Method, device, electronic device and medium for establishing SCD risk prediction model

Abstract

The present disclosure relates to the field of medical big data technology, and specifically to a method, device, electronic device and medium for establishing a sudden cardiac death (SCD) risk prediction model based on time-dependent electrocardiogram parameters. The method comprises: obtaining electrocardiogram data of a training data set, wherein the training data set comprises electrocardiogram data of a high-risk SCD group and a healthy control group; using the same segmentation method, segmenting each electrocardiogram data into a plurality of electrocardiogram data segments, each electrocardiogram data segment corresponds to a corresponding time period, and segmenting each electrocardiogram data segment into a plurality of sub-segments; for each time period, using the average electrocardiogram parameter group of the electrocardiogram data segment corresponding to the time period to train the SCD risk prediction model of the time period. The present disclosure provides a low-cost, easy-to-operate and relatively reliable SCD risk prediction method, which achieves the technical effect of fully stratifying SCD risk patients based on electrocardiogram parameters.

Description

Technical Field

[0001] The present disclosure relates to the field of medical big data technology, and in particular to a method, device, electronic device and medium for establishing a SCD risk prediction model based on time-dependent electrocardiogram parameters. Background Art

[0002] Sudden cardiac death (SCD) is caused by irregular electrical disturbances in the heart, which causes the heart to quiver and stop pumping. When the heart stops pumping, it blocks blood flow to the body's organs, leading to a sudden loss of heart function, breathing, and consciousness. In this case, the patient will die quickly.

[0003] Widely available and low-cost electrocardiogram (ECG) is a potential non-invasive tool for SCD risk stratification. However, currently, although the combination of known ECG parameters such as J wave amplitude, QRS prolongation, QT prolongation, T wave alternans, and QRS grade can improve SCD risk prediction, existing technologies are still unable to fully stratify patients at risk of SCD based on ECG parameters. Summary of the invention

[0004] In order to solve the problems in the related art, the embodiments of the present disclosure provide a method, device, electronic device and medium for establishing a SCD risk prediction model based on time-dependent ECG parameters, which can clinically identify high-risk SCD patients from healthy people.

[0005] In a first aspect, the present disclosure provides a method for establishing a SCD risk prediction model based on time-dependent ECG parameters, comprising: obtaining a training data set, the training data set comprising ECG data of a high-risk SCD group and a healthy control group, the high-risk SCD group and the healthy control group both comprising a plurality of subjects;

[0006] In the same segmentation manner, each electrocardiogram data is segmented into a plurality of electrocardiogram data segments, each electrocardiogram data segment corresponds to a corresponding time period, and each electrocardiogram data segment is segmented into a plurality of sub-segments;

[0007] Extracting an ECG parameter group from each sub-segment, wherein the ECG parameter group includes a plurality of specified ECG parameters;

[0008] For each electrocardiogram data segment, calculating an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter groups of the sub-segments of the electrocardiogram data segment;

[0009] For each time period, the average ECG parameter group of the ECG data segment corresponding to the time period is used to train the SCD risk prediction model of the time period, and the SCD risk prediction model is used to predict the SCD risk based on the ECG data of the subjects obtained in the time period.

[0010] In a second aspect, an embodiment of the present disclosure provides a device for establishing a SCD risk prediction model based on time-dependent ECG parameters, comprising:

[0011] A first acquisition module is configured to acquire a training data set, wherein the training data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects;

[0012] A first segmentation module is configured to segment each electrocardiogram data into a plurality of electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponds to a corresponding time period, and segment each electrocardiogram data segment into a plurality of sub-segments;

[0013] A first extraction module is configured to extract an ECG parameter group from each sub-segment, wherein the ECG parameter group includes a plurality of specified ECG parameters;

[0014] A first calculation module is configured to calculate, for each electrocardiogram data segment, an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter groups of sub-segments of the electrocardiogram data segment;

[0015] The training module is configured to train, for each time period, an SCD risk prediction model for the time period using an average electrocardiogram parameter group of the electrocardiogram data segment corresponding to the time period, wherein the SCD risk prediction model is used to perform SCD risk prediction based on the electrocardiogram data of the subject obtained in the time period.

[0016] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer instructions, and wherein the one or more computer instructions are executed by the processor to implement a method as described in any one of the first aspects.

[0017] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium on which computer instructions are stored. When the computer instructions are executed by a processor, the method described in the first aspect is implemented.

[0018] In a fifth aspect, a computer program product is provided in an embodiment of the present disclosure, comprising computer instructions, which, when executed by a processor, implement the method described in the first aspect.

[0019] According to the technical solution provided by the embodiment of the present disclosure, a training data set is obtained, the training data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, and the high-risk SCD group and the healthy control group each include multiple subjects; each electrocardiogram data is segmented into multiple electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponds to a corresponding time period, and each electrocardiogram data segment is segmented into multiple sub-segments; an electrocardiogram parameter group is extracted from each sub-segment, and the electrocardiogram parameter group includes multiple specified electrocardiogram parameters; for each electrocardiogram data segment, the average electrocardiogram parameter group of the electrocardiogram data segment is calculated according to the electrocardiogram parameter groups of the sub-segments of the electrocardiogram data segment; for each time period, the average electrocardiogram parameter group of the electrocardiogram data segment corresponding to the time period is used to train an SCD risk prediction model for the time period, and the SCD risk prediction model is used to perform SCD risk prediction based on the electrocardiogram data of the subjects obtained in the time period. The present invention can accurately identify high-risk SCD patients based on widely available and low-cost electrocardiogram data. The method takes into account the differences in electrocardiogram data obtained in different time periods when identifying high-risk SCD patients and healthy people, and trains models for different time periods respectively for SCD risk prediction based on electrocardiogram data obtained in different time periods. It provides a low-cost, easy-to-operate and relatively reliable SCD risk prediction method, and achieves the technical effect of fully stratifying SCD risk patients based on electrocardiogram parameters.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Other features, objectives and advantages of the present disclosure will become more apparent through the following detailed description of non-limiting embodiments in conjunction with the accompanying drawings. In the accompanying drawings:

[0022] Figure 1 A flowchart showing a method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0023] Figure 2 A schematic diagram of ECG parameters showing the maximum value of the R peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0024] Figure 3 A schematic diagram of ECG parameters showing the minimum value of the R peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0025] Figure 4A schematic diagram of ECG parameters showing the maximum value of the S peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0026] Figure 5 A schematic diagram of ECG parameters showing the minimum value of the S peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0027] Figure 6 A schematic diagram of ECG parameters showing the maximum value of the T peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0028] Figure 7 A schematic diagram of ECG parameters showing the minimum value of the T peak within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0029] Figure 8 A schematic diagram of ECG parameters showing an average value of R peak value within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0030] Fig. 9 A schematic diagram of ECG parameters showing an average value of an S peak value within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0031] Fig.10 A schematic diagram of ECG parameters showing an average value of T peak values ​​within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0032] Fig.11 A schematic diagram of ECG parameters showing the difference between the R peak and the S peak in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0033] Fig.12 A schematic diagram of ECG parameters showing the maximum area of ​​an R peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0034] Fig.13 A schematic diagram of ECG parameters showing the minimum area of ​​an R peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0035] Fig.14A schematic diagram of ECG parameters showing the maximum area of ​​the S peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0036] Fig.15 A schematic diagram of ECG parameters showing the minimum area of ​​an S peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0037] Fig.16 A schematic diagram of ECG parameters showing the maximum area of ​​a T peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0038] Fig.17 A schematic diagram of ECG parameters showing the minimum area of ​​a T peak within a sub-segment in a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0039] Fig.18 A schematic diagram of ECG parameters showing the average area of ​​the R peak within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0040] Fig.19 A schematic diagram of ECG parameters showing the average area of ​​the S peak within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0041] Fig. 20 A schematic diagram of ECG parameters showing the average area of ​​T peaks within sub-segments of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0042] Fig.21 A schematic diagram of ECG parameters for a time interval from the start of the S wave to the end of the T wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown;

[0043] Fig. 22 A schematic diagram of ECG parameters of a time interval from the start of the R wave to the end of the T wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown;

[0044] Fig.23 A schematic diagram of ECG parameters of a time interval from the beginning of the R wave to the end of the S wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown;

[0045] Fig.24 A schematic diagram of ECG parameters of a time interval from the beginning of a T wave to the end of a T wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown;

[0046] Fig.25 A schematic diagram of an electrocardiographic parameter showing an average value of the difference between the S peak value and the T peak value in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent electrocardiographic parameters according to an embodiment of the present disclosure;

[0047] Fig.26 A schematic diagram of ECG parameters showing the total area of ​​R peaks within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0048] Fig. 27 A schematic diagram of ECG parameters showing the total area of ​​S peaks within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0049] Fig.28 A schematic diagram of ECG parameters showing the total area of ​​T peaks within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0050] Fig.29 A schematic diagram of ECG parameters showing the total time interval between the start of an R wave and the end of an R wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0051] Fig.30 A schematic diagram of ECG parameters showing the total time interval between the start of an S wave and the end of an S wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0052] Fig.31 A schematic diagram of ECG parameters showing the total time interval between the start of a T wave and the end of a T wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0053] Fig.32 A schematic diagram of ECG parameters showing a standard deviation of T peak values ​​within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0054] Fig.33 A schematic diagram of an ECG parameter showing an average value of a difference between an R peak value and an S peak value within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0055] Fig.34 A schematic diagram of ECG parameters showing the average time interval between the start of an R wave and the end of a T wave in a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0056] Fig.35 A schematic diagram of ECG parameters showing the standard deviation of the time interval between the start of the R wave and the end of the T wave within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0057] Fig.36 A schematic diagram of ECG parameters showing the standard deviation of the time interval between the start and end of a T wave within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0058] Fig.37 A schematic diagram of ECG parameters showing the standard deviation of T peak areas within a sub-segment of a method for establishing an SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure;

[0059] Fig.38 A structural block diagram of a device for establishing a SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown;

[0060] Fig.39 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown;

[0061] Fig.40 A schematic diagram showing the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0062] Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement them. In addition, for the sake of clarity, parts not related to the description of the exemplary embodiments are omitted in the accompanying drawings.

[0063] In the present disclosure, it should be understood that terms such as "include" or "have" are intended to indicate the presence of features, numbers, steps, behaviors, components, parts, or a combination thereof disclosed in the present specification, and are not intended to exclude the possibility that one or more other features, numbers, steps, behaviors, components, parts, or a combination thereof exist or are added.

[0064] It should also be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other. The present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0065] In the present disclosure, if it involves operations of obtaining user information or user data or displaying user information or user data to others, the operations are all authorized and confirmed by the user, or actively selected by the user.

[0066] Widely available and low-cost ECG is a potential non-invasive tool for SCD risk stratification. However, currently, although the combination of known ECG parameters such as J wave amplitude, QRS prolongation, QT prolongation, T wave alternans, and QRS grade can improve SCD risk prediction, existing technologies are still unable to fully stratify patients at risk of SCD based on ECG parameters. The present disclosure provides a method for establishing a sudden cardiac death (SCD) risk prediction model based on time-dependent electrocardiogram parameters, the method comprising: obtaining a training data set, the training data set comprising electrocardiogram data of a high-risk SCD group and a healthy control group, the high-risk SCD group and the healthy control group both comprising a plurality of subjects; segmenting each electrocardiogram data into a plurality of electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponding to a corresponding time period, and segmenting each electrocardiogram data segment into a plurality of sub-segments; extracting an electrocardiogram parameter group from each sub-segment, the electrocardiogram parameter group comprising a plurality of specified electrocardiogram parameters; for each electrocardiogram data segment, calculating an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter groups of the sub-segments of the electrocardiogram data segment; for each time period, using the average electrocardiogram parameter group of the electrocardiogram data segment corresponding to the time period to train a SCD risk prediction model for the time period, the SCD risk prediction model being used to perform SCD risk prediction based on the electrocardiogram data of the subjects acquired in the time period. According to the embodiments of the present disclosure, high-risk SCD patients can be accurately identified based on widely available and low-cost electrocardiogram data. The method takes into account the differences in electrocardiogram data acquired in different time periods when identifying high-risk SCD patients and healthy people, and trains models for different time periods respectively for SCD risk prediction based on electrocardiogram data acquired in different time periods, thereby providing a low-cost, easy-to-operate and relatively reliable SCD risk prediction method.

[0067] Figure 1 A flow chart showing a method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown. Figure 1 As shown, the method for establishing a SCD risk prediction model based on time-dependent ECG parameters includes steps S101 to S105.

[0068] In step S101, a training data set is obtained, the training data set including electrocardiogram data of a high-risk SCD group and a healthy control group, the high-risk SCD group and the healthy control group each including a plurality of subjects. According to an embodiment of the present disclosure, the electrocardiogram data is 24-hour single-lead electrocardiogram data.

[0069] According to an embodiment of the present disclosure, multiple subjects are recruited to divide SCD survivors or patients with hemodynamic disorders caused by sustained ventricular tachycardia / ventricular fibrillation into a high-risk SCD group (SCDHR group); healthy controls without heart disease are divided into a healthy control group (HC group).

[0070] According to an embodiment of the present disclosure, subjects were screened, and 24-hour single-lead electrocardiogram data of 49 high-risk SCD patients and 125 healthy controls were collected, among which 30 high-risk SCD patients and 80 healthy controls were divided into a training data set.

[0071] According to an embodiment of the present disclosure, when screening subjects, subjects with the following conditions are excluded: (1) the degree of atrioventricular block at resting heart rate is ≥ II degree; (2) the NYHA IV standard is met; (3) there is a history of myocardial infarction within 1 month before the start of the test period; (4) treatment with coronary artery revascularization within 3 months before the start of the test period; (5) suffering from advanced cerebrovascular or renal disease and any non-cardiac disease with a higher probability of death during the test period; (6) life expectancy is less than 1 year old; (7) there is a history of valvular heart disease.

[0072] In step S102, each electrocardiogram data is segmented into a plurality of electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponds to a corresponding time period, and each electrocardiogram data segment is segmented into a plurality of sub-segments.

[0073] The inventors noticed that the ECG data of patients at high risk of SCD show different characteristics in different time periods. Therefore, according to an embodiment of the present disclosure, the ECG data of each subject in the training data set is segmented into multiple ECG data segments in the same segmentation method, and the SCD risk prediction model of the corresponding time period is trained for each ECG data segment, so as to fully consider the characteristic differences of the ECG data obtained in different time periods in SCD risk prediction.

[0074] According to an embodiment of the present disclosure, the electrocardiogram data of each subject in the training data set is segmented into multiple electrocardiogram data segments in the same segmentation manner, including segmenting the electrocardiogram data of each subject at the same time point, and the time lengths between the time points can be the same or different. For example, assuming that the total length of the electrocardiogram data of each subject is T hours, it can be evenly segmented into N electrocardiogram data segments, and the length of each electrocardiogram data segment is T / N hours.

[0075] According to an embodiment of the present disclosure, the electrocardiogram data is 24-hour single-lead electrocardiogram data, that is, T is 24 hours, but is not limited thereto. The length of the time period corresponding to each electrocardiogram data segment is a specified time length between 0.5 hours and 2 hours, for example, it can be 1 hour. For example, the 24-hour electrocardiogram data of each subject can be divided into 24 electrocardiogram data segments, and the length of each electrocardiogram data segment is 1 hour.

[0076] According to an embodiment of the present disclosure, in order to extract ECG parameters more accurately, each ECG data segment is further divided into multiple sub-segments, and the duration of the sub-segments can be set as needed, for example, 1 minute, but not limited thereto. In a specific embodiment, each 1-hour ECG data segment is further divided into 60 1-minute sub-segments.

[0077] According to an embodiment of the present disclosure, in order to remove noise in ECG parameters, data filtering is performed on sub-segments. When data filtering is performed on sub-segments, sub-segments that meet the following criteria are discarded: the data length of the sub-segment is less than half of the time length of the sub-segment.

[0078] According to an embodiment of the present disclosure, in order to extract ECG parameters more accurately, the multiple sub-segments of each ECG data segment are further divided into more sub-segments, and the duration of the sub-segments can be set as needed, for example, 1 second, but not limited thereto. In a specific embodiment, each 1-minute ECG data segment is further divided into 1-second sub-segments.

[0079] According to an embodiment of the present disclosure, in order to remove noise in ECG parameters, data filtering is performed on the segments. When data filtering is performed on the segments, the segments that meet the following filtering criteria are discarded: (1) the maximum value in the segment is lower than 0.2 mV; (2) the maximum value in the segment is higher than 5 mV; (3) the difference between the maximum value and the minimum value in the segment is greater than twice the median difference in the time length of its sub-segments.

[0080] In step S103, an ECG parameter group is extracted from each sub-segment, where the ECG parameter group includes a plurality of designated ECG parameters.

[0081] According to an embodiment of the present disclosure, a plurality of candidate ECG parameters are extracted from each sub-segment. In a specific embodiment, a plurality of candidate ECG parameters are extracted from each 1-minute sub-segment.

[0082] According to an embodiment of the present disclosure, the candidate ECG parameters include: peak-related ECG parameters, area-related ECG parameters and time-related ECG parameters, but are not limited thereto. In a specific embodiment, the plurality of candidate ECG parameters is 37, including: maxR, minR, maxS, minS, maxT, minT, mean_R, mean_S, mean_T, RS, inteRM, inteRm, inteSM, inteSm, inteTM, inteTm, inteR_mean, inteS_mean, inteT_mean, t_ST, t_RT, t_RS, t_T, len_ST, inteR_sum, inteS_sum, inteT_sum, t_R_sum, t_S_sum, t_T_sum, T_sd, mean_RS, t_QT, QT_sd, t_T_sd, inteT_sd and t_plus;

[0083] in:

[0084] maxR is the maximum value of the R peak value in the corresponding sub-segment, see Figure 2 ;

[0085] minR is the minimum value of the R peak in the corresponding sub-segment, see Figure 3 ;

[0086] maxS is the maximum value of the S peak value in the corresponding sub-segment, see Figure 4 ;

[0087] minS is the minimum value of the S peak value in the corresponding sub-segment, see Figure 5 ;

[0088] maxT is the maximum value of the T peak value in the corresponding sub-segment, see Figure 6 ;

[0089] minT is the minimum value of the T peak in the corresponding subsegment, see Figure 7 ;

[0090] mean_R is the average value of the R peak value in the corresponding sub-segment, see Figure 8 ;

[0091] mean_S is the average value of the S peak value in the corresponding sub-segment, see Fig. 9 ;

[0092] mean_T is the average value of T peaks in the corresponding sub-segment, see Fig.10 ;

[0093] RS is the difference between the peak R value and the peak S value in the corresponding sub-segment, see Fig.11 ;

[0094] len_ST is the average of the differences between the S peak and the T peak within the corresponding subsegment, see Fig.12 ;

[0095] T_sd is the standard deviation of the T peak within the corresponding sub-fragment, see Fig.13 ;

[0096] mean_RS is the average of the differences between the R peak and the S peak in the corresponding sub-segment, see Fig.14 ;

[0097] inteRM is the maximum area of ​​the R peak within the corresponding subfragment, see Fig.15 ;

[0098] inteSM is the maximum area of ​​the S peak in the corresponding subfragment, see Fig.16 ;

[0099] inteTM is the maximum area of ​​the T peak within the corresponding subfragment, see Fig.17 ;

[0100] inteRm is the minimum area of ​​the R peak in the corresponding subfragment, see Fig.18 ;

[0101] inteSm is the minimum area of ​​the S peak in the corresponding subfragment, see Fig.19 ;

[0102] inteTm is the minimum area of ​​the T peak within the corresponding subfragment, see Fig. 20 ;

[0103] inteR_mean is the average area of ​​the R peak in the corresponding subfragment, see Fig.21 ;

[0104] inteS_mean is the average area of ​​the S peak in the corresponding sub-fragment, see Fig. 22 ;

[0105] inteT_mean is the average area of ​​the T peak in the corresponding subfragment, see Fig.23 ;

[0106] inteR_sum is the total area of ​​the R peaks within the corresponding subfragment, see Fig.24 ;

[0107] inteS_sum The total area of ​​the S peak in the corresponding sub-fragment, see Fig.25 ;

[0108] inteT_sum The total area of ​​the T peaks in the corresponding subfragment, see Fig.26 ;

[0109] inteT_sd is the standard deviation of the T peak area within the corresponding subfragment, see Fig. 27 ;

[0110] t_ST is the time interval from the beginning of the S wave to the end of the T wave in the corresponding subsegment, see Fig.28 ;

[0111] t_RT is the time interval from the beginning of the R wave to the end of the T wave in the corresponding subsegment, see Fig.29 ;

[0112] t_RS is the time interval from the beginning of the R wave to the end of the S wave in the corresponding subsegment, see Fig.30 ;

[0113] t_T is the time interval from the beginning of the T wave to the end of the T wave in the corresponding sub-segment, see Fig.31 ;

[0114] t_R_sum is the total time interval between the start of the R wave and the end of the R wave within the corresponding subsegment, see Fig.32 ;

[0115] t_S_sumS is the total time interval between the start of the S wave and the end of the S wave in the corresponding subsegment, see Fig.33 ;

[0116] t_T_sum is the total time interval between the start and end of the T wave in the corresponding subsegment, see Fig.34 ;

[0117] t_QT is the average time interval between the beginning of the R wave and the end of the T wave within the corresponding subsegment, see Fig.35 ;

[0118] QT_sd is the standard deviation of the time interval between the start of the R wave and the end of the T wave within the corresponding subsegment, see Fig.36 ;

[0119] t_T_sd is the standard deviation of the time interval between the start and end of the T wave in the corresponding subsegment, see Fig.37 ;

[0120] t_plus is the ratio of the positive T wave within the corresponding sub-segment.

[0121] According to an embodiment of the present disclosure, the candidate ECG parameters whose differences between the high-risk SCD group subjects and the healthy control group subjects meet preset conditions are selected as the designated ECG parameters; the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, inteSsum, mean_RS, t_T_sd and t_plus; or the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, mean_RS, t_T_sd and t_plus, but are not limited to this.

[0122] The embodiments of the present disclosure are not simply based on the commonly used ECG parameters in the art (such as J wave amplitude, QRS prolongation, QT prolongation, T wave alternation and QRS grade). When predicting SCD during the PR interval, it lacks the extraction and need to use relevant data such as S peak value and T peak value and the location of the peak area to make a judgment. Instead, it calculates the area and peak value of the S wave, T wave and R wave, and selects the candidate ECG parameters from the candidate ECG parameters whose differences between the high-risk SCD group subjects and the healthy control group subjects meet the preset conditions as the designated ECG parameters for SCD prediction. It can effectively identify the ECG parameters that are more suitable for SCD early warning, and improve the accuracy of the SCD risk prediction model. According to the embodiments of the present disclosure, the candidate ECG parameters are not limited to the above 37 ECG parameters, and the designated ECG parameters are not limited to the above 8 ECG parameters. Those skilled in the art can select more, fewer, or different ECG parameters according to the principles of the present disclosure for implementing the scheme of the present disclosure.

[0123] According to an embodiment of the present disclosure, the preset condition is: using a statistical test method, such as a T test, but not limited to this, to check the differences in candidate ECG parameters between subjects in the high-risk SCD group and subjects in the healthy control group, and the obtained P value is less than the set value.

[0124] According to an embodiment of the present disclosure, the set value is a value obtained by dividing 0.05 by the number of candidate ECG parameters. In a specific embodiment, when the number of candidate ECG parameters is 37, the set value is a value obtained by dividing 0.05 by 37.

[0125] Specifically, for each candidate ECG parameter of each data segment, the average value of the candidate ECG parameter of each sub-segment in the data segment is obtained by averaging the values ​​of the candidate ECG parameter of each sub-segment in the data segment. For example, when each data segment is 1 hour and the sub-segment is 1 minute, for the candidate ECG parameter X in a single data segment, the average value of the candidate ECG parameter X in this 1-hour data segment is obtained by averaging 60 values ​​of the candidate ECG parameter X for 60 minutes in 1 hour. The difference in the average values ​​of the candidate ECG parameters between the SCDHR and HC samples was compared by the Student's T test. Within 17 hours, the integrated means of SCDHR and HC were significantly different. Within more than 5 hours, there were 8 candidate ECG parameter averages that were significantly different between SCDHR and HC, and the corresponding candidate ECG parameters included inteSm, inteSmean, mean_RS, t_RT, t_T, inteSsum, t_T_sd and t_plus. Among them, three ECG features are related to the S wave area and three are related to the T wave interval. Therefore, inteSm, inteSmean, t_RT, t_T, inteSsum, mean_RS, t_T_sd and t_plus are taken as designated ECG parameters, and a set of designated ECG parameters is taken as an ECG parameter group.

[0126] In step S104, for each ECG data segment, the average ECG parameter group of the ECG data segment is calculated based on the ECG parameter group of the sub-segment of the ECG data segment. In a specific embodiment, the average ECG parameter of the 1-hour ECG data segment is calculated based on the ECG parameter group of the 1-minute sub-segment of the 1-hour ECG data segment.

[0127] According to an embodiment of the present disclosure, the same designated ECG parameters in the ECG parameter group of the sub-segment of the ECG data segment are averaged to obtain the average ECG parameter group of the ECG data segment. In a specific embodiment, for each 1-hour ECG data segment, the same 8 designated ECG parameters in the ECG parameter group of its 1-minute ECG data segment are averaged, and the 8 designated ECG parameters of the one-minute scale are combined into 8 designated ECG parameters of the one-hour scale to obtain the average ECG parameter group of each 1-hour ECG data segment.

[0128] In step S105, for each time period, an average ECG parameter group of the ECG data segment corresponding to the time period is used to train an SCD risk prediction model for the time period, and the SCD risk prediction model is used to predict SCD risk based on the ECG data of the subjects obtained in the time period.

[0129] According to an embodiment of the present disclosure, a regression model is trained using an average electrocardiogram parameter group of the electrocardiogram data segments corresponding to the time period, and the regression model includes LASSO regression, but is not limited thereto.

[0130] According to an embodiment of the present disclosure, the average ECG parameter group of the ECG data segment corresponding to the time period is used to train the SCD risk prediction model for the time period by cross-validation, wherein the cross-validation method adopted may be a three-fold cross-validation method, but is not limited thereto.

[0131] According to an embodiment of the present disclosure, the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period is selected for SCD risk prediction.

[0132] The inventors noticed that the prediction performance of the SCD risk prediction models obtained in different time periods is different. Therefore, selecting the SCD risk prediction model with the best performance in each time period to predict SCD risk can obtain better prediction results.

[0133] According to an embodiment of the present disclosure, the time period corresponding to the SCD risk prediction model with the best performance includes the time point 18:00 and / or the time point 19:00; or the SCD risk prediction model with the best performance is the SCD risk prediction model of 18:00-19:00.

[0134] The inventors found that the characteristics related to the S wave region were most significantly different between the high-risk SCD group subjects and the healthy control group subjects at around 1:00-4:00 and 18:00-19:00, while the characteristics related to the T wave region were most significantly different between the high-risk SCD group subjects and the healthy control group subjects at around 18:00-19:00. Therefore, the prediction effect of the SCD risk prediction model including the time point 18:00 and / or the time point 19:00 is better.

[0135] According to an embodiment of the present disclosure, the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period is selected for SCD risk prediction, including: during the training process, calculating the area under the first ROC curve of each SCD risk prediction model; and determining the performance of the SCD risk prediction model based on the area under the first ROC curve.

[0136] In a specific embodiment, when the SCD risk prediction model is established for each hour, the SCD risk prediction model with the best performance is the SCD risk prediction model of 18:00-19:00, which has better performance than the model in any other time range. Specifically, 24 LASSO regression models were constructed from 1:00 to 24:00. The area under the ROC curve of the model tested in the training set ranged from 0.73 to 0.91. The area under the ROC curve of the SCD risk prediction model of 18:00-19:00 reached a maximum value of 0.91, and the electrocardiogram features used in the model were inteSm, inteSmean, t_RT, t_T, mean_RS, t_T_sd and t_plus. Among them, inteSsum was not included because of its significantly short time period.

[0137] According to an embodiment of the present disclosure, the step of selecting the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period for SCD risk prediction further includes:

[0138] Acquire a test data set, wherein the test data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects;

[0139] In the same segmentation manner as that for the training data set, each electrocardiogram data in the test data set is segmented into a plurality of test electrocardiogram data segments, and each test electrocardiogram data segment is segmented into a plurality of test sub-segments, each test electrocardiogram data segment corresponds to a corresponding time period;

[0140] extracting from each test sub-segment the same set of ECG parameters as in the sub-segment;

[0141] For each test electrocardiogram data segment, calculating an average electrocardiogram parameter group of the test electrocardiogram data segment according to the electrocardiogram parameter group of the test sub-segment of the test electrocardiogram data segment;

[0142] For the SCD risk prediction model of each time period, using the average electrocardiogram parameter group of the corresponding test electrocardiogram data segment to calculate the area under the second ROC curve of the SCD risk prediction model;

[0143] The performance of the SCD risk prediction model determined according to the area under the first ROC curve is verified using the area under the second ROC curve.

[0144] In order to verify the accuracy of the data in the training data set, the test data set was used to verify the data in the training data set. According to the embodiments of the present disclosure, the subjects were screened, and 24-hour single-lead ECG data of 49 high-risk SCD patients and 125 healthy controls were collected, among which 19 high-risk SCD patients and 45 healthy controls were divided into the training data set.

[0145] According to an embodiment of the present disclosure, each electrocardiogram data in the test data set is segmented into a plurality of test electrocardiogram data segments in the same segmentation manner as for the training data set, and each test electrocardiogram data segment is segmented into a plurality of test sub-segments, each test electrocardiogram data segment corresponding to a corresponding time period. In a specific embodiment, each electrocardiogram data in the test data set is segmented into 24 1-hour test electrocardiogram data segments, and each 1-hour test electrocardiogram data segment is segmented into 60 1-minute test sub-segments.

[0146] According to an embodiment of the present disclosure, the same ECG parameter group as in the sub-segment is extracted from each test sub-segment. In a specific embodiment, the same ECG parameter group as in the 1-minute sub-segment is extracted from each 1-minute test sub-segment.

[0147] According to an embodiment of the present disclosure, for each test ECG data segment, the average ECG parameter group of the test ECG data segment is calculated based on the ECG parameter group of the test sub-segment of the test ECG data segment. In a specific embodiment, the average ECG parameter group of the 1-hour test ECG data segment is calculated based on the ECG parameter group of the 1-minute test sub-segment of the 1-hour test ECG data segment.

[0148] According to an embodiment of the present disclosure, for the SCD risk prediction model of each time period, the average electrocardiogram parameter group of the corresponding test electrocardiogram data segment is used to calculate the second ROC curve area of ​​the SCD risk prediction model. In a specific embodiment, for the SCD risk prediction model of 1 hour, the average electrocardiogram parameter group of the test electrocardiogram data segment of 1 hour is used to calculate the second ROC curve area of ​​the SCD risk prediction model.

[0149] According to an embodiment of the present disclosure, the performance of the SCD risk prediction model determined according to the area under the first ROC curve is verified using the area under the second ROC curve, including: comparing the values ​​of the area under the first ROC curve and the area under the second ROC curve. In a specific embodiment, in the test data set, from 1:00 to 24:00, the 24 SCD risk prediction models obtained based on the training data set were tested using the average ECG parameters including 8 specified ECG parameters, and it was found that the SCD risk prediction model in the time period of 18:00-19:00 reached the maximum area under the second ROC curve (0.832), which was close to the maximum area under the first ROC curve (0.91). Therefore, the accuracy of the predictive performance of the SCD risk prediction model obtained in the training data set was proved in the test data set.

[0150] Fig.38 The structural block diagram of the device for establishing a SCD risk prediction model based on time-dependent ECG parameters according to an embodiment of the present disclosure is shown. The device can be implemented as part or all of an electronic device through software, hardware, or a combination of both.

[0151] like Fig.38 As shown, an apparatus 500 for establishing a SCD risk prediction model based on time-dependent ECG parameters includes: a first acquisition module 3801 , a first segmentation module 3802 , a first extraction module 3803 , a first calculation module 3804 and a training module 3805 .

[0152] A first acquisition module 3801 is configured to acquire a training data set, wherein the training data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects;

[0153] The first segmentation module 3802 is configured to segment each electrocardiogram data into a plurality of electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponds to a corresponding time period, and segment each electrocardiogram data segment into a plurality of sub-segments;

[0154] A first extraction module 3803 is configured to extract an ECG parameter group from each sub-segment, wherein the ECG parameter group includes a plurality of specified ECG parameters;

[0155] A first calculation module 3804 is configured to calculate, for each electrocardiogram data segment, an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter groups of the sub-segments of the electrocardiogram data segment;

[0156] The training module 3805 is configured to train, for each time period, a SCD risk prediction model for the time period using an average ECG parameter group of the ECG data segments corresponding to the time period, wherein the SCD risk prediction model is used to perform SCD risk prediction based on the ECG data of the subjects obtained in the time period.

[0157] According to an embodiment of the present disclosure, for each ECG data segment, the average ECG parameter group of the ECG data segment is calculated based on the ECG parameter group of the sub-segment of the ECG data segment, including: averaging the same specified ECG parameters in the ECG parameter group of the sub-segment of the ECG data segment to obtain the average ECG parameter group of the ECG data segment.

[0158] According to an embodiment of the present disclosure, the method of using the average ECG parameter group of the ECG data segments corresponding to the time period to train the SCD risk prediction model for the time period includes: training a regression model using the average ECG parameter group of the ECG data segments corresponding to the time period.

[0159] According to an embodiment of the present disclosure, for each time period, the average ECG parameter group of the ECG data segment corresponding to the time period is used to train the SCD risk prediction model of the time period, including: using the average ECG parameter group of the ECG data segment corresponding to the time period to train the SCD risk prediction model of the time period by cross-validation.

[0160] According to an embodiment of the present disclosure, the length of the time period is a specified time length between 0.5 hours and 2 hours.

[0161] According to an embodiment of the present disclosure, the length of the time period is 1 hour; the time length of the sub-segment is 1 minute.

[0162] According to an embodiment of the present disclosure, the electrocardiogram data is 24-hour single-lead electrocardiogram data.

[0163] According to an embodiment of the present disclosure, the device 500 further includes:

[0164] A second extraction module is configured to extract a plurality of candidate ECG parameters from each sub-segment, wherein the candidate ECG parameters include: a peak-related ECG parameter, an area-related ECG parameter, and a time-related ECG parameter;

[0165] The first selection module is configured to select, from the candidate ECG parameters, the candidate ECG parameters whose differences between the high-risk SCD group subjects and the healthy control group subjects meet a preset condition as the designated ECG parameters.

[0166] According to an embodiment of the present disclosure, the peak-related ECG parameters include: maxR, minR, maxS, minS, maxT, minT, mean_R, mean_S, mean_T, RS, len_ST, T_sd and mean_RS; the area-related ECG parameters include: inteRM, inteRm, inteSM, inteSm, inteTM, inteTm, inteR_mean, inteS_mean, inteT_mean, inteR_sum, inteS_sum, inteT_sum and inteT_sd; the time-related ECG parameters include: t_ST, t_RT, t_RS, t_T, t_R_sum, t_S_sum, t_T_sum, t_QT, QT_sd, t_T_sd and t_plus;

[0167] in:

[0168] maxR is the maximum value of the R peak in the corresponding sub-segment;

[0169] minR is the minimum value of the R peak within the corresponding sub-segment;

[0170] maxS is the maximum value of the S peak value in the corresponding sub-segment;

[0171] minS is the minimum value of the S peak in the corresponding sub-segment;

[0172] maxT is the maximum value of the T peak in the corresponding subsegment;

[0173] minT is the minimum value of the T peak within the corresponding subsegment;

[0174] mean_R is the average value of the R peak value in the corresponding sub-segment;

[0175] mean_S is the average value of the S peak value in the corresponding sub-segment;

[0176] mean_T is the average value of T peaks in the corresponding sub-segment;

[0177] RS is the difference between the R peak and the S peak in the corresponding sub-segment;

[0178] len_ST is the average of the differences between the S peak and the T peak within the corresponding subsegment;

[0179] T_sd is the standard deviation of the T peak within the corresponding sub-fragment;

[0180] mean_RS is the average of the differences between the R peak and the S peak within the corresponding sub-segment;

[0181] inteRM is the maximum area of ​​the R peak within the corresponding subfragment;

[0182] inteSM is the maximum area of ​​the S peak within the corresponding subfragment;

[0183] inteTM is the maximum area of ​​the T peak within the corresponding subfragment;

[0184] inteRm is the minimum area of ​​the R peak within the corresponding subfragment;

[0185] inteSm is the minimum area of ​​the S peak within the corresponding subfragment;

[0186] inteTm is the minimum area of ​​the T peak within the corresponding subfragment;

[0187] inteR_mean is the average area of ​​the R peak in the corresponding subfragment;

[0188] inteS_mean is the average area of ​​the S peak in the corresponding subfragment;

[0189] inteT_mean is the average area of ​​the T peak within the corresponding subfragment;

[0190] inteR_sum is the total area of ​​the R peaks within the corresponding subfragment;

[0191] inteS_sum the total area of ​​the S peak in the corresponding sub-fragment;

[0192] inteT_sum the total area of ​​the T peaks in the corresponding subfragment;

[0193] inteT_sd is the standard deviation of the T peak area within the corresponding subfragment;

[0194] t_ST is the time interval from the beginning of the S wave to the end of the T wave in the corresponding subsegment;

[0195] t_RT is the time interval from the beginning of the R wave to the end of the T wave in the corresponding subsegment;

[0196] t_RS is the time interval from the beginning of the R wave to the end of the S wave in the corresponding sub-segment;

[0197] t_T is the time interval from the beginning of the T wave to the end of the T wave in the corresponding sub-segment;

[0198] t_R_sum is the total time interval between the start of the R wave and the end of the R wave within the corresponding subsegment;

[0199] t_S_sumS is the total time interval between the start of the S wave and the end of the S wave in the corresponding subsegment;

[0200] t_T_sum is the total time interval between the start of the T wave and the end of the T wave within the corresponding subsegment;

[0201] t_QT is the average time interval between the beginning of the R wave and the end of the T wave within the corresponding subsegment;

[0202] QT_sd is the standard deviation of the time interval between the start of the R wave and the end of the T wave within the corresponding subsegment;

[0203] t_T_sd is the standard deviation of the time interval between the start and end of the T wave within the corresponding subsegment;

[0204] t_plus is the ratio of the positive T wave within the corresponding sub-segment.

[0205] According to an embodiment of the present disclosure, the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, inteSsum, mean_RS, t_T_sd and t_plus; or the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, mean_RS, t_T_sd and t_plus.

[0206] According to an embodiment of the present disclosure, the preset condition is: using a T test to examine the differences in candidate ECG parameters between subjects in the high-risk SCD group and subjects in the healthy control group, the obtained P value is less than a set value.

[0207] According to an embodiment of the present disclosure, the set value is a value obtained by dividing 0.05 by the number of candidate ECG parameters.

[0208] According to an embodiment of the present disclosure, the apparatus 500 for establishing a SCD risk prediction model based on time-dependent ECG parameters further includes:

[0209] The second selection module is configured to select the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period for SCD risk prediction.

[0210] According to an embodiment of the present disclosure, the time period corresponding to the SCD risk prediction model with the best performance includes the time point 18:00 and / or the time point 19:00; the SCD risk prediction model with the best performance is the SCD risk prediction model of 18:00-19:00.

[0211] According to an embodiment of the present disclosure, the step of selecting the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period for SCD risk prediction includes:

[0212] During the training process, the area under the first ROC curve (AUC) of each SCD risk prediction model was calculated;

[0213] The performance of the SCD risk prediction model is determined based on the area under the first ROC curve.

[0214] According to an embodiment of the present disclosure, the step of selecting the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period for SCD risk prediction further includes:

[0215] Acquire a test data set, wherein the test data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects;

[0216] In the same segmentation manner as that for the training data set, each electrocardiogram data in the test data set is segmented into a plurality of test electrocardiogram data segments, and each test electrocardiogram data segment is segmented into a plurality of test sub-segments, each test electrocardiogram data segment corresponds to a corresponding time period;

[0217] extracting from each test sub-segment the same set of ECG parameters as in the sub-segment;

[0218] For each test electrocardiogram data segment, calculating an average electrocardiogram parameter group of the test electrocardiogram data segment according to the electrocardiogram parameter groups of the test sub-segments of the test electrocardiogram data segment;

[0219] For the SCD risk prediction model of each time period, using the average electrocardiogram parameter group of the corresponding test electrocardiogram data segment to calculate the area under the second ROC curve of the SCD risk prediction model;

[0220] The performance of the SCD risk prediction model determined according to the area under the first ROC curve is verified using the area under the second ROC curve.

[0221] The present disclosure also discloses an electronic device, Fig.39 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown.

[0222] like Fig.39 As shown, the electronic device includes a memory and a processor, wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method according to an embodiment of the present disclosure.

[0223] The disclosed embodiment provides a method for establishing a SCD risk prediction model based on time-dependent ECG parameters, comprising: acquiring a training data set, the training data set comprising ECG data of a high-risk SCD group and a healthy control group, the high-risk SCD group and the healthy control group both comprising a plurality of subjects;

[0224] In the same segmentation manner, each electrocardiogram data is segmented into a plurality of electrocardiogram data segments, each electrocardiogram data segment corresponds to a corresponding time period, and each electrocardiogram data segment is segmented into a plurality of sub-segments;

[0225] Extracting an ECG parameter group from each sub-segment, wherein the ECG parameter group includes a plurality of specified ECG parameters;

[0226] For each electrocardiogram data segment, calculating an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter groups of the sub-segments of the electrocardiogram data segment;

[0227] For each time period, the average ECG parameter group of the ECG data segment corresponding to the time period is used to train the SCD risk prediction model of the time period, and the SCD risk prediction model is used to predict the SCD risk based on the ECG data of the subjects obtained in the time period.

[0228] According to an embodiment of the present disclosure, for each ECG data segment, an average ECG parameter group of the ECG data segment is calculated based on the ECG parameter group of the sub-segment of the ECG data segment, including: averaging the same specified ECG parameters in the ECG parameter group of the sub-segment of the ECG data segment to obtain the average ECG parameter group of the ECG data segment.

[0229] According to an embodiment of the present disclosure, the SCD risk prediction model for the time period is trained using the average ECG parameter group of the ECG data segments corresponding to the time period, including: training a regression model using the average ECG parameter group of the ECG data segments corresponding to the time period.

[0230] According to an embodiment of the present disclosure, for each time period, the average ECG parameter group of the ECG data segment corresponding to the time period is used to train the SCD risk prediction model of the time period, including: using the average ECG parameter group of the ECG data segment corresponding to the time period, and adopting a cross-validation method to train the SCD risk prediction model of the time period.

[0231] According to an embodiment of the present disclosure, the length of the time period is a specified time length between 0.5 hours and 2 hours.

[0232] According to an embodiment of the present disclosure, the length of the time period is 1 hour; the time length of the sub-segment is 1 minute.

[0233] According to an embodiment of the present disclosure, the electrocardiogram data is 24-hour single-lead electrocardiogram data.

[0234] According to an embodiment of the present disclosure, it also includes:

[0235] Extracting a plurality of candidate ECG parameters from each sub-segment, the candidate ECG parameters comprising: peak-related ECG parameters, area-related ECG parameters and time-related ECG parameters;

[0236] Among the candidate ECG parameters, the candidate ECG parameters whose differences between the high-risk SCD group subjects and the healthy control group subjects meet preset conditions are selected as the designated ECG parameters.

[0237] According to an embodiment of the present disclosure, the peak-related ECG parameters include: maxR, minR, maxS, minS, maxT, minT, mean_R, mean_S, mean_T, RS, len_ST, T_sd and mean_RS; the area-related ECG parameters include: inteRM, inteRm, inteSM, inteSm, inteTM, inteTm, inteR_mean, inteS_mean, inteT_mean, inteR_sum, inteS_sum, inteT_sum and inteT_sd; the time-related ECG parameters include: t_ST, t_RT, t_RS, t_T, t_R_sum, t_S_sum, t_T_sum, t_QT, QT_sd, t_T_sd and t_plus;

[0238] in:

[0239] maxR is the maximum value of the R peak in the corresponding sub-segment;

[0240] minR is the minimum value of the R peak within the corresponding sub-segment;

[0241] maxS is the maximum value of the S peak value in the corresponding sub-segment;

[0242] minS is the minimum value of the S peak in the corresponding sub-segment;

[0243] maxT is the maximum value of the T peak in the corresponding subsegment;

[0244] minT is the minimum value of the T peak within the corresponding subsegment;

[0245] mean_R is the average value of the R peak value in the corresponding sub-segment;

[0246] mean_S is the average value of the S peak value in the corresponding sub-segment;

[0247] mean_T is the average value of T peaks in the corresponding sub-segment;

[0248] RS is the difference between the R peak and the S peak in the corresponding sub-segment;

[0249] len_ST is the average of the differences between the S peak and the T peak within the corresponding subsegment;

[0250] T_sd is the standard deviation of the T peak within the corresponding sub-fragment;

[0251] mean_RS is the average of the differences between the R peak and the S peak within the corresponding sub-segment;

[0252] inteRM is the maximum area of ​​the R peak within the corresponding subfragment;

[0253] inteSM is the maximum area of ​​the S peak within the corresponding subfragment;

[0254] inteTM is the maximum area of ​​the T peak within the corresponding subfragment;

[0255] inteRm is the minimum area of ​​the R peak within the corresponding subfragment;

[0256] inteSm is the minimum area of ​​the S peak within the corresponding subfragment;

[0257] inteTm is the minimum area of ​​the T peak within the corresponding subfragment;

[0258] inteR_mean is the average area of ​​the R peak in the corresponding subfragment;

[0259] inteS_mean is the average area of ​​the S peak in the corresponding subfragment;

[0260] inteT_mean is the average area of ​​the T peak within the corresponding subfragment;

[0261] inteR_sum is the total area of ​​the R peaks within the corresponding subfragment;

[0262] inteS_sum the total area of ​​the S peak in the corresponding sub-fragment;

[0263] inteT_sum the total area of ​​the T peaks in the corresponding subfragment;

[0264] inteT_sd is the standard deviation of the T peak area within the corresponding subfragment;

[0265] t_ST is the time interval from the beginning of the S wave to the end of the T wave in the corresponding subsegment;

[0266] t_RT is the time interval from the beginning of the R wave to the end of the T wave in the corresponding subsegment;

[0267] t_RS is the time interval from the beginning of the R wave to the end of the S wave in the corresponding sub-segment;

[0268] t_T is the time interval from the beginning of the T wave to the end of the T wave in the corresponding sub-segment;

[0269] t_R_sum is the total time interval between the start of the R wave and the end of the R wave within the corresponding subsegment;

[0270] t_S_sumS is the total time interval between the start of the S wave and the end of the S wave in the corresponding subsegment;

[0271] t_T_sum is the total time interval between the start of the T wave and the end of the T wave within the corresponding subsegment;

[0272] t_QT is the average time interval between the beginning of the R wave and the end of the T wave within the corresponding subsegment;

[0273] QT_sd is the standard deviation of the time interval between the start of the R wave and the end of the T wave within the corresponding subsegment;

[0274] t_T_sd is the standard deviation of the time interval between the start and end of the T wave within the corresponding sub-segment.

[0275] According to an embodiment of the present disclosure, the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, inteSsum, mean_RS, t_T_sd and t_plus; or the designated ECG parameters include: inteSM, inteSmean, t_RT, t_T, mean_RS, t_T_sd and t_plus.

[0276] According to an embodiment of the present disclosure, the preset condition is: using a T test to examine the differences in candidate ECG parameters between subjects in the high-risk SCD group and subjects in the healthy control group, the obtained P value is less than a set value.

[0277] According to an embodiment of the present disclosure, the value is set to a value obtained by dividing 0.05 by the number of candidate ECG parameters.

[0278] According to an embodiment of the present disclosure, the method further includes:

[0279] The SCD risk prediction model with the best performance among the SCD risk prediction models in each time period was selected for SCD risk prediction.

[0280] According to an embodiment of the present disclosure, the time period corresponding to the SCD risk prediction model with the best performance includes the time point 18:00 and / or the time point 19:00; the SCD risk prediction model with the best performance is the SCD risk prediction model of 18:00-19:00.

[0281] According to an embodiment of the present disclosure, the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period is selected for SCD risk prediction, including:

[0282] During the training process, the area under the first ROC curve was calculated for each SCD risk prediction model;

[0283] The performance of the SCD risk prediction model is determined based on the area under the first ROC curve.

[0284] According to an embodiment of the present disclosure, the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period is selected for SCD risk prediction, and further includes:

[0285] Acquire a test data set, wherein the test data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects;

[0286] In the same segmentation manner as that for the training data set, each electrocardiogram data in the test data set is segmented into a plurality of test electrocardiogram data segments, and each test electrocardiogram data segment is segmented into a plurality of test sub-segments, each test electrocardiogram data segment corresponds to a corresponding time period;

[0287] extracting from each test sub-segment the same set of ECG parameters as in the sub-segment;

[0288] For each test electrocardiogram data segment, calculating an average electrocardiogram parameter group of the test electrocardiogram data segment according to the electrocardiogram parameter group of the test sub-segment of the test electrocardiogram data segment;

[0289] For the SCD risk prediction model of each time period, using the average electrocardiogram parameter group of the corresponding test electrocardiogram data segment to calculate the area under the second ROC curve of the SCD risk prediction model;

[0290] The performance of the SCD risk prediction model determined according to the area under the first ROC curve is verified using the area under the second ROC curve.

[0291] Fig.40 A schematic diagram showing the structure of a computer system suitable for implementing the method according to an embodiment of the present disclosure is shown.

[0292] like Fig.40 As shown, the computer system includes a processing unit, which can perform the various methods in the above-mentioned embodiments according to the program stored in the read-only memory (ROM) or the program loaded from the storage part into the random access memory (RAM). In the RAM, various programs and data required for the operation of the computer system are also stored. The processing unit, ROM and RAM are connected to each other through a bus. The input / output (I / O) interface is also connected to the bus.

[0293] The following components are connected to the I / O interface: an input part including a keyboard, a mouse, etc.; an output part including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage part including a hard disk, etc.; and a communication part including a network interface card such as a LAN card, a modem, etc. The communication part performs a communication process via a network such as the Internet. The drive is also connected to the I / O interface as needed. Removable media, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive as needed so that the computer program read therefrom is installed into the storage part as needed. Among them, the processing unit can be implemented as a processing unit such as a CPU, a GPU, a TPU, an FPGA, an NPU, etc.

[0294] In particular, according to an embodiment of the present disclosure, the method described above can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes a program code for executing the above method. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium.

[0295] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0296] The units or modules involved in the embodiments described in the present disclosure may be implemented by software or programmable hardware. The units or modules described may also be set in a processor, and the names of these units or modules do not constitute limitations on the units or modules themselves in some cases.

[0297] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present disclosure is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other.

Claims

1. A method for establishing a sudden cardiac death (SCD) risk prediction model based on time-dependent electrocardiographic parameters, characterized in that: The following steps are involved: Acquire a training data set, wherein the training data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the electrocardiogram data is 24-hour single-lead electrocardiogram data, wherein both the high-risk SCD group and the healthy control group include multiple subjects, and the high-risk SCD group includes SCD survivors or patients with hemodynamic disorders caused by sustained ventricular tachycardia / ventricular fibrillation; In the same segmentation manner, each 24-hour single-lead electrocardiogram data is segmented into a plurality of electrocardiogram data segments, each electrocardiogram data segment corresponds to a corresponding time period; For each time period, the corresponding ECG data segment is divided into a plurality of sub-segments, a plurality of candidate ECG parameters are extracted from each sub-segment, and the difference of the candidate ECG parameters between the subjects in the high-risk SCD group and the subjects in the healthy control group in the time period is checked by using a statistical test method, and the candidate ECG parameters with a P value less than a set value are taken as designated ECG parameters; the candidate ECG parameters include: peak-related ECG parameters, area-related ECG parameters and time-related ECG parameters; For each electrocardiogram data segment, calculating an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter group of a sub-segment of the electrocardiogram data segment, wherein the electrocardiogram parameter group includes a plurality of specified electrocardiogram parameters; For each time period, using the average electrocardiogram parameter group of the electrocardiogram data segments corresponding to the time period to train the SCD risk prediction model of the time period, the SCD risk prediction model is used to perform SCD risk prediction based on the electrocardiogram data of the subject obtained in the time period; The SCD risk prediction model with the best performance among the SCD risk prediction models in each time period was selected for SCD risk prediction.

2. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: For each ECG data segment, the average ECG parameter group of the ECG data segment is calculated according to the ECG parameter group of the sub-segment of the ECG data segment, including: averaging the same designated ECG parameters in the ECG parameter group of the sub-segment of the ECG data segment to obtain the average ECG parameter group of the ECG data segment.

3. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The using the average electrocardiogram parameter group of the electrocardiogram data segments corresponding to the time period to train the SCD risk prediction model for the time period includes: using the average electrocardiogram parameter group of the electrocardiogram data segments corresponding to the time period to train a regression model.

4. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: For each time period, the SCD risk prediction model of the time period is trained using the average ECG parameter group of the ECG data segments corresponding to the time period, including: using the average ECG parameter group of the ECG data segments corresponding to the time period to train the SCD risk prediction model of the time period by cross-validation.

5. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The length of the time period is a specified time period between 0.5 hours and 2 hours.

6. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 5, characterized in that: The length of the time period is 1 hour, and the length of the sub-segment is 1 minute.

7. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The peak-related ECG parameters include: maxR, minR, maxS, minS, maxT, minT, mean_R, mean_S, mean_T, RS, len_ST, T_sd and mean_RS; the area-related ECG parameters include: inteRM, inteRm, inteSM, inteSm, inteTM, inteTm, inteR_mean, inteS_mean, inteT_mean, inteR_sum, inteS_sum, inteT_sum and inteT_sd; the time-related ECG parameters include: t_ST, t_RT, t_RS, t_T, t_R_sum, t_S_sum, t_T_sum, t_QT, QT_sd, t_T_sd and t_plus; in: maxR is the maximum value of the R peak in the corresponding sub-segment; minR is the minimum value of the R peak within the corresponding sub-segment; maxS is the maximum value of the S peak value in the corresponding sub-segment; minS is the minimum value of the S peak in the corresponding sub-segment; maxT is the maximum value of the T peak in the corresponding subsegment; minT is the minimum value of the T peak within the corresponding subsegment; mean_R is the average value of the R peak value in the corresponding sub-segment; mean_S is the average value of the S peak value in the corresponding sub-segment; mean_T is the average value of T peaks in the corresponding sub-segment; RS is the difference between the R peak and the S peak in the corresponding sub-segment; len_ST is the average of the differences between the S peak and the T peak within the corresponding subsegment; T_sd is the standard deviation of T peaks within the corresponding sub-fragment; mean_RS is the average of the differences between the R peak and the S peak within the corresponding sub-segment; inteRM is the maximum area of ​​the R peak within the corresponding subfragment; inteSM is the maximum area of ​​the S peak within the corresponding subfragment; inteTM is the maximum area of ​​the T peak within the corresponding subfragment; inteRm is the minimum area of ​​the R peak within the corresponding subfragment; inteSm is the minimum area of ​​the S peak within the corresponding subfragment; inteTm is the minimum area of ​​the T peak within the corresponding subfragment; inteR_mean is the average area of ​​the R peak in the corresponding subfragment; inteS_mean is the average area of ​​the S peak in the corresponding subfragment; inteT_mean is the average area of ​​the T peak within the corresponding subfragment; inteR_sum is the total area of ​​the R peaks within the corresponding subfragment; inteS_sum the total area of ​​the S peak in the corresponding sub-fragment; inteT_sum the total area of ​​the T peaks in the corresponding subfragment; inteT_sd is the standard deviation of the T peak area within the corresponding subfragment; t_ST is the time interval from the beginning of the S wave to the end of the T wave in the corresponding subsegment; t_RT is the time interval from the beginning of the R wave to the end of the T wave in the corresponding subsegment; t_RS is the time interval from the beginning of the R wave to the end of the S wave in the corresponding sub-segment; t_T is the time interval from the beginning of the T wave to the end of the T wave in the corresponding sub-segment; t_R_sum is the total time interval between the start of the R wave and the end of the R wave within the corresponding subsegment; t_S_sumS is the total time interval between the start of the S wave and the end of the S wave in the corresponding subsegment; t_T_sum is the total time interval between the start of the T wave and the end of the T wave within the corresponding subsegment; t_QT is the average time interval between the beginning of the R wave and the end of the T wave within the corresponding subsegment; QT_sd is the standard deviation of the time interval between the start of the R wave and the end of the T wave within the corresponding subsegment; t_T_sd is the standard deviation of the time interval between the start and end of the T wave within the corresponding subsegment; t_plus is the ratio of the positive T wave within the corresponding sub-segment.

8. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 7, characterized in that: The specified ECG parameters include: inteSM, inteSmean, t_RT, t_T, inteSsum, mean_RS, t_T_sd and t_plus; Or the specified ECG parameters include: inteSM, inteSmean, t_RT, t_T, mean_RS, t_T_sd and t_plus.

9. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The set value is a value obtained by dividing 0.05 by the number of candidate ECG parameters.

10. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The time period corresponding to the best performing SCD risk prediction model includes the time point 18:00 and / or the time point 19:00; the best performing SCD risk prediction model is the SCD risk prediction model from 18:00 to 19:

00.

11. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 1, characterized in that: The SCD risk prediction model with the best performance among the SCD risk prediction models in each time period is selected for SCD risk prediction, including: During the training process, the area under the first ROC curve (AUC) of each SCD risk prediction model was calculated; The performance of the SCD risk prediction model is determined based on the area under the first ROC curve.

12. The method for establishing a SCD risk prediction model based on time-dependent ECG parameters according to claim 11, characterized in that: The step of selecting the SCD risk prediction model with the best performance among the SCD risk prediction models in each time period for SCD risk prediction further includes: Acquire a test data set, wherein the test data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the high-risk SCD group and the healthy control group each include a plurality of subjects; In the same segmentation manner as that for the training data set, each electrocardiogram data in the test data set is segmented into a plurality of test electrocardiogram data segments, and each test electrocardiogram data segment is segmented into a plurality of test sub-segments, each test electrocardiogram data segment corresponds to a corresponding time period; extracting from each test sub-segment the same set of ECG parameters as in the sub-segment; For each test electrocardiogram data segment, calculating an average electrocardiogram parameter group of the test electrocardiogram data segment according to the electrocardiogram parameter group of the test sub-segment of the test electrocardiogram data segment; For the SCD risk prediction model of each time period, using the average electrocardiogram parameter group of the corresponding test electrocardiogram data segment to calculate the area under the second ROC curve of the SCD risk prediction model; The performance of the SCD risk prediction model determined according to the area under the first ROC curve is verified using the area under the second ROC curve.

13. A device for establishing a SCD risk prediction model based on time-dependent ECG parameters, characterized in that: include: A first acquisition module is configured to acquire a training data set, wherein the training data set includes electrocardiogram data of a high-risk SCD group and a healthy control group, wherein the electrocardiogram data is 24-hour single-lead electrocardiogram data, wherein the high-risk SCD group and the healthy control group both include multiple subjects, and the high-risk SCD group includes SCD survivors or patients with hemodynamic disorders caused by sustained ventricular tachycardia / ventricular fibrillation; A first segmentation module is configured to segment each 24-hour single-lead electrocardiogram data into a plurality of electrocardiogram data segments in the same segmentation manner, each electrocardiogram data segment corresponds to a corresponding time period; The first extraction module is configured to divide the corresponding ECG data segment into a plurality of sub-segments for each time period, extract a plurality of candidate ECG parameters from each sub-segment, use a statistical test method to check the difference of the candidate ECG parameters between the subjects in the high-risk SCD group and the subjects in the healthy control group in the time period, and take the candidate ECG parameters with a P value less than a set value as the designated ECG parameters; the candidate ECG parameters include: peak-related ECG parameters, area-related ECG parameters and time-related ECG parameters; A first calculation module is configured to calculate, for each electrocardiogram data segment, an average electrocardiogram parameter group of the electrocardiogram data segment according to the electrocardiogram parameter group of a sub-segment of the electrocardiogram data segment, wherein the electrocardiogram parameter group includes a plurality of specified electrocardiogram parameters; A training module is configured to train, for each time period, a SCD risk prediction model for the time period using an average electrocardiogram parameter group of the electrocardiogram data segments corresponding to the time period, wherein the SCD risk prediction model is used to perform SCD risk prediction based on the electrocardiogram data of the subject obtained in the time period; Among them, the SCD risk prediction model with the best performance in each time period is selected for SCD risk prediction.

14. An electronic device, characterized in that: The method comprises a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method steps described in any one of claims 1 to 12.

15. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the method steps of any one of claims 1 to 12 are implemented.

16. A computer program product, comprising computer instructions, which, when executed by a processor, implement the method steps of any one of claims 1 to 12.

Images