Exercise electrocardiogram-based coronary heart disease risk assessment method and system
Through the coronary heart disease risk assessment method and system based on exercise electrocardiogram, the evaluation model is used to output the probability value of coronary heart disease, and the problem of inconsistent time and accuracy of exercise electrocardiogram interpretation in the prior art is solved, and the rapid and Accurate diagnosis and medical treatment decisions for coronary heart disease.
Patent Information
- Application Number
- PCT/CN2023/139440
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-16
- Filing Date
- 2023-12-18
- Publication Date
- 2024-11-21
AI Technical Summary
In the prior art, the interpretation of exercise electrocardiograms takes a long time and depends on physician experience, resulting in inconsistent diagnosis time and accuracy, inability to conduct immediate interpretation, and lack of auxiliary diagnostic methods and systems to accelerate and improve the diagnosis of coronary heart disease.
Coronary heart disease risk assessment method and system based on exercise electrocardiogram, motor ECG data is obtained from memory through a processor, data screening and processing is performed, and the evaluation model is used to output the probability value of coronary heart disease, and auxiliary physicians decide on medical treatment.
It achieves rapid and accurate risk judgment of coronary heart disease, reduces the clinical workload of physicians, and improves the accuracy and efficiency of diagnosis.
Smart Images

Figure CN2023139440_21112024_PF_FP_ABST
Abstract
Description
Coronary heart disease risk assessment method and system based on exercise electrocardiogram Technical Field
[0001] The present invention relates to a coronary heart disease risk assessment method and system thereof, and in particular to a coronary heart disease risk assessment method and system thereof based on exercise electrocardiogram. Background Art
[0002] The most commonly used clinical tool for detecting coronary heart disease is the exercise electrocardiogram (ECG). However, interpreting an exercise ECG is not only time-consuming but also relies on the physician's extensive experience. Furthermore, the time and accuracy of interpretation vary between physicians, and due to factors such as busy clinical schedules, immediate interpretation of an exercise ECG is often not possible.
[0003] In view of this, the market currently lacks a method and system to help doctors accelerate and improve the diagnosis of coronary heart disease, so relevant industry players are all seeking solutions.
[0004] Summary of the Invention
[0005] The purpose of the present invention is to provide a coronary heart disease risk assessment method and system based on exercise electrocardiogram, which can assist physicians in determining the risk of coronary heart disease and deciding subsequent medical treatment methods.
[0006] According to one embodiment of a method aspect of the present invention, a coronary heart disease risk assessment method based on exercise electrocardiogram (ECG) is provided for assessing a coronary heart disease risk value based on exercise electrocardiogram (ECG) data of a user. The coronary heart disease risk assessment method based on exercise electrocardiogram (ECG) data includes a data acquisition step, a first processing step, and a screening step. The data acquisition step includes driving a processor to acquire the user's exercise ECG data from a memory. The exercise ECG data includes multiple ECG signal segments and multiple physiological measurement features. The first processing step includes driving the processor to pair each ECG signal segment with multiple exercise stages and select multiple evaluation signal segments corresponding to each exercise stage. The screening step includes driving the processor to screen each evaluation signal segment and each physiological measurement feature based on a first screening condition and a second screening condition different from the first screening condition to generate a screening result, and then determining whether to generate a coronary heart disease risk value based on the screening result. When the processor determines to generate a coronary heart disease risk value based on the screening result, the processor is driven to generate multiple filtered evaluation signal segments and multiple filtered physiological measurement features. The exercise electrocardiogram-based coronary heart disease risk assessment method further includes a second processing step and an evaluation step. The second processing step includes driving a processor to generate a plurality of derived features based on each post-screening physiological measurement feature, and performing a normalization process on each post-screening physiological measurement feature and each derived feature to generate a plurality of normalized physiological measurement features. The evaluation step includes driving the processor to retrieve an evaluation model from a memory and input each post-screening evaluation signal segment and each normalized physiological measurement feature into the evaluation model to generate a coronary heart disease probability value.
[0007] Other embodiments of the aforementioned embodiment are as follows: the screening step further includes a first screening step and a second screening step. The first screening step includes driving the processor to confirm whether each evaluation signal segment meets a first screening condition, thereby generating a first screening result. The second screening step includes driving the processor to confirm whether each physiological measurement feature meets a second screening condition, thereby generating a second screening result. When the first screening result is yes, the processor generates each post-screening evaluation signal segment and proceeds to the second screening step; when the first screening result is no, the processor terminates the screening step; when the second screening result is yes, the processor generates each post-screening physiological measurement feature and proceeds to the second processing step; when the second screening result is no, the processor terminates the screening step.
[0008] Other embodiments of the aforementioned embodiment are as follows: Each exercise stage includes a pre-exercise stage, an exercise stage, and a recovery stage. Each evaluation signal segment includes a pre-stage ECG signal segment with the last measurement time in the pre-exercise stage, a mid-stage ECG signal segment with the fastest measured heartbeat in the exercise stage, and a recovery stage ECG signal segment with the earliest measurement time in the recovery stage.
[0009] Other examples of the aforementioned embodiment are as follows: the first screening condition is that none of the early stage ECG signal segments, the middle stage ECG signal segments, and the recovery stage ECG signal segments are missing.
[0010] Other examples of the aforementioned embodiment are as follows: the second screening condition is that a missing number of each physiological measurement feature is less than a preset value.
[0011] Other examples of the aforementioned embodiment are as follows: the second processing step further includes driving the processor to compensate for each post-screening physiological measurement feature.
[0012] Other examples of the aforementioned embodiment are as follows: The exercise ECG-based coronary heart disease risk assessment method further includes a model training step. The model training step includes a training data acquisition step, a data pre-processing step, and a weight training step. The training data acquisition step includes driving a processor to retrieve multiple historical ECG data and corresponding multiple cardiac catheterization data from a memory. Each historical ECG data includes multiple training ECG signal segments and multiple training physiological measurement features corresponding to each exercise stage. Each cardiac catheterization data is cardiac catheterization data measured within six months. The data pre-processing step includes driving the processor to pair each training ECG signal segment with each exercise stage to select multiple training signal segments corresponding to each exercise stage, supplement missing training physiological measurement features to generate multiple training derived features, and normalize each training physiological measurement feature and each training derived feature to generate multiple normalized training physiological measurement features. The weight training step includes driving the processor to provide weights for each training signal segment and each normalized training physiological measurement feature based on each cardiac catheterization data, and to train the assessment model.
[0013] Other embodiments of the aforementioned implementation are as follows: the model training step further includes a model correction step, which includes driving the processor to input each training signal segment and each normalized training physiological measurement feature into the evaluation model to generate a training result, and the processor can correct the evaluation model by comparing the training result with each cardiac catheter data.
[0014] Other embodiments of the aforementioned implementation manner are as follows: each exercise stage is a pre-exercise stage, an exercise stage and a recovery stage; the three training signal segments are a pre-stage training signal segment with the last measurement time in the pre-exercise stage, a mid-stage training signal segment with the fastest measured heart rate in the exercise stage, and a recovery stage training signal segment with the earliest measurement time in the recovery stage.
[0015] Other examples of the aforementioned embodiment are as follows: a file format of the exercise electrocardiogram data and each historical electrocardiogram data is an Extensible Markup Language (XML) file; the exercise electrocardiogram-based coronary heart disease risk assessment method further includes: when the probability value of coronary heart disease is greater than or equal to a probability threshold value, determining a subsequent medical treatment method for the user.
[0016] According to one embodiment of the present invention, a coronary heart disease risk assessment system based on exercise electrocardiogram (ECG) is provided for assessing a coronary heart disease risk value based on exercise electrocardiogram (ECG) data of a user. The coronary heart disease risk assessment system includes a memory and a processor. The memory stores the exercise electrocardiogram (ECG) data and an assessment model. The exercise electrocardiogram (ECG) data includes multiple ECG signal segments and multiple physiological measurement features. A processor is coupled to the memory and configured to perform an operation comprising the following steps: a data acquisition step, a first processing step, and a screening step. The data acquisition step includes acquiring the user's exercise ECG data from the memory. The first processing step includes pairing each ECG signal segment with multiple exercise stages and selecting multiple evaluation signal segments corresponding to each exercise stage. The screening step includes screening each evaluation signal segment and each physiological measurement feature according to a first screening condition and a second screening condition different from the first screening condition to generate a screening result, and determining whether to generate a coronary heart disease risk value based on the screening result. When the processor determines to generate a coronary heart disease probability value based on the screening results, the processor generates multiple post-screening evaluation signal segments and multiple post-screening physiological measurement features. The processor also performs the following operations: a second processing step and an evaluation step. The second processing step includes generating multiple derived features based on each post-screening physiological measurement feature, and performing a normalization process on each post-screening physiological measurement feature and each derived feature to generate multiple normalized physiological measurement features. The evaluation step includes retrieving an evaluation model from a memory and inputting each post-screening evaluation signal segment and each normalized physiological measurement feature into the evaluation model to generate a coronary heart disease probability value.
[0017] Other embodiments of the aforementioned embodiment are as follows: The screening step further includes a first screening step and a second screening step. The first screening step includes determining whether each evaluation signal segment satisfies a first screening condition, thereby generating a first screening result. The second screening step includes determining whether each physiological measurement feature satisfies a second screening condition, thereby generating a second screening result. When the first screening result is positive, the processor generates each post-screening evaluation signal segment and proceeds to the second screening step. When the first screening result is negative, the processor terminates the screening step. When the second screening result is positive, the processor generates each post-screening physiological measurement feature and proceeds to the second processing step. When the second screening result is negative, the processor terminates the screening step.
[0018] Other embodiments of the aforementioned embodiment are as follows: Each exercise stage includes a pre-exercise stage, an exercise stage, and a recovery stage. Each evaluation signal segment includes a pre-stage ECG signal segment with the last measurement time in the pre-exercise stage, a mid-stage ECG signal segment with the fastest measured heartbeat in the exercise stage, and a recovery stage ECG signal segment with the earliest measurement time in the recovery stage.
[0019] Other examples of the aforementioned embodiment are as follows: the first screening condition is that none of the early stage ECG signal segments, the middle stage ECG signal segments, and the recovery stage ECG signal segments are missing.
[0020] Other examples of the aforementioned embodiment are as follows: the second screening condition is that a missing number of each physiological measurement feature is less than a preset value.
[0021] Other examples of the aforementioned embodiment are as follows: the second processing step further includes performing a value compensation on each post-screening physiological measurement feature.
[0022] Other embodiments of the aforementioned embodiment are as follows: The processor further includes the following steps: a model training step including a training data acquisition step, a data pre-processing step, and a weight training step. The training data acquisition step includes acquiring a plurality of historical ECG data and corresponding plurality of cardiac catheterization data from a memory, each historical ECG data including a plurality of training ECG signal segments and a plurality of training physiological measurement features corresponding to each exercise stage, and each cardiac catheterization data is cardiac catheterization data measured within six months. The data pre-processing step includes pairing each training ECG signal segment with each exercise stage to select a plurality of training signal segments corresponding to each exercise stage, supplementing missing training physiological measurement features to generate a plurality of training derived features, and normalizing each training physiological measurement feature and each training derived feature to generate a plurality of normalized training physiological measurement features. The weight training step includes driving the system to provide weights for each training signal segment and each normalized training physiological measurement feature based on each cardiac catheterization data, and training the evaluation model.
[0023] Other embodiments of the aforementioned implementation are as follows: the model training step further includes a model correction step, which includes inputting each training signal segment and each normalized training physiological measurement feature into the evaluation model to generate a training result, and the processor can correct the evaluation model by comparing the training result with each cardiac catheter data.
[0024] Another embodiment of the aforementioned embodiment is as follows: Each exercise stage is a pre-exercise stage, an exercise stage, and a recovery stage. The three training signal segments are a pre-exercise stage training signal segment with the last measurement time in the pre-exercise stage, a mid-exercise stage training signal segment with the fastest heart rate measurement in the exercise stage, and a recovery stage training signal segment with the earliest measurement time in the recovery stage.
[0025] Other examples of the aforementioned embodiment are as follows: a file format of the exercise electrocardiogram data and each historical electrocardiogram data is an Extensible Markup Language file. The processor further includes the following operation: when the probability value of coronary heart disease is greater than or equal to a probability threshold value, determining a subsequent medical treatment method for the user.
[0026] Thus, the present invention's exercise electrocardiogram-based coronary heart disease risk assessment method and system, by capturing and screening coronary heart disease-related characteristic data and combining it with an assessment model to output the probability of coronary heart disease, can assist physicians in determining the risk of coronary heart disease and deciding subsequent medical treatment methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] FIG1 is a schematic diagram illustrating a coronary heart disease risk assessment system based on exercise electrocardiogram according to a first embodiment of the present invention;
[0028] FIG2 is a flow chart illustrating a method for assessing coronary heart disease risk based on exercise electrocardiogram according to a second embodiment of the present invention;
[0029] FIG3 is a flow chart illustrating the model training steps in the exercise electrocardiogram-based coronary heart disease risk assessment method according to FIG2 ; and
[0030] FIG. 4 is a schematic diagram illustrating exercise electrocardiogram data according to the present invention.
[0031] [Description of symbols] 100: Coronary heart disease risk assessment system based on exercise electrocardiogram 110: Memory 111: Exercise electrocardiogram data 112: Assessment model 113: Historical electrocardiogram data 114: Cardiac catheter data 120: Processor 200: Coronary heart disease risk assessment method based on exercise electrocardiogram C1: First screening condition C2: Second screening condition S01: Model training step S011: Training data acquisition step S012: Data pre-processing step S013: Weight training step S014: Model correction step S02: Data acquisition step S03: First processing step S04: Screening step S041: First screening step S042: Second screening step S05: Second processing step S06: Assessment step DETAILED DESCRIPTION
[0032] The following describes various embodiments of the present invention with reference to the accompanying drawings. For clarity, many practical details will be included in the following description. However, it should be understood that these practical details are not intended to limit the present invention. In other words, in some embodiments of the present invention, these practical details are not essential. Furthermore, to simplify the drawings, some commonly used structures and components are depicted in simplified schematic form; and duplicate components may be represented using the same reference numerals.
[0033] In addition, in this article, when a certain element (or unit or module, etc.) is "connected" to another element, it may mean that the element is directly connected to the other element, or it may mean that the element is indirectly connected to the other element, that is, there are other elements between the element and the other element. When it is explicitly stated that a certain element is "directly connected" to another element, it means that there are no other elements between the element and the other element. The terms "first", "second", "third", etc. are only used to describe different elements and do not limit the elements themselves. Therefore, the first element can also be renamed as the second element. Moreover, the combination of elements / units / circuits in this article is not a generally known, conventional or customary combination in this field. Whether the elements / units / circuits themselves are conventional cannot be used to determine whether their combination relationship can be easily completed by a person of ordinary skill in the technical field.
[0034] Referring to Figures 1 and 2 , Figure 1 is a schematic diagram illustrating a first embodiment of an exercise ECG-based coronary heart disease risk assessment system 100, and Figure 2 is a flow chart illustrating a second embodiment of an exercise ECG-based coronary heart disease risk assessment method 200. The exercise ECG-based coronary heart disease risk assessment system 100 is configured to implement the exercise ECG-based coronary heart disease risk assessment method 200, which assesses a coronary heart disease risk probability value based on exercise ECG data 111 of a user. It should be noted that the exercise ECG-based coronary heart disease risk assessment method 200 of the present invention is not limited to implementation by the exercise ECG-based coronary heart disease risk assessment system 100 of the present invention. In the first and second embodiments, severe coronary heart disease is defined as a condition with ≥50% stenosis of the main coronary artery or ≥70% stenosis of other coronary arteries.
[0035] In FIG1 , a coronary heart disease risk assessment system 100 based on exercise electrocardiogram (ECG) includes a memory 110 and a processor 120 coupled to the memory 110. The memory 110 stores exercise electrocardiogram (ECG) data 111, an assessment model 112, a plurality of historical ECG data 113, and a plurality of cardiac catheterization data 114 corresponding to the plurality of historical ECG data 113. The processor 120 is configured to implement the coronary heart disease risk assessment method 200 based on exercise electrocardiogram (ECG) (as shown in FIG2 ). The memory 110 may be a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions for execution by the processor 120, but the present invention is not limited thereto. The processor 120 may be a processor, a microprocessor, a central processing unit (CPU), a computer, a mobile device processor, a cloud processor, or other electronic computing processor, but the present invention is not limited thereto. The evaluation model 112 may be a Convolutional Recurrent Neural Network (CRNN) model.
[0036] In FIG2 , the exercise electrocardiogram-based coronary heart disease risk assessment method 200 includes a model training step S01, a data acquisition step S02, a first processing step S03, and a screening step S04, which are performed sequentially. When the processor 120 determines to generate a coronary heart disease probability value based on a screening result in screening step S04, the exercise electrocardiogram-based coronary heart disease risk assessment method 200 further includes a second processing step S05 and an evaluation step S06.
[0037] 2 and 3 , FIG3 is a flow chart illustrating the model training step S01 of the exercise ECG-based coronary heart disease risk assessment method 200 according to FIG2 . The model training step S01 includes a training data acquisition step S011 , a data pre-processing step S012 , a weight training step S013 , and a model modification step S014 , which are sequentially performed.
[0038] 3 , the training data acquisition step S011 includes driving the processor 120 to acquire a plurality of historical ECG data 113 and corresponding cardiac catheterization data 114 from the memory 110 . The historical ECG data 113 and the corresponding cardiac catheterization data 114 are obtained from historical medical records of different subjects.
[0039] The historical ECG data 113 is formatted as an XML file and contains multiple training ECG signal segments and multiple training physiological measurement features corresponding to different exercise phases, where the exercise phases include a pre-exercise phase, an exercise phase, and a recovery phase. The training ECG signal segments are 12-lead ECG signals for each exercise phase of the subject, with the 12-lead ECG signals for each phase recorded for 10 seconds at a sampling rate of 500 Hz. The training physiological measurement features include physiological features and measurement features. Physiological features include physiological age (Age), gender (Sex), height, and weight information. Measurement features include peak heart rate (PHR), maximum predicted heart rate (Max Predicted Heart Rate (%)), resting heart rate (RHR), maximum heart rate and systolic blood pressure product (RPP), maximum metabolic equivalent (MW), maximum ST-segment depression (Max ST), ST-segment / heart rate index (ST / HR index), and exercise termination reason, but the present invention is not limited thereto. Cardiac catheterization data 114 is cardiac catheterization data measured over a six-month period for different subjects. Specifically, if a subject has previously been diagnosed with severe coronary heart disease via exercise electrocardiography, the physician will perform a cardiac catheterization on the subject within six months to obtain cardiac catheterization data 114. The cardiac catheterization data will indicate the degree of stenosis in each of the subject's coronary arteries.
[0040] Data pre-processing step S012 includes driving processor 120 to pair training ECG signal segments with pre-exercise, mid-exercise, and recovery phases, thereby selecting a training signal segment corresponding to each of the pre-exercise, mid-exercise, and recovery phases. Further, processor 120 pairs exercise phases and heart rates using the time in the XML file, and selects the three most meaningful training signal segments for analyzing coronary heart disease based on the following conditions: the three training signal segments are a pre-exercise training signal segment with the last measurement time in the pre-exercise phase, a mid-exercise training signal segment with the fastest measured heart rate in the mid-exercise phase, and a recovery training signal segment with the first measurement time in the recovery phase. If any of the pre-exercise training signal segment, the mid-exercise training signal segment, and the recovery training signal segment are missing, the corresponding subject's historical ECG data 113 will not be used.
[0041] The data pre-processing step S012 further includes driving the processor 120 to fill in the missing training physiological measurement features to generate a plurality of training derived features, and performing a normalization process on the training physiological measurement features and the training derived features to generate a plurality of normalized training physiological measurement features. It is further explained that the method of filling in the missing values of the training physiological measurement features is to fill in the average value. In the normalization process of the training physiological measurement features and the training derived features, the gender feature is processed by one hot encoding (One Hot Encoder), and the other features are all processed by min-max normalization (Min-Max normalization). The training derived features are the body mass index (BMI) calculated based on height and weight, the chronotropic index calculated based on heart rate, resting heart rate, and age, and the predicted metabolic equivalent percentage (Percent predicted METs) calculated based on gender and age, but the present invention is not limited thereto.
[0042] In FIG3 , the weight training step S013 includes driving the processor 120 to provide weights for each training signal segment and each normalized training physiological measurement feature based on the cardiac catheterization data 114, and to train the evaluation model 112. Further, the processor 120 divides each subject's training signal segment and each normalized training physiological measurement feature into ten groups based on the number of subjects, eight of which serve as training and validation sets. Two groups are used as test sets, and are assigned different training weights based on the presence or absence of severe coronary heart disease according to the cardiac catheterization results through class weight adjustment to train the evaluation model 112 and the classifier. The model correction step S014 includes driving the processor 120 to input the training signal segment and normalized training physiological measurement feature into the evaluation model 112 to generate a training result. The processor 120 can then compare the training result with the cardiac catheterization data 114 to correct the evaluation model 112 for subsequent evaluation of the user's probability of coronary heart disease.
[0043] Referring to Figures 2 and 4, Figure 4 is a schematic diagram illustrating the exercise electrocardiogram data 111 of the present invention. The data acquisition step S02 includes driving the processor 120 to obtain the user's exercise electrocardiogram data 111 from the memory 110. A file format of the exercise electrocardiogram data 111 is an XML file, and includes multiple electrocardiogram signal segments and multiple physiological measurement features. Among them, the electrocardiogram signal segments are 12-lead electrocardiogram signals of each exercise stage of the user, and their recording time and sampling frequency are the same as the aforementioned training electrocardiogram signal segments, so they are not repeated here; the physiological measurement features are the same as the aforementioned training physiological measurement features, so they are not repeated here. The first processing step S03 includes driving the processor 120 to pair the multiple electrocardiogram signal segments of the exercise electrocardiogram data 111 with the pre-exercise stage, the mid-exercise stage and the recovery stage, and select an evaluation signal segment corresponding to the pre-exercise stage, the mid-exercise stage and the recovery stage respectively. The three evaluation signal segments are respectively a pre-stage ECG signal segment with the last measurement time in the pre-exercise stage, a mid-stage ECG signal segment with the fastest heartbeat in the exercise stage, and a recovery stage ECG signal segment with the earliest measurement time in the recovery stage. <phasename>In the , the exercise stage name corresponding to the ECG signal segment will be displayed, among which the pre-exercise stage is PRETEST, the exercise stage is EXERCISE, and the recovery stage is RECOVERY.
[0044] FIG4 further illustrates a detailed method of pairing each ECG signal segment with a motion stage: Assuming that one wants to know the motion stage corresponding to an ECG signal segment, first, the motion ECG data 111 is obtained. <stripdata>Get the measurement time of the ECG signal segment <time>28 seconds. Then, compare to <trenddata>of <entrytime>, can be confirmed <entrytime>At 28 seconds <heartrate>is 94bpm. Finally, <phasetime>28 seconds, you can <phasename>It is known that the exercise stage corresponding to this ECG signal segment is the pre-exercise stage. It should be noted that in this embodiment of the present invention, since the file format of the exercise ECG data 111 is an XML file, the probability of coronary heart disease can be directly analyzed using the raw data in the XML file, eliminating the need for clinical physicians to make judgments based on ECG waveform PDF files. This reduces the physician's clinical workload and improves the accuracy of the examination.
[0045] In Figure 2, screening step S04 includes driving the processor 120 to screen the evaluation signal segments and physiological measurement features based on a first screening condition C1 and a second screening condition C2 different from the first screening condition C1 to generate a screening result, and determining whether to generate a coronary heart disease probability value based on the screening result. When the processor 120 determines to generate a coronary heart disease probability value based on the screening result, it drives the processor 120 to generate multiple post-screening evaluation signal segments and multiple post-screening physiological measurement features. Specifically, screening step S04 includes a first screening step S041 and a second screening step S042, which are performed sequentially. The first screening step S041 involves driving the processor 120 to determine whether the three evaluation signal segments meet the first screening condition C1 and generate a first screening result. If the first screening result is yes, the processor 120 generates the post-screening evaluation signal segments and proceeds to the second screening step S042. If the first screening result is no, the processor 120 terminates screening step S04 and stops the subsequent coronary heart disease assessment. The second screening step S042 involves driving the processor 120 to determine whether the physiological measurement characteristics meet the second screening condition C2 and generate a second screening result. If the second screening result is yes, the processor 120 generates the screened physiological measurement characteristics and proceeds to the second processing step S05. If the second screening result is no, the processor 120 terminates the screening step S04 and stops the subsequent coronary heart disease assessment. The first screening condition C1 is that none of the early-stage ECG signal segments, mid-stage ECG signal segments, or recovery-stage ECG signal segments are missing. The second screening condition C2 is that the number of missing physiological measurement characteristics is less than a preset value. In the second embodiment, the preset value is 5.
[0046] In FIG2 , the second processing step S05 includes driving the processor 120 to fill in missing values in the post-screening physiological measurement features, generate multiple derived features based on the post-screening physiological measurement features, and perform normalization processing on the post-screening physiological measurement features and the derived features to generate multiple normalized physiological measurement features. The normalization processing method for the post-screening physiological measurement features and the derived features is the same as the normalization processing for the trained physiological measurement features and the trained derived features, and thus will not be repeated here. The derived features are the same as the trained derived features, and thus will not be repeated here. The evaluation step S06 includes driving the processor 120 to obtain the evaluation model 112 from the memory 110, and input the post-screening evaluation signal segment and the normalized physiological measurement features into the evaluation model 112 to generate an auxiliary diagnosis report to generate a coronary heart disease probability value. When the coronary heart disease probability value is greater than or equal to a probability threshold value, a subsequent medical treatment method for the user is determined. Specifically, the auxiliary diagnostic report includes the user's normalized physiological measurements, exercise data, coronary heart disease risk, and ECG waveforms for the initial, intermediate, and recovery phases of the ECG signal. It should be noted that in the second screening step S042, if the user's physiological measurements meet the second screening condition C2 but still contain missing data, a warning indicating missing data will be included in the auxiliary diagnostic report.
[0047] The following is a further example, using a user with coronary heart disease and a user without coronary heart disease as case studies:
[0048] <Case 1> User suffers from coronary heart disease
[0049] In data acquisition step S02, exercise ECG data 111 corresponding to the user is acquired. In first processing step S03, multiple ECG signal segments in exercise ECG data 111 are paired with the pre-exercise phase, the exercise phase, and the recovery phase, and corresponding evaluation signal segments are acquired. The pairing results and the corresponding evaluation signal segments are shown in Table 1A below.
[0050] In screening step S04, Table 1A above shows that none of the early-stage, mid-stage, and recovery-stage ECG signal segments are missing, thus satisfying the first screening condition C1. The corresponding physiological measurement characteristics of the user are shown in Table 1B below. Table 1B shows that the number of missing physiological measurement characteristics is 1, thus satisfying the second screening condition C2.
[0051] In the second processing step S05, missing values (NaNs) in the post-screening physiological measurement features are imputed, and multiple derived features are generated based on the post-screening physiological measurement features. The post-screening physiological measurement features and derived features corresponding to the user are shown in Table 1C below, and the normalized physiological measurement features are shown in Table 1D below. As can be seen from Table 1C below, the missing feature Max ST in Table 1B has been imputed using the mean value.
[0052] In evaluation step S06, an auxiliary diagnosis report is generated to produce a coronary heart disease probability value. The auxiliary diagnosis report is shown in Table 1E below. As shown in Table 1E, the coronary heart disease probability value for user Case 1 is 0.9587124. When the coronary heart disease probability value is greater than or equal to a probability threshold (e.g., 50%), the user's subsequent medical treatment is determined. Since the user in Case 1 is assessed to have coronary heart disease, subsequent medical treatment is required, and the medical treatment for Case 1 is the placement of a heart stent.
[0053] <Case 2> The user does not have coronary heart disease
[0054] In data acquisition step S02, exercise ECG data 111 corresponding to the user is acquired. In first processing step S03, multiple ECG signal segments in exercise ECG data 111 are paired with the pre-exercise phase, the exercise phase, and the recovery phase, and corresponding evaluation signal segments are acquired. The pairing results and corresponding evaluation signal segments are shown in Table 2A below.
[0055] In screening step S04, Table 2A above shows that none of the early-stage, mid-stage, and recovery-stage ECG signal segments are missing, thus satisfying the first screening condition C1. The corresponding physiological measurement characteristics of the user are shown in Table 2B below. Table 2B shows that the number of missing physiological measurement characteristics is 1, thus satisfying the second screening condition C2.
[0056] In the second processing step S05, missing values in the post-screening physiological measurement features are supplemented, and multiple derived features are generated based on the post-screening physiological measurement features. The post-screening physiological measurement features and derived features corresponding to the user are shown in Table 2C below, and the normalized physiological measurement features are shown in Table 2D below. As can be seen from Table 2C below, the missing feature Max ST in Table 2B has been supplemented by the mean value.
[0057] In evaluation step S06, an auxiliary diagnosis report is generated to produce a coronary heart disease probability value. The auxiliary diagnosis report is shown in Table 2E below. As shown in Table 2E, the coronary heart disease probability value for Case 2 is 0.04812117. When the coronary heart disease probability value is greater than or equal to a probability threshold (e.g., 50%), a subsequent medical treatment method is determined for the user. Since Case 2 is assessed as not having coronary heart disease, no subsequent medical treatment is required.
[0058] As can be seen from the above embodiments, the present invention has the following advantages: First, by outputting the probability of coronary heart disease through an evaluation model, it can assist physicians in making a rapid diagnosis and deciding on subsequent medical treatment. Second, by capturing and filtering characteristic data related to coronary heart disease and combining it with the evaluation model, the accuracy of coronary heart disease diagnosis can be improved, while reducing the clinical workload of physicians.
[0059] Although the present invention has been disclosed above with reference to the embodiments, they are not intended to limit the present invention. Anyone with ordinary knowledge in the technical field may make slight changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.< / phasename> < / phasetime> < / heartrate> < / entrytime> < / entrytime> < / trenddata> < / time> < / stripdata> < / phasename>
Claims
1. A coronary heart disease risk assessment method based on exercise electrocardiogram, for assessing a coronary heart disease risk value based on an exercise electrocardiogram data of a user, the coronary heart disease risk assessment method based on exercise electrocardiogram is characterized in that Include: a data acquisition step, comprising driving a processor to acquire the exercise electrocardiogram data of the user from a memory, wherein the exercise electrocardiogram data comprises a plurality of electrocardiogram signal segments and a plurality of physiological measurement features; A first processing step includes driving the processor to pair the plurality of ECG signal segments with a plurality of motion stages, and selecting a plurality of evaluation signal segments corresponding to each of the motion stages; as well as a screening step, comprising driving the processor to screen the plurality of evaluation signal segments and the plurality of physiological measurement features according to a first screening condition and a second screening condition different from the first screening condition to generate a screening result, and determining whether to generate the coronary heart disease probability value according to the screening result; When the processor determines to generate the coronary heart disease probability value according to the screening result, the processor is driven to generate a plurality of post-screening evaluation signal segments and a plurality of post-screening physiological measurement features. The coronary heart disease risk assessment method based on exercise electrocardiogram further includes: a second processing step, comprising driving the processor to generate a plurality of derived features according to the plurality of post-screening physiological measurement features, and performing a normalization process on the plurality of post-screening physiological measurement features and the plurality of derived features to generate a plurality of normalized physiological measurement features; and An evaluation step includes driving the processor to obtain an evaluation model from a memory, and inputting the plurality of filtered evaluation signal segments and the plurality of normalized physiological measurement features into the evaluation model to generate the coronary heart disease susceptibility value.
2. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 1, characterized in that: This screening step also includes: A first screening step, comprising driving the processor to determine whether the plurality of evaluation signal segments satisfy the first screening condition to generate a first screening result; and A second screening step, comprising driving the processor to confirm whether the plurality of physiological measurement characteristics meet the second screening condition to generate a second screening result; When the first screening result is yes, the processor generates the plurality of post-screening evaluation signal segments and continues to perform the second screening step; when the first screening result is no, the processor ends. The screening step; When the second screening result is yes, the processor generates the plurality of post-screening physiological measurement characteristics and continues to execute the second processing step. When the second screening result is no, the processor ends the screening step.
3. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 2, characterized in that: The plurality of exercise phases include a pre-exercise phase, an exercise phase, and a recovery phase; and The multiple evaluation signal segments include a pre-stage ECG signal segment with the last measurement time in the pre-exercise stage, a mid-stage ECG signal segment with the fastest measured heartbeat in the exercise stage, and a recovery stage ECG signal segment with the earliest measurement time in the recovery stage.
4. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 3, characterized in that: The first screening condition is that any one of the pre-stage ECG signal segment, the middle-stage ECG signal segment, and the recovery-stage ECG signal segment is not missing.
5. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 2, characterized in that: The second screening condition is that a missing number of the plurality of physiological measurement features is less than a preset value.
6. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 5, characterized in that: The second processing step further includes driving the processor to supplement the plurality of post-screening physiological measurement features.
7. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 1, characterized in that Also includes: A model training step includes: a training data acquisition step, comprising driving the processor to acquire a plurality of historical electrocardiogram data and a plurality of corresponding cardiac catheter data from the memory, wherein the plurality of historical electrocardiogram data comprises a plurality of training electrocardiogram signal segments and a plurality of training physiological measurement features corresponding to each of the exercise stages, and the plurality of training electrocardiogram signal segments and a plurality of corresponding cardiac catheter data; The cardiac catheterization data are the cardiac catheterization data measured within six months; a data pre-processing step, comprising driving the processor to pair the plurality of training ECG signal segments with the plurality of motion stages to select a plurality of training signal segments corresponding to each of the motion stages, supplementing the missing plurality of training physiological measurement features to generate a plurality of training derived features, and performing the normalization process on the plurality of training physiological measurement features and the plurality of training derived features to generate a plurality of normalized training physiological measurement features; and A weight training step includes driving the processor to provide the plurality of training signal segments and the plurality of normalized training physiological measurement feature weights according to the plurality of cardiac catheter data, and to train the evaluation model.
8. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 7, characterized in that: The model training steps also include: A model modification step includes driving the processor to input the plurality of training signal segments and the plurality of normalized training physiological measurement features into the evaluation model to generate a training result. The processor can modify the evaluation model by comparing the training result with the plurality of cardiac catheter data.
9. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 7, characterized in that: The plurality of exercise phases are respectively a pre-exercise phase, an exercise phase and a recovery phase; and The three training signal segments are respectively a pre-stage training signal segment with the last measurement time in the pre-exercise stage, a mid-stage training signal segment with the fastest measured heart rate in the mid-exercise stage, and a recovery stage training signal segment with the earliest measurement time in the recovery stage.
10. The method for assessing the risk of coronary heart disease based on exercise electrocardiogram according to claim 7, characterized in that: A file format of the exercise electrocardiogram data and the plurality of historical electrocardiogram data is an eXtensible Markup Language file; The exercise electrocardiogram-based coronary heart disease risk assessment method also includes: When the probability value of coronary heart disease is greater than or equal to a probability threshold value, it is determined that A follow-up medical treatment method for the user.
11. A coronary heart disease risk assessment system based on exercise electrocardiogram, for assessing a coronary heart disease risk value based on an exercise electrocardiogram data of a user, wherein the coronary heart disease risk assessment system based on exercise electrocardiogram is characterized in that Include: a memory storing the exercise electrocardiogram data and an evaluation model, wherein the exercise electrocardiogram data includes a plurality of electrocardiogram signal segments and a plurality of physiological measurement features; and A processor is coupled to the memory, and the processor is configured to perform operations including the following steps: a data acquisition step, comprising acquiring the exercise electrocardiogram data of the user from the memory; a first processing step, comprising pairing the plurality of ECG signal segments with a plurality of motion stages, and selecting a plurality of evaluation signal segments corresponding to each of the motion stages; and a screening step, comprising screening the plurality of evaluation signal segments and the plurality of physiological measurement features according to a first screening condition and a second screening condition different from the first screening condition to generate a screening result, and determining whether to generate the coronary heart disease probability value according to the screening result; When the processor determines to generate the coronary heart disease probability value according to the screening result, the processor generates a plurality of post-screening evaluation signal segments and a plurality of post-screening physiological measurement features, and the processor further comprises the following steps: a second processing step, comprising generating a plurality of derived features according to the plurality of post-screening physiological measurement features, and performing a normalization process on the plurality of post-screening physiological measurement features and the plurality of derived features to generate a plurality of normalized physiological measurement features; and An evaluation step includes obtaining the evaluation model from the memory, and inputting the plurality of filtered evaluation signal segments and the plurality of normalized physiological measurement features into the evaluation model to generate the coronary heart disease susceptibility value.
12. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 11, characterized in that: This screening step also includes: a first screening step, comprising confirming whether the plurality of evaluation signal segments satisfy the first screening condition to generate a first screening result; and A second screening step includes confirming whether the plurality of physiological measurement characteristics meet the second screening conditions. The second screening result is generated by Wherein, when the first screening result is yes, the processor generates the plurality of post-screening evaluation signal segments and continues to perform the second screening step, and when the first screening result is no, the processor ends the screening step; When the second screening result is yes, the processor generates the plurality of post-screening physiological measurement characteristics and continues to execute the second processing step. When the second screening result is no, the processor ends the screening step.
13. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 12, characterized in that: The plurality of exercise phases include a pre-exercise phase, an exercise phase, and a recovery phase; and The multiple evaluation signal segments include a pre-stage ECG signal segment with the last measurement time in the pre-exercise stage, a mid-stage ECG signal segment with the fastest measured heartbeat in the exercise stage, and a recovery stage ECG signal segment with the earliest measurement time in the recovery stage.
14. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 13, characterized in that: The first screening condition is that any one of the pre-stage ECG signal segment, the middle-stage ECG signal segment, and the recovery-stage ECG signal segment is not missing.
15. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 12, characterized in that: The second screening condition is that a missing number of the plurality of physiological measurement features is less than a preset value.
16. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 15, characterized in that: The second processing step further includes performing compensation on the plurality of post-screening physiological measurement features.
17. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 11, characterized in that: The processor also includes the following steps: A model training step includes: a training data acquisition step, comprising acquiring a plurality of historical electrocardiogram data and a plurality of corresponding cardiac catheter data from the memory, wherein the plurality of historical electrocardiogram data comprises a plurality of training electrocardiogram signal segments and a plurality of training physiological measurement features corresponding to each exercise stage, and the plurality of cardiac catheter data are cardiac catheter data measured within six months; a data pre-processing step, comprising pairing the plurality of training ECG signal segments with the plurality of motion stages to select a plurality of training signal segments corresponding to the plurality of motion stages, supplementing the missing plurality of training physiological measurement features to generate a plurality of training derived features, and performing the normalization process on the plurality of training physiological measurement features and the plurality of training derived features to generate a plurality of normalized training physiological measurement features; and A weight training step includes driving the plurality of training signal segments and the plurality of normalized training physiological measurement feature weights according to the plurality of cardiac catheter data, and training the evaluation model.
18. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 17, characterized in that: The model training steps also include: A model modification step includes inputting the plurality of training signal segments and the plurality of normalized training physiological measurement features into the evaluation model to generate a training result. The processor can modify the evaluation model by comparing the training result with the plurality of cardiac catheter data.
19. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 17, characterized in that: The plurality of exercise phases are respectively a pre-exercise phase, an exercise phase and a recovery phase; and The three training signal segments are respectively a pre-stage training signal segment with the last measurement time in the pre-exercise stage, a mid-stage training signal segment with the fastest measured heart rate in the mid-exercise stage, and a recovery stage training signal segment with the earliest measurement time in the recovery stage.
20. The coronary heart disease risk assessment system based on exercise electrocardiogram according to claim 17, characterized in that: The file format of the exercise electrocardiogram data and the plurality of historical electrocardiogram data is an eXtensible markup language files; The processor also includes the following operations: When the probability value of suffering from coronary heart disease is greater than or equal to a probability threshold value, a subsequent medical treatment method of the user is determined.
Citation Information
Patent Citations
Exercise electrocardiogram data processing method and device, computer equipment and storage medium
CN114052746A
Modeling method of coronary heart disease motion reactivity prediction model, model and equipment
CN115394448A
Coronary heart disease risk assessment method, coronary heart disease risk assessment device, sudden cardiac death risk assessment method and sudden cardiac death risk assessment system
CN115581465A
Risk monitoring
US20020095094A1
Electrocardiographic assessment of arrhythmia risk
US20120265086A1
