Health condition prediction system using asynchronous electrocardiograms

Abstract

The present invention discloses a health condition prediction system using an asynchronous electrocardiogram, which includes a single-lead electrocardiogram measurement unit (110) having two electrodes and measuring electrocardiograms of two or more electrical axes with a time difference to obtain asynchronous electrocardiogram data, and a prediction unit (120) for predicting the presence or absence of a disease and the degree of a disease from the asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit (110) and segmented by a specific time unit according to a diagnostic algorithm (121) previously constructed by learning a plurality of asynchronous segmented standard electrocardiogram datasets in which asynchronous segmented standard lead electrocardiograms at different points in time of an asynchronous standard lead electrocardiogram segmented by a specific time unit from a synchronous standard lead electrocardiogram measured on the same electrical axis and stored in a server of a medical institution are matched with the presence or absence of a disease and the degree of a disease corresponding to the asynchronous segmented standard lead electrocardiogram, and the system segments the synchronous electrocardiogram of the medical institution to generate an asynchronous electrocardiogram and generate a dataset for learning a prediction model, thereby predicting a disease from asynchronous electrocardiogram data of two or more leads measured by a single-lead electrocardiogram device.

Description

[Technical Field]

[0001] The present invention relates to a health condition prediction system using asynchronous electrocardiograms, which can predict diseases from asynchronous electrocardiogram data of two or more leads measured by a single-lead electrocardiogram device by segmenting various types of synchronized electrocardiograms accumulated in medical institutions, extracting asynchronous electrocardiograms of two or more leads, and generating a dataset for training a prediction model.

[0002] [Background technology]

[0003] As is well known, after the development of the electrocardiogram, knowledge related to electrocardiograms has expanded exponentially, and electrocardiograms can be used to obtain information about the electrical function of the heart and diagnose various heart diseases such as arrhythmias, coronary artery disease, and myocardial disease.

[0004] Recently, active research has been conducted on AI algorithms for electrocardiograms, and AI algorithms can be used to detect heart failure, predict atrial fibrillation from arrhythmia rhythms, or even determine gender.

[0005] In this way, by overcoming human limitations, AI algorithms can detect subtle changes in ECG waveforms, thereby improving ECG interpretation.

[0006] Meanwhile, the electrocardiogram used in the medical field is a 12-lead electrocardiogram, which is measured by attaching 10 electrodes: 3 limb electrodes, 6 chest electrodes, and 1 ground electrode, and the measured electrocardiogram data can be transmitted remotely.

[0007] However, since it is inconvenient to expose the chest and attach 10 electrodes in daily life, some people use portable pad measuring devices, Galaxy Watch, Apple Watch, etc. that can measure single-lead ECGs.

[0008] Although the single-lead electrocardiogram measured in this manner can be used to diagnose arrhythmia, it has limitations in its use in diagnosing diseases that require electrocardiogram information from multiple leads, such as myocardial infarction.

[0009] Therefore, there is a demand for technology that can easily measure two or more leads of asynchronous electrocardiograms using a single-lead electrocardiogram device or a six-lead electrocardiogram device, and that can predict health status from the measured two or more leads of asynchronous electrocardiogram data.

[0010] Summary of the Invention [Problem to be solved by the invention]

[0011] The technical problem that the concept of the present invention aims to achieve is to provide a health condition prediction system using asynchronous electrocardiograms that can predict diseases from asynchronous electrocardiogram data of two or more leads measured by a single-lead electrocardiogram device by segmenting various types of synchronized electrocardiograms accumulated in medical institutions, extracting asynchronous electrocardiograms of two or more leads, and generating a dataset for training a prediction model.

[0012] [Means for solving the problem]

[0013] To achieve the above-mentioned object, an embodiment of the present invention provides a health condition prediction system using an asynchronous electrocardiogram, including: a single-lead electrocardiogram measurement unit having two electrodes and measuring electrocardiograms of two or more electrical axes with a time difference to obtain asynchronous electrocardiogram data; and a prediction unit that predicts the presence or absence of a disease and the severity of a disease from the asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit and segmented by a specific time unit using a diagnostic algorithm that is pre-constructed by learning asynchronous segmental standard electrocardiograms at different times of asynchronous standard lead electrocardiograms segmented by a specific time unit from asynchronous standard lead electrocardiograms measured on the same electrical axis and stored in a server of a medical institution, and a plurality of asynchronous segmental standard electrocardiogram data sets that match the presence or absence of a disease and the severity of a disease corresponding to the asynchronous segmental standard lead electrocardiograms.

[0014]

[0015] Here, the device may further include an electrocardiogram data generating unit that grasps characteristics of a plurality of asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit, identifies specific lead electrocardiogram data by grasping characteristics of the plurality of asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit, and generates a plurality of remaining standard lead electrocardiogram data that do not correspond to the identified specific lead electrocardiogram data. The diagnostic algorithm may input the measured asynchronous electrocardiogram data and the generated standard lead electrocardiogram data, or may input the generated standard lead electrocardiogram data, and output and predict the presence or absence of a disease and the severity of the disease.

[0016]

[0017] The electrocardiogram data generating unit can generate a plurality of synchronized standard lead electrocardiogram data.

[0018]

[0019] In addition, the diagnostic algorithm may generate the plurality of asynchronous segmented standard electrocardiogram data sets by asynchronously extracting synchronized segmented standard electrocardiograms obtained by segmenting standard lead electrocardiogram data stored in a synchronized form at specific time units.

[0020]

[0021] The diagnostic algorithm may be configured to input standard lead ECG data obtained by removing time-series information from synchronously measured standard lead ECG data, and output and predict the presence or absence and severity of a disease corresponding to the standard lead ECG data.

[0022]

[0023] In addition, the diagnostic algorithm learns by reflecting the individual characteristic information of the subject, and the prediction unit can predict the presence or absence of disease and the severity of the disease from the asynchronous electrocardiogram data obtained by the single-lead electrocardiogram measurement unit by reflecting the individual characteristic information.

[0024]

[0025] In addition, the individual characteristic information may include demographic information such as gender and age, basic health information such as weight, height, and obesity level, disease-related information such as past history, drug history, and family history, and test information on vital signs and vital signs such as blood tests, genetic tests, blood pressure, and oxygen saturation.

[0026]

[0027] [Effects of the Invention]

[0028] According to the present invention, it is possible to easily measure asynchronous electrocardiograms of two or more leads using a single-lead electrocardiogram device or a six-lead electrocardiogram device, and it is possible to segment various types of synchronized electrocardiograms stored in medical institutions to extract asynchronous electrocardiograms of two or more leads, thereby generating a dataset for training a prediction model.

[0029] In addition, there is an effect that diseases can be predicted from asynchronous electrocardiogram data of two or more leads measured by a single-lead electrocardiogram device or a six-lead electrocardiogram device.

[0030] Furthermore, by generating multi-channel electrocardiogram data from asynchronous electrocardiogram data of two or more leads, it is possible to output and predict the presence or absence of a disease and the severity of the disease with higher accuracy.

[0031]

[0032] [Brief explanation of the drawings]

[0033] [Figure 1] 1 is a schematic diagram illustrating the configuration of a health condition prediction system using an asynchronous electrocardiogram according to an embodiment of the present invention.

[0034] [Figure 2] 2 is a flowchart of a prediction method by the health condition prediction system using asynchronous electrocardiograms of FIG. 1.

[0035] DETAILED DESCRIPTION OF THE INVENTION

[0036] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the preferred embodiments of the present invention having the above-mentioned features will be described in more detail with reference to the accompanying drawings.

[0037]

[0038] A health condition prediction system using an asynchronous electrocardiogram according to an embodiment of the present invention includes a single-lead electrocardiogram measurement unit 110 having two electrodes and measuring electrocardiograms of two or more electrical axes with a time difference to obtain asynchronous electrocardiogram data, and a prediction unit 120 that predicts the presence or absence of a disease and the severity of a disease from the asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit 110 and segmented by a specific time unit using a diagnostic algorithm 121 that is pre-constructed by learning a plurality of asynchronous segmented standard electrocardiogram datasets that match asynchronous segmented standard lead electrocardiograms at different points in time of asynchronous standard lead electrocardiograms segmented by a specific time unit from asynchronous standard lead electrocardiograms measured on the same electrical axis and stored in a server of a medical institution, and the presence or absence of a disease and the severity of a disease corresponding to the asynchronous segmented standard lead electrocardiograms. The gist of the health condition prediction system using an asynchronous electrocardiogram is to segment the synchronized electrocardiograms of the medical institution to generate asynchronous electrocardiograms and generate a dataset for training a prediction model, and predict a disease from the asynchronous electrocardiogram data of two or more leads measured by a single-lead electrocardiogram device.

[0039]

[0040] The health condition prediction system using an asynchronous electrocardiogram having the above-described configuration will be specifically described in detail below with reference to the drawings.

[0041]

[0042] First, the single-lead electrocardiogram measurement unit 110 is equipped with two electrodes and contacts two points on the subject's body to measure electrocardiograms of two or more electrical axes with a time lag, thereby obtaining asynchronous electrocardiogram data and transmitting it to the prediction unit 120 via a short-range network.

[0043] For example, by placing electrodes on both hands to measure a lead I electrocardiogram corresponding to one electrical axis, and then placing electrodes on the right hand and left ankle, which are different from the previous electrode contact body combination, to measure a lead II electrocardiogram corresponding to another electrical axis, it is possible to measure multiple asynchronous electrocardiograms for two or more different electrical axes.

[0044] In addition, the single-lead electrocardiogram measurement unit 110 can measure asynchronous or synchronous electrocardiograms using a wearable electrocardiogram patch 111 capable of contact or non-contact electrocardiogram measurement during daily life, a smart watch 112, or a six-lead electrocardiogram bar that can be measured in a short period of time, and multiple electrocardiograms can be further generated by the electrocardiogram data generation units 130, 130 and input into a diagnostic algorithm 121 constructed from the standard lead electrocardiogram data to predict the corresponding health condition.

[0045] Here, the single-lead electrocardiogram measurement unit 110 can measure the subject's continuous electrocardiogram and transmit it to the prediction unit 120, or can measure two electrocardiograms at a time interval and transmit them to the prediction unit 120.

[0046]

[0047] Next, the prediction unit 120 is configured to predict the health condition from the segmented asynchronous electrocardiogram data of the single-lead electrocardiogram measurement unit 110 using a pre-trained diagnostic algorithm 121, and predicts the presence or absence of disease and the severity of disease from the asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit 110 and segmented in specific time units using the diagnostic algorithm 121 pre-constructed by learning a plurality of asynchronous segmented standard electrocardiogram data sets that match asynchronous segmented standard lead electrocardiograms at different points in time of asynchronous standard lead electrocardiograms segmented in specific time units from synchronized standard lead electrocardiograms measured on the same electrical axis and stored in the medical institution server, and the presence or absence of disease and the severity of disease corresponding to the asynchronous segmented standard lead electrocardiograms, thereby understanding the health condition of the examinee.

[0048] Here, the asynchronous standard lead ECG can be provided by segmenting a standard lead ECG measured by an ECG device at a medical institution in a specific time unit, for example, in 2.5-second intervals, to obtain segmented standard lead ECGs at different time points.

[0049]

[0050] Meanwhile, the diagnostic algorithm 121 is constructed to input standard lead ECG data measured and segmented asynchronously with respect to any electrical axis, and to output and predict the presence or absence and severity of a disease corresponding to the asynchronous segmented standard lead ECG data, thereby making it possible to predict the health condition of the examinee from the asynchronous ECG data transmitted from the single-lead ECG measurement unit 110.

[0051] Alternatively, the diagnostic algorithm 121 may be configured to input standard lead electrocardiogram data obtained by removing time-series information from standard lead electrocardiogram data measured synchronously for multiple electrical axes, and output and predict the presence or absence and severity of a disease corresponding to the standard lead electrocardiogram data, thereby making it possible to predict the health condition from the asynchronous electrocardiogram data transmitted from the single-lead electrocardiogram measurement unit 110.

[0052] Alternatively, the diagnostic algorithm 121 can be constructed to convert multiple asynchronous electrocardiogram data measured by the single-lead electrocardiogram measurement unit 110 and input at different times into synchronized electrocardiogram data, and predict the presence or absence of a disease and the severity of the disease from the converted synchronized electrocardiogram data.

[0053] Alternatively, the diagnostic algorithm 121 can generate multiple asynchronous segmented standard electrocardiogram data sets by asynchronously extracting synchronized segmented standard lead electrocardiograms that are segmented in specific time units from standard lead electrocardiogram data stored in synchronized form.

[0054] Therefore, the diagnostic algorithm 121 can match multiple asynchronous electrocardiogram data measured by the single-lead electrocardiogram measurement unit 110, input with a time difference, and segmented to standard lead electrocardiogram data measured asynchronously and segmented, or segmented standard lead electrocardiogram data from which time series information has been deleted, or segmented synchronized standard lead electrocardiogram data, respectively.

[0055]

[0056] In addition, the electrocardiogram data generating unit 130 grasps the individual characteristics of the plurality of asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit 110, identifies them as specific lead electrocardiogram data, and generates the remaining plurality of standard lead electrocardiogram data that do not correspond to the identified specific lead electrocardiogram data.The diagnostic algorithm 121 receives the asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit 110 and the standard lead electrocardiogram data generated by the electrocardiogram data generating unit 130 as input, or receives the standard lead electrocardiogram data generated by the electrocardiogram data generating unit 130 as input, and can output and predict the presence or absence of a disease and the severity of the disease.

[0057] For example, when generating multiple standard lead electrocardiogram data by the electrocardiogram data generating unit 130, the individual characteristics can be grasped based on the inherent characteristics of each lead electrocardiogram, the asynchronous electrocardiogram data input from the single-lead electrocardiogram measuring unit 110 can be matched to the corresponding specific standard lead electrocardiogram data, and the remaining standard lead electrocardiogram data that does not match the specific standard lead electrocardiogram data can be generated to generate new multiple standard lead electrocardiogram data.

[0058] The electrocardiogram data generating unit 130 can also generate a plurality of synchronized standard lead electrocardiogram data. That is, the standard lead electrocardiogram data generated by the electrocardiogram data generating unit 130 can be asynchronous or synchronized electrocardiogram. As described above, when the diagnostic algorithm 121 uses standard lead electrocardiogram data measured asynchronously or standard lead electrocardiogram data from which time series information has been removed, the electrocardiogram data generating unit 130 can generate asynchronous electrocardiogram data or electrocardiogram data without considering synchronization.

[0059] As a result, by utilizing the asynchronous electrocardiogram measured and segmented by the single-lead electrocardiogram measurement unit 110 to predict the presence or absence of a disease that should be diagnosed using a multi-lead electrocardiogram and to diagnose the severity of the disease, it is possible to more accurately analyze diseases based on multiple electrocardiogram information and measure, diagnose, examine, and predict health conditions.

[0060] For example, in the case of a disease that can be diagnosed with a single lead, such as arrhythmia, a two-lead electrocardiogram and multiple electrocardiograms generated based on it can be used to generate electrocardiograms for each bit on various electrical axes, enabling more accurate measurement, diagnosis, screening, and prediction of health status.In the case of a disease that can be diagnosed with multiple leads, such as myocardial infarction, multiple electrocardiograms can be further generated to diagnose myocardial infarction.

[0061] Alternatively, the diagnostic algorithm 121 may be a model that measures, diagnoses, examines, and predicts health status using only the electrocardiogram data corresponding to the two leads to be used among the standard lead electrocardiogram data for disease diagnosis, but as mentioned above, it is not limited to two leads and electrocardiogram data of further generated leads can also be used.

[0062] In other words, the electrocardiogram data accumulated at medical institutions is standard 12-lead electrocardiogram data, and the electrocardiogram data generation unit 130 generates a standard 12-lead electrocardiogram from a 2-lead electrocardiogram. By inputting the generated electrocardiogram into an algorithm that measures, diagnoses, examines, and predicts health status using the medical institution's standard 12-lead electrocardiogram data, more accurate prediction results can be output by utilizing the 2-lead electrocardiogram, and a wider range of health status can be predicted.

[0063] For example, when the single-lead electrocardiogram measurement unit 110 measures the electrocardiogram data, if some leads or sections of the electrocardiogram data contain a lot of noise or if the electrode contact is loose and measurement cannot be performed normally, the electrocardiogram data generation unit 130 generates an electrocardiogram with noise-free leads to fill in the missing electrocardiogram data, thereby enabling more accurate measurement, diagnosis, examination, and prediction of health status.

[0064] In addition, the single-lead electrocardiogram measurement unit 110 and the prediction unit 120 measure a reference electrocardiogram of the examinee when he or she is in a normal healthy state to monitor the health condition. Thereafter, the electrocardiogram measured and input in real time by the single-lead electrocardiogram measurement unit 110 during daily life is compared with the reference electrocardiogram to predict whether the electrocardiogram is an electrocardiogram measured without error or whether there is any abnormality in the examinee's health condition. When an error or abnormality is predicted, the warning unit 140 generates warning information and transmits the warning information together with a beep via the single-lead electrocardiogram measurement unit 110 in the form of a smartwatch or a separate smart device.

[0065] Specifically, after initially storing a reference electrocardiogram by the single-lead electrocardiogram measurement unit 110, an additional lead electrocardiogram can be generated and a 12-lead electrocardiogram can be continuously monitored. Here, the smartwatch 112 can be held in the right hand and touched to the abdomen to measure a lead II electrocardiogram and store it as a reference electrocardiogram. Then, the smartwatch 112 can be worn on the left hand and touched with the right hand to measure a lead I electrocardiogram, and multiple electrocardiogram leads including a lead II electrocardiogram can be generated using the smartwatch 112, thereby enabling measurement, diagnosis, examination and prediction of a wider variety of health conditions.

[0066] Alternatively, before putting on the wearable electrocardiogram patch 111, the wearer can take the patch with both hands to measure a lead I electrocardiogram and save it as a reference electrocardiogram, and then put on the wearable electrocardiogram patch 111 to measure a lead V electrocardiogram, thereby generating a synchronized multi-lead electrocardiogram based on the continuously measured or monitored lead V electrocardiogram, thereby enabling more accurate measurement, diagnosis, screening, and prediction of health status.

[0067]

[0068] In addition, the diagnostic algorithm 121 learns by reflecting the individual characteristic information of the subject, and the prediction unit 120 can predict the presence or absence of disease and the severity of the disease from the asynchronous electrocardiogram data obtained by the single-lead electrocardiogram measurement unit 110 by reflecting the individual characteristic information.

[0069] For example, when applying standard lead electrocardiogram data utilized in learning the diagnostic algorithm 121, the presence or absence of a corresponding disease and the severity of the disease can be predicted by reflecting individual characteristic information, or the presence or absence of a corresponding disease and the severity of the disease can be predicted by generating standard lead electrocardiogram data by the electrocardiogram data generating unit 130 by reflecting individual characteristic information of the examinee whose electrocardiogram is measured by the single lead electrocardiogram measuring unit 110.

[0070] Here, the individual characteristic information may include demographic information such as the subject's gender and age, basic health information such as weight, height, and obesity level, disease-related information such as past history, drug history, and family history, and test information on vital signs and vital signs such as blood tests, genetic tests, blood pressure, and oxygen saturation.

[0071]

[0072] On the other hand, the prediction unit 120 can input not only a graphed electrocardiogram but also a digitized electrocardiogram, and further includes, for example, a data conversion module 122 that converts the asynchronous electrocardiogram data measured by the single-lead electrocardiogram measurement unit 110 into numerical data corresponding to an electrocardiogram using a specific formula, and the diagnostic algorithm 121 can also predict the health condition using numerical data corresponding to the asynchronous electrocardiogram data as input.

[0073]

[0074] In addition, the diagnostic algorithm 121 can be constructed by deep learning models of various methods such as convolutional neural networks, LSTM, RNN, and MLP, and by various machine learning models such as logistic regression, principle-based models, random forests, and support vector machines.

[0075] For example, the diagnostic algorithm 121 may be constructed by a deep learning model using two or more electrocardiogram data measured at different times with a time lag. Here, the two or more electrocardiograms may be integrated into one electrocardiogram data and input, or the two or more electrocardiograms may be input into the deep learning model, and then intermediately measured semantic features, i.e., spatial and time series features, may be extracted. Then, the two electrocardiograms may be compared based on the extracted semantic features, thereby enabling measurement, diagnosis, examination, and prediction of the health state at the time of measurement by the single-lead electrocardiogram measurement unit 110 or the future health state.

[0076] Here, a method for combining electrocardiograms measured at two or more time points can be to divide a single electrocardiogram into bits and then input pairs of bits of the same lead measured at different time points, or to input the bits and then compare the extracted semantic features or result values ​​with each other.

[0077] Alternatively, electrocardiograms measured at two or more time points may be integrated without being divided into bits. Here, electrocardiogram data from two or more time points may be fused as is and input, or may be separated by lead and fused before being input. Electrocardiograms measured at different time points may be input to the deep learning layer of a deep learning model, and semantic features or output values ​​may be extracted and fused to output a final conclusion. Alternatively, electrocardiograms measured at different time points may be separated by lead and input to the deep learning layer, and the features or output values ​​extracted by deep learning may be fused in a later stage to output a final conclusion.

[0078] Here, when electrocardiograms from two or more points in time are mixed and used, they can be input in an asynchronous state, synchronized on a bit-by-bit basis, or synchronized based on deep learning to fuse the electrocardiograms for use.

[0079] That is, a deep learning model can be constructed using standard lead ECG data. Here, a deep learning model can be developed that inputs one ECG at each time point to measure, diagnose, examine, and predict health conditions, and then the deep learning model can be used to input ECGs measured at two or more time points into the deep learning model, and the semantic features or final output values ​​output from the deep learning model can be combined to predict results.

[0080] As mentioned above, various deep learning methods such as convolution neural networks, LSTM, RNN, and MLP can be used to synthesize the final output value, and various machine learning methods such as logistic regression, principled models, random forests, and support vector machines can also be used.

[0081]

[0082] The prediction unit 120 as described above can diagnose and predict diseases of the circulatory system, endocrine, nutritional and metabolic diseases, neoplastic diseases, mental and behavioral disorders, nervous system diseases, eye and adnexal diseases, ear and mastoid diseases, respiratory system diseases, digestive system diseases, skin and skin tissue diseases, musculoskeletal and connective tissue diseases, genitourinary system diseases, pregnancy, childbirth and postpartum diseases, congenital malformations, deformities and chromosomal abnormalities.

[0083] In addition, the prediction unit 120 can identify injuries due to physical trauma, determine prognosis, measure pain, predict the risk of death or worsening due to trauma, capture or predict complications, and identify specific pathologies that appear before and after birth.

[0084] In addition, the prediction unit 120 can measure, diagnose, examine and predict the health condition of the examinee, which can lead to services in the healthcare field such as aging, sleep, weight, blood pressure, blood sugar, oxygen saturation, metabolism, stress, tension, fear, drinking, smoking, problematic behaviors, lung capacity, amount of exercise, pain management, obesity, body mass, body composition, menus, types of exercise, lifestyle pattern recommendations, emergency situation management, late-onset disease management, drug prescriptions, test recommendations, medical checkup recommendations, nursing care, remote health management, remote medical treatment, vaccination and post-vaccination management.

[0085]

[0086] Meanwhile, Fig. 2 shows a flowchart of a prediction method by the health condition prediction system using asynchronous electrocardiograms of Fig. 1. The method will be described in detail below with reference to this flowchart.

[0087] First, a single-lead electrocardiogram measurement unit 110 equipped with two electrodes measures electrocardiograms of two or more electrical axes with a time lag by contacting two points on the subject's body with two electrodes, and obtains asynchronous electrocardiogram data, which is then transmitted to a prediction unit 120 via a short-range network (S110).

[0088] Thereafter, the prediction unit 120 segments the asynchronous electrocardiogram data of the single-lead electrocardiogram measurement unit 110 using a pre-trained diagnostic algorithm 121 to generate segmented asynchronous electrocardiogram data (S121), and predicts the health condition from the segmented asynchronous electrocardiogram data (S122).The health condition is grasped by predicting the presence or absence of disease and the severity of disease from the segmented asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit 110 using the pre-constructed diagnostic algorithm 121 that learns the standard multi-electrocardiogram data set that matches the presence or absence of disease and the severity of disease corresponding to the segmented standard lead electrocardiogram and the standard multi-electrocardiogram that are measured and segmented on the same electrical axis as the electrocardiogram measured by the single-lead electrocardiogram measurement unit 110.

[0089] Meanwhile, the electrocardiogram data generating unit 130 grasps the individual characteristics of the plurality of asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit 110, identifies them as specific lead electrocardiogram data, and generates the remaining plurality of standard lead electrocardiogram data that do not correspond to the identified specific lead electrocardiogram data (S130). The diagnostic algorithm 121 receives the asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit 110 and the standard lead electrocardiogram data generated by the electrocardiogram data generating unit 130 as input, or receives the standard lead electrocardiogram data generated by the electrocardiogram data generating unit 130 as input, and can output and predict the presence or absence of a disease and the severity of the disease.

[0090] Then, the warning unit 140 monitors the health condition by measuring a reference electrocardiogram of the examinee's usual healthy state using the single-lead electrocardiogram measurement unit 110 and the prediction unit 120, and then compares the electrocardiogram measured and input in real time by the single-lead electrocardiogram measurement unit 110 during daily life with the reference electrocardiogram to predict whether the electrocardiogram is an electrocardiogram measured without error or whether there is any abnormality in the examinee's health condition.If an error or abnormality is predicted, warning information is generated and the warning information can be transmitted along with a beep via the single-lead electrocardiogram measurement unit 110 in the form of a smartwatch or a separate smart device.

[0091]

[0092] Therefore, by configuring the health condition prediction system using asynchronous electrocardiograms as described above, asynchronous electrocardiograms of two or more leads can be easily measured using a single-lead electrocardiogram device or a six-lead electrocardiogram device, various types of synchronized electrocardiograms accumulated in medical institutions can be segmented to extract asynchronous electrocardiograms of two or more leads to generate a dataset for learning a prediction model, and diseases can be predicted from asynchronous electrocardiogram data of two or more leads measured using a single-lead electrocardiogram device or a six-lead electrocardiogram device.By generating multi-channel electrocardiogram data from asynchronous electrocardiogram data of two or more leads, the presence or absence of disease and the severity of disease can be output and predicted with higher accuracy.

[0093]

[0094] The embodiments described in this specification and the configurations shown in the drawings are merely the best embodiment of the present invention and do not fully represent the technical ideas of the present invention, so it should be understood that there may be various equivalents and modifications that can replace them at the time of this application. [Explanation of symbols]

[0095] 110: Single lead electrocardiogram measurement unit 111: Wearable electrocardiogram patch 112: Smartwatch 120: Prediction section 121: Diagnostic Algorithm 122: Data conversion module 130: Electrocardiogram data generation unit 140: Warning section

Claims

1. A health condition prediction system using asynchronous electrocardiograms, comprising: a single-lead electrocardiogram measurement unit having two electrodes and measuring electrocardiograms of two or more electrical axes with a time difference to obtain asynchronous electrocardiogram data; and a prediction unit that predicts the presence or absence of disease and the degree of disease from the asynchronous electrocardiogram data input from the single-lead electrocardiogram measurement unit and segmented in specific time units using a diagnostic algorithm that is pre-constructed by learning from asynchronous segmented standard lead electrocardiograms at different times in asynchronous standard lead electrocardiograms segmented in specific time units from synchronous standard lead electrocardiograms measured and stored on a server of a medical institution, and a plurality of asynchronous segmented standard electrocardiogram data sets that match the presence or absence of disease and the degree of disease corresponding to the asynchronous segmented standard lead electrocardiograms.

2. The method further includes an electrocardiogram data generating unit that identifies a plurality of asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit as specific lead electrocardiogram data by grasping characteristics of the asynchronous electrocardiogram data measured by the single-lead electrocardiogram measuring unit, and generates a plurality of remaining standard lead electrocardiogram data that do not correspond to the identified specific lead electrocardiogram data, 2. The health condition prediction system using an asynchronous electrocardiogram according to claim 1, wherein the diagnostic algorithm takes the measured asynchronous electrocardiogram data and the generated standard lead electrocardiogram data as input, or takes the generated standard lead electrocardiogram data as input, and outputs and predicts the presence or absence of a disease and the severity of the disease.

3. 3. The health condition prediction system using an asynchronous electrocardiogram according to claim 2, wherein the electrocardiogram data generating unit generates a plurality of synchronized standard lead electrocardiogram data.

4. 2. The health condition prediction system using asynchronous electrocardiograms according to claim 1, wherein the diagnostic algorithm generates the plurality of asynchronous segmented standard electrocardiogram data sets by asynchronously extracting synchronized segmented standard lead electrocardiograms obtained by segmenting standard lead electrocardiogram data stored in a synchronized form in specific time units.

5. 2. The health condition prediction system using asynchronous electrocardiograms according to claim 1, wherein the diagnostic algorithm is configured to input standard lead electrocardiogram data obtained by removing time series information from synchronously measured standard lead electrocardiogram data, and to output and predict the presence or absence and severity of a disease corresponding to the standard lead electrocardiogram data.

6. The diagnostic algorithm learns by reflecting individual characteristic information of the examinee, The health condition prediction system using an asynchronous electrocardiogram according to claim 1, characterized in that the prediction unit reflects the individual characteristic information and predicts the presence or absence of a disease and the severity of the disease from the asynchronous electrocardiogram data obtained by the single-lead electrocardiogram measurement unit.

7. The health condition prediction system using asynchronous electrocardiograms as described in claim 6, characterized in that the individual characteristic information includes demographic information such as gender and age, basic health information such as weight, height and obesity level, disease-related information such as past history, drug history and family history, and test information of vital signs and vital signs such as blood tests, genetic tests, blood pressure and oxygen saturation.

Images