network or a set of neural networks that are a combination of multiple neural networks.

[0038] The explanation of the above terms is intended to aid understanding of the present disclosure. Therefore, unless explicitly stated as limiting the contents of the present disclosure, it should be noted that the above terms are not intended to limit the technical ideas of the contents of the present disclosure.

[0039] FIG. 2 is a block diagram of a computing device according to one embodiment of the present disclosure.

[0040] A 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 intensive data processing functions and shares resources, or may be a client that shares resources through interaction with a server. In addition, the computing device (100) may be a cloud system in which multiple servers and clients interact to comprehensively process data. Since the above description is only one example related to the type of computing device (100), the type of computing device (100) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0041] Referring to FIG. 2, 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. 2 is merely an example, and the computing device (100) may include other components for implementing a computing environment. In addition, only some of the disclosed components may be included in the computing device (100).

[0042] 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 input data for machine learning, feature extraction for machine learning, and error calculation based on backpropagation. The processor (110) for performing such data processing may include a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA). The above-described type of the processor (110) 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.

[0043] When electrocardiogram (ECG) data is transmitted to an ECG interpretation center, the processor (110) can classify the ECG data to determine how to proceed with the interpretation process through the ECG interpretation center. The processor (110) can classify the ECG data transmitted to the ECG interpretation center as either normal, abnormal, or a boundary between normal and abnormal. In this case, normal or abnormal may be a characteristic related to the possibility of a disease or the health status of the subject whose ECG data was measured. If the possibility of a disease is high, the ECG data may be classified as abnormal. Conversely, if the possibility of a disease is low, the ECG data may be classified as normal. The high or low possibility of a disease can be determined using a pre-trained artificial intelligence model based on the interpretation text generated from the ECG data interpretation. The health status can be classified as good, average, or poor as a measure of health awareness, and this classification can be determined using a pre-trained artificial intelligence model based on the interpretation text generated from the ECG data interpretation. In addition, the boundary between normal and abnormal can be determined based on the ECG data determined to be normal as described above. The boundary between normal and abnormal can be determined using a pre-trained artificial intelligence model using the interpretation text generated as a result of reading electrocardiogram data.

[0044] For example, the processor (110) may input electrocardiogram data into a pre-trained first model to perform a first classification that classifies the electrocardiogram data as normal or abnormal. Through the first classification, the processor (110) may distinguish data that requires precise reading from data that does not require precise reading. At this time, the first model may be a model trained based on training data constructed with electrocardiogram data that includes abnormalities in the reading text. Specifically, the first model may receive training data including electrocardiogram data labeled as abnormal through a previous reading and classify it as normal or abnormal. The results of the first model classifying as normal or abnormal and the labels of the training data input to the first model may be compared through a loss function such as mean squared error (MSE) cross entropy. If an error is calculated during the comparison process using the loss function, the first model may be trained through a task in which the parameters constituting the first model are adjusted based on the calculated error. The first model can be trained through supervised learning as described above, but is not limited thereto and can also be trained through unsupervised learning, self-supervised learning, etc.

[0045] And, if the electrocardiogram data is classified as normal through the first classification, the processor (110) can input the electrocardiogram data classified as normal into a pre-trained second model to perform a second classification that distinguishes between data that falls on the border between normal and abnormal and data that does not fall on the border. The processor (110) can identify data that is on the border between abnormal and abnormal among the normal data through the second classification, and distinguish between data that requires interpretation and data that does not require interpretation itself. At this time, the second model may be a model learned based on learning data constructed from electrocardiogram data that is interpreted as very normal with an extremely low possibility of disease among data interpreted as normal through existing interpretation. At this time, the very normal data may be electrocardiogram data that includes a normal sinus rhythm in the interpretation text and an electrocardiogram feature value that falls within the normal category. And, the normal category of the electrocardiogram feature value can be understood to mean a category in which electrocardiogram feature values ​​such as heart rate, QTc, PR interval, QRS duration, P axis, and R axis are judged to be normal based on clinical experience and research. Specifically, the second model can receive training data including electrocardiogram data labeled as non-boundary because normal sinus rhythm is described in the reading text and the electrocardiogram feature values ​​are described as within the normal category, and classify them as boundary or non-boundary. The results classified by the second model and the labels of the training data input to the second model can be compared using a loss function such as mean squared error (MSE) cross entropy. When an error is calculated in the comparison process using the loss function, the second model can be trained by adjusting the parameters constituting the second model based on the calculated error.The second model can be trained through supervised learning as described above, but is not limited thereto and can also be trained through unsupervised learning, self-supervised learning, etc.

[0046] The processor (110) may determine which of several reading processes that can be performed at the electrocardiogram reading center will be performed based on the classification result. When the electrocardiogram reading center performs a multi-step reading process that basically passes through multiple readers to perform readings, the processor (110) may perform readings through only some of the readers or decide not to perform readings based on the classification result. If the multi-step reading process is performed on data that does not require reading or does not require precise reading, a significant amount of resources may be required for the operation of the reading center. Therefore, the processor (110) can create an environment in which the reading center can operate efficiently by classifying the electrocardiogram data before performing the reading and determining an appropriate reading process based on the classification.

[0047] For example, the processor (110) may determine whether to perform a first reading process that analyzes the possibility of a disease or the health status of a subject whose electrocardiogram data has been measured in multiple stages based on the result of the first classification that distinguishes between normal and abnormal. If the electrocardiogram data is classified as abnormal, the processor (110) may determine that the electrocardiogram reading center performs the first reading process based on the electrocardiogram data classified as abnormal. That is, if the electrocardiogram data is classified as abnormal by the processor (110), the electrocardiogram reading center may perform the first reading process to analyze the possibility of a disease or the health status of the subject whose electrocardiogram data has been measured. On the other hand, if the electrocardiogram data is classified as normal, the processor (110) may perform a second classification that distinguishes between data that falls on the border between normal and abnormal and data that does not fall on the border, as described above. And, based on the result of the second classification, the processor (110) can determine whether to perform a second reading process that analyzes the disease possibility or health condition of the subject whose electrocardiogram data was measured as part of the multi-step included in the first reading process. If the first reading process performs a precise reading through multiple readers, the second reading process can be understood as performing a quick reading through some of the multiple readers included in the first reading process. If the electrocardiogram data is classified as data corresponding to the borderline between normal and abnormal, the processor (110) can determine to perform the second reading process based on the electrocardiogram data classified as borderline data by the electrocardiogram reading center. That is, if the electrocardiogram data is classified as borderline by the processor (110), the electrocardiogram reading center can perform a second reading process that is simpler than the first reading process to analyze the disease possibility or health condition of the subject whose electrocardiogram data was measured.On the other hand, if the electrocardiogram data is classified as non-boundary data, the processor (110) may determine that the electrocardiogram data is normal. That is, if the electrocardiogram data is classified as non-boundary by the processor (110), the electrocardiogram interpretation center may inform the electrocardiogram interpretation requester that the subject whose electrocardiogram was measured is normal without performing the first interpretation process or the second interpretation process.

[0048] 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 random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory (120) may also include a database system that controls and manages data in a predetermined system. The type of memory (120) described above is only one example, and thus the type of memory (120) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.

[0049] The memory (120) can structure and organize and manage data, combinations of data, and program codes executable by the processor (110) required for the processor (110) to perform operations. For example, the memory (120) can store electrocardiogram data received through the network unit (130) described below. The memory (120) can store program codes that operate a neural network model to receive electrocardiogram data as input and perform learning, program codes that operate a neural network model to receive electrocardiogram data as input and perform inference according to the purpose of use of the computing device (100), and processed data generated as the program codes are executed.

[0050] The network unit (130) according to one embodiment of the present disclosure may be understood as a component that transmits and receives data through any type of known wired or wireless communication system. For example, the network unit (130) may perform data transmission and reception using a wired or wireless communication system such as a local area network (LAN), wideband code division multiple access (WCDMA), long term evolution (LTE), wireless broadband internet (WiBro), fifth generation mobile communication (5G), 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.

[0051] The network unit (130) can receive data necessary for the processor (110) to perform calculations through wired or wireless communication with any system or any client, etc. In addition, the network unit (130) can transmit data generated through calculations of the processor (110) through wired or wireless communication with any system or any client, etc. For example, the 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, etc. The network unit (130) can transmit output data of a neural network 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.

[0052] FIG. 3 is a conceptual diagram illustrating an electrocardiogram reading process according to one embodiment of the present disclosure.

[0053] The existing reading center (300) of Fig. 1 performs readings through two readers (310, 320) in a batch manner for all cases where precise reading is not necessary or reading itself is not necessary, and therefore the reading itself is bound to be performed inefficiently. Therefore, the electrocardiogram reading center (400) according to an embodiment of the present disclosure as shown in Fig. 3 does not perform readings as a single process through two readers (310, 320) like the existing reading center (300), but aims to increase the efficiency of reading by dividing the reading process into several branches. In order to perform the reading process by dividing it into several branches, the electrocardiogram reading center (400) may include a classifier (410) that classifies data received from the reading requester (200) into normal, abnormal, or the boundary between normal and abnormal. The electrocardiogram interpretation center (400) can selectively perform one of three interpretation processes based on the classification results obtained through the classifier (410).

[0054] Specifically, the classifier (410) can determine whether the electrocardiogram data received from the reading requester (200) is abnormal using the pre-trained first model. The classifier (410) can filter out data requiring precise reading by determining whether the data for which reading is requested is abnormal using the pre-trained first model. At this time, the first model may be a model in which multiple neural networks are combined. If the electrocardiogram data is determined to be abnormal, the reading center (400) can perform process ⓐ of performing reading using both the primary reader (420) and the secondary reader (430). If the electrocardiogram data is determined to be normal, the classifier (410) can determine whether the data determined to be normal corresponds to the boundary between normal and abnormal using the pre-trained second model. The classifier (410) can filter out data that does not require reading by additionally determining whether the data determined to be normal is located on the boundary between normal and abnormal using the pre-trained second model. At this time, the second model may be a model that combines multiple neural networks and may be an independent model that is distinct from the first model. If the electrocardiogram data is determined to not be on the borderline between normal and abnormal, the interpretation center (400) may perform process ⓑ to immediately notify the interpretation requester (200) that the data is in a normal state and does not require interpretation, without going through both the primary interpreter (420) and the secondary interpreter (430). If the electrocardiogram data is determined to be on the borderline between normal and abnormal, the interpretation center (400) may perform process ⓒ to perform interpretation through the secondary interpreter (430) without going through the primary interpreter (420). That is, the interpretation center (400) may not perform only process ⓐ for all data requested for interpretation, but may selectively perform process ⓐ, process ⓑ, or process ⓒ based on the classification result using the classifier (410).Through this, the reading center (400) can optimize the resources required for reading using the primary reader (420) and secondary reader (430).

[0055] FIG. 4 is a flowchart illustrating a data classification process for electrocardiogram reading according to one embodiment of the present disclosure.

[0056] A computing device (100) according to one embodiment of the present disclosure may perform a classification task on electrocardiogram data (10) requested for reading to determine whether precise reading of the electrocardiogram data is required, and if precise reading is not required, whether reading itself is not required. In addition, the computing device (100) may determine which of the reading processes that can be performed at an electrocardiogram reading center is appropriate for reading the electrocardiogram data (10) based on the classification result of the electrocardiogram data (10).

[0057] Referring to FIG. 4, the computing device (100) can perform a first classification to distinguish between normal and abnormal electrocardiogram data (10) requested for interpretation (S110). The computing device (100) can input the electrocardiogram data (10) into a pre-learned first model to distinguish the electrocardiogram data (10) as normal or abnormal. At this time, data classified as abnormal can be understood to correspond to data that is interpreted as indicating that the condition of the person who measured the data is abnormal or has a high probability of being ill. The abnormality of the person's condition or high probability of being ill can be determined based on a reference value determined based on clinical experience and research. Therefore, if the electrocardiogram data (10) is classified as abnormal through the first classification (S110), the computing device (100) can decide to perform a first interpretation process at an electrocardiogram interpretation center (S120). The first interpretation process is a multi-stage process that analyzes the disease possibility or health status of the subject whose electrocardiogram data (10) has been measured, and can be understood as a process that performs the most precise interpretation at the electrocardiogram interpretation center. Specifically, the first interpretation process may be a process that performs a first interpretation through a neural network-based interpretation model or a first interpretation expert terminal based on the electrocardiogram data (10), and performs a second interpretation through a second interpretation expert terminal based on the electrocardiogram data (10) and the first interpretation result. The neural network-based interpretation model may be a model that interprets the possibility that the subject whose electrocardiogram data (10) has been measured is suffering from any one of a plurality of diseases or the health status of the subject whose electrocardiogram data (10) has been measured.

[0058] Meanwhile, if the electrocardiogram data (10) is classified as normal through the first classification (S110), the computing device (100) can perform a second classification to distinguish whether the electrocardiogram data (10) classified as normal corresponds to the borderline of normal or abnormal or not corresponds to the borderline (S130). The computing device (100) can input the electrocardiogram data (10) into a pre-trained second model to classify the electrocardiogram data (10) into borderline or non-borderline. At this time, data classified as non-borderline can be understood to correspond to data that is judged to be very normal or have a significantly low possibility of disease for the person who measured the data. The fact that the person is very normal or has a significantly low possibility of disease can be determined based on whether the electrocardiogram has a normal sinus rhythm and whether the electrocardiogram feature values ​​fall within the normal range. Whether the electrocardiogram feature values ​​fall within the normal range can be determined based on whether the criteria set based on clinical experience and research as shown in [Table 1] below are satisfied.

[0059] -30<=R axis<=90-30<=P axis<=90120<=PR interval<200QRS duration<11060<=Heart rate<=100Male: QTc<450 / Female: QTc<460

[0060] If the electrocardiogram data (10) is classified as non-borderline through the second classification (S130), the computing device (100) may determine that the electrocardiogram data is normal (S140). Since the electrocardiogram data is normal, the computing device (100) may determine not to perform an additional reading process at the electrocardiogram interpretation center. If the electrocardiogram data (10) is classified as borderline through the second classification (S130), the computing device (100) may determine to perform a second reading process at the electrocardiogram interpretation center (S150). The second reading process is a process of analyzing the possibility of a disease or the health status of the subject whose electrocardiogram data (10) was measured by performing some of the multi-steps that constitute the first reading process, and may be understood as a process in which the electrocardiogram interpretation center performs a reading that is not as precise as the first reading process but is quick. Specifically, the second reading process may be a process of performing a single reading through a second reading expert terminal based on electrocardiogram data (10) classified as normal and borderline. That is, the second reading process may be understood as performing only a second reading without performing a first reading on the first reading process. Since the electrocardiogram data (10) has already been determined to correspond to the borderline between normal and abnormal through the first classification (S110) and the second classification (130), the electrocardiogram reading center can obtain results with as high an accuracy as the first reading process while being more efficient than the first reading process by performing the second reading process.

[0061] FIG. 5 is a flowchart illustrating a data classification method of an electrocardiogram reading center according to one embodiment of the present disclosure.

[0062] Referring to FIG. 5, an electrocardiogram interpretation center according to one embodiment of the present disclosure may input electrocardiogram data into a pre-trained first model and perform a first classification to classify the electrocardiogram data as normal or abnormal (S210). In this case, normal or abnormal may be related to the possibility of a disease or the health status of the subject whose electrocardiogram data was measured. For example, if the possibility of a disease of the subject whose electrocardiogram data was measured is determined to be high through the first model, the electrocardiogram data may be classified as abnormal. In addition, the electrocardiogram interpretation center may determine, based on the results of the first classification, whether to perform a first interpretation process that analyzes the possibility of a disease of the subject whose electrocardiogram data was measured in multiple stages.

[0063] If the ECG data is classified as abnormal, the ECG interpretation center may decide to perform a first interpretation process based on the ECG data classified as abnormal (S220). For example, the ECG interpretation center may perform a first interpretation based on the ECG data classified as abnormal using a neural network-based interpretation model or a first interpretation expert terminal. The ECG interpretation center may perform a second interpretation based on the ECG data classified as abnormal and the first interpretation results using a second interpretation expert terminal. Through this multi-step process, the ECG interpretation center can perform a precise interpretation of the ECG data classified as abnormal.

[0064] If the electrocardiogram data is classified as normal, the electrocardiogram interpretation center can input the electrocardiogram data classified as normal into a pre-trained second model to perform a second classification that distinguishes data that falls within the boundary between normal and abnormal from data that does not fall within the boundary (S230). At this time, data that does not fall within the boundary may correspond to electrocardiogram data that includes a normal sinus rhythm in the interpretation text and whose electrocardiogram feature values ​​fall within the normal category. Based on the results of the second classification, the electrocardiogram interpretation center can determine whether to perform a second interpretation process that analyzes the possibility of a disease or health condition of the subject whose electrocardiogram data was measured as part of the multi-step included in the first interpretation process.

[0065] If the electrocardiogram data is classified as data corresponding to the borderline between normal and abnormal, the electrocardiogram interpretation center may determine to perform a second interpretation process based on the electrocardiogram data classified as borderline data (S240). For example, the electrocardiogram interpretation center may perform an interpretation corresponding to the second interpretation process, excluding the first interpretation of the first interpretation process, based on the electrocardiogram data classified as borderline data. In other words, the electrocardiogram interpretation center may perform an interpretation through the second interpretation expert terminal based on the electrocardiogram data classified as borderline data. The electrocardiogram interpretation center can perform a rapid interpretation of the electrocardiogram data classified as borderline between normal and abnormal through the second interpretation process.

[0066] If the ECG data is classified as non-borderline, the ECG interpretation center may determine that the data is normal (S250). In other words, the ECG interpretation center may determine that no further interpretation of the ECG data is necessary. Furthermore, the ECG interpretation center may provide feedback, such as a report indicating the ECG data is normal, to the requester.

[0067] Meanwhile, the electrocardiogram reading center described with reference to FIG. 5 may be the computing device (100) of FIG. 2 itself or may be an upper device including the computing device (100).

[0068] The various embodiments of the present disclosure described above can be combined with additional embodiments and modified within the scope understood by those skilled in the art in light of the detailed description above. It should be understood that the embodiments of the present disclosure are illustrative in all respects and not restrictive. For example, each component described as a single component may be implemented in a distributed manner, and likewise, components described as distributed may be implemented in a combined manner. Accordingly, all changes or modifications derived from the meaning, scope, and equivalent concepts of the claims of the present disclosure should be construed as being included within the scope of the present disclosure.

Claims

1. A data classification method of an electrocardiogram reading center, performed by a computing device including at least one processor, A step of inputting electrocardiogram data into a pre-trained first model and performing a first classification to classify the electrocardiogram data as normal or abnormal; and A step of determining whether to perform a first reading process for analyzing the disease possibility or health status of the subject whose electrocardiogram data was measured in multiple stages based on the result of the first classification; Including, The above normal or abnormal, Related to the possibility of disease or health status of the subject whose electrocardiogram data was measured above, method.

2. In paragraph 1, The above first model is, It is learned based on training data constructed from electrocardiogram data that contains abnormalities in the reading text. method.

3. In paragraph 1, The above first reading process is, A first reading is performed through a neural network-based reading model or a first reading expert terminal based on the above electrocardiogram data, and a second reading is performed through a second reading expert terminal based on the above electrocardiogram data and the first reading result. method.

4. In paragraph 1, Based on the results of the first classification, the step of determining whether to perform the first reading process for analyzing the disease possibility or health status of the subject whose electrocardiogram data was measured in multiple stages is as follows: If the electrocardiogram data is classified as abnormal, a step of determining that the electrocardiogram reading center performs the first reading process based on the electrocardiogram data classified as abnormal; and If the electrocardiogram data is classified as normal, a step of inputting the electrocardiogram data classified as normal into a pre-trained second model to perform a second classification to distinguish between data corresponding to the boundary between normal and abnormal and data not corresponding to the boundary; including, method.

5. In paragraph 4, The second model above is, The training data is built based on electrocardiogram data that contains normal sinus rhythm in the reading text and electrocardiogram feature values ​​that fall within the normal range. method.

6. In paragraph 4, A step of determining whether to perform a second reading process for analyzing the disease possibility or health status of the subject whose electrocardiogram data was measured based on the result of the second classification as part of the multi-step included in the first reading process; including more, method.

7. In paragraph 6, The above second reading process is, Based on the electrocardiogram data classified as normal above, a reading is performed through a second reading expert terminal. method.

8. In paragraph 6, Based on the results of the second classification, the step of determining whether to perform a second reading process that analyzes the disease possibility or health status of the subject whose electrocardiogram data was measured as part of the multi-step included in the first reading process is as follows: If the electrocardiogram data is classified as data corresponding to the boundary between normal and abnormal, a step of determining that the second reading process is performed based on the electrocardiogram data classified as data corresponding to the boundary at the electrocardiogram reading center; and A step of determining that the electrocardiogram data is normal when the electrocardiogram data is classified as data that does not fall within the above boundary; including, method.

9. A computer program stored in a computer-readable storage medium, wherein the computer program, when executed on one or more processors, causes an electrocardiogram interpretation center to perform operations for classifying data. The above actions are, An operation of inputting electrocardiogram data into a pre-trained first model and performing a first classification to classify the electrocardiogram data as normal or abnormal; and An operation of determining whether to perform a first reading process that analyzes the possibility of a disease or health condition of a subject whose electrocardiogram data was measured in multiple stages based on the result of the first classification; Including, The above normal or abnormal, Related to the possibility of disease or health status of the subject whose electrocardiogram data was measured above, including, Computer program.

10. As a computing device for classifying data in an electrocardiogram interpretation center, A processor comprising at least one core; A memory containing program codes executable by the processor; and A network unit for acquiring electrocardiogram data; Including, The above processor, By inputting the acquired electrocardiogram data into the pre-trained first model, a first classification is performed to classify the electrocardiogram data as normal or abnormal, Based on the results of the first classification, it is determined whether to perform a first reading process that analyzes the disease possibility or health status of the subject whose electrocardiogram data was measured in multiple stages. The above normal or abnormal, Related to the possibility of disease or health status of the subject whose electrocardiogram data was measured above, device.

Images