A method for pulse feature extraction in cardiovascular function assessment of depressive patients
By acquiring two-dimensional pulse signals from electrocardiogram signals and radial artery pulse reference points, spatial disharmony index and arterial vasospasm index are determined. Combined with the autonomic nervous system regulation efficiency index, a cardiovascular function risk factor for depression is constructed, which solves the problem of low assessment accuracy in existing technologies and achieves efficient assessment for early screening.
Patent Information
- Application Number
- CN202611325026.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-28
- Publication Date
- 2026-11-10
AI Technical Summary
Existing cardiovascular assessment tools have limitations in capturing early functional changes in cardiovascular function in depressive disorders. Feature extraction methods have failed to effectively construct a mapping relationship with the pathological mechanisms of cardiovascular dysfunction in depressive disorders, resulting in low accuracy in diagnosis.
By acquiring two-dimensional pulse signals from electrocardiogram signals and radial artery pulse reference points, spatial disharmony index and arterial vasospasm index were determined. Combined with the autonomic nervous system regulation efficiency index, neurovascular decompensation indicators were analyzed, and cardiovascular function risk factors for depression were constructed. Machine learning models were then used for training and evaluation.
It improves the accuracy of early screening and assessment of depressive cardiovascular dysfunction, fully maps the pathological chain from neural activation to vasospasm to compensatory failure, and enhances the sensitivity and accuracy of assessment.
Smart Images

Figure CN123058686A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical data processing technology, specifically to a method for extracting pulse features for assessing cardiovascular function in patients with depression. Background Technology
[0002] Depression is a common mental disorder, and in addition to psychiatric symptoms, patients often experience cardiovascular dysfunction. Existing cardiovascular assessment tools have limitations in capturing early functional changes. Traditional Chinese medicine pulse diagnosis contains rich cardiovascular information. By combining it with modern sensing technology and advanced signal feature extraction methods, a novel non-invasive objective assessment method for "depressive cardiovascular dysfunction" can be developed, providing a new perspective and tool for clinical risk stratification, early intervention, and efficacy monitoring.
[0003] Current pulse feature extraction methods suffer from core defects such as lack of pathological correlation and failure to capture micro-features. Most features are general signal parameters that fail to establish a mapping relationship with the specific pathological mechanisms of depressive cardiovascular dysfunction. As a result, subtle pulse variations with high diagnostic value are drowned out by noise or filtered out as invalid information during the extraction process. Consequently, the feature set cannot sensitively characterize early compensatory changes and progressive damage, leading to low accuracy in cardiovascular function assessment. Summary of the Invention
[0004] To address the technical problem of low accuracy in cardiovascular function assessment, the present invention aims to provide a pulse feature extraction method for cardiovascular function assessment in patients with depression. The specific technical solution adopted is as follows: In a first aspect, embodiments of the present invention provide a method for extracting pulse features for assessing cardiovascular function in patients with depression, the method comprising: Acquire the subject's electrocardiogram signals and two-dimensional pulse signals at various radial artery pulse reference points; The spatial misalignment index is determined based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point; The envelope phase difference of the two-dimensional pulse signal during diastole is extracted and combined with the spatial disharmony index to determine the arterial spasm index. The correlation between heart rate intensity and pulse arrival time in subjects was analyzed to determine the autonomic nervous system regulation efficiency index; By combining the rate of change of the arterial vasospasm index and the autonomic nervous system regulation efficiency index, the neurovascular decompensation index was determined. By analyzing the temporal correlation between the decompensation index and the arterial vasospasm index, cardiovascular function risk factors for depression can be identified.
[0005] Furthermore, the pulse reference points include: the cun point, the guan point, and the chi point; the two-dimensional pulse signal is a two-dimensional pulse pressure signal.
[0006] Furthermore, determining the spatial misalignment index based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point includes: Based on the propagation order between each pulse reference point, the pulse propagation time difference between adjacent pulse reference points is determined; Based on the fluctuation of the two-dimensional pulse signal at each pulse reference point, the pulse fluctuation value of each pulse reference point is determined. Based on the attenuation of pulse wave values at different pulse reference points, the vascular sclerosis characteristic values of the subjects were determined. The spatial disharmony index is determined by combining the vascular sclerosis characteristic value and the pulse propagation time difference.
[0007] Furthermore, determining the subject's arteriosclerosis characteristic values based on the attenuation of pulse wave values at different pulse reference points includes: Based on the propagation order between each pulse reference point, the average value of the ratio of the pulse wave fluctuation value of the next pulse reference point to the previous pulse reference point is calculated among all adjacent pulse reference points, and used as the characteristic value of arteriosclerosis.
[0008] Further, the extraction of the envelope phase difference of the diastolic two-dimensional pulse signal, combined with the spatial discocoordination index, to determine the arterial vasospasm index includes: For each pulse reference point, obtain the maximum and minimum phases of the two-dimensional pulse signal during diastole; By comparing the maximum and minimum phases, the pulse rhythm desynchronization index at the pulse reference point in the radial artery of the subject was determined; By combining the pulse rhythm asynchrony index of each pulse reference point within the same heartbeat cycle, the pulse rhythm asynchrony coefficient for each heartbeat cycle is determined. The arterial vasospasm index is determined by combining the pulse rhythm disorder coefficient and the spatial disharmony index; both the pulse rhythm disorder coefficient and the spatial disharmony index are positively correlated with the arterial vasospasm index.
[0009] Furthermore, the method for obtaining the pulse arrival time is as follows: The time corresponding to the R wave in the ECG signal of the subject is obtained as the initial time; the peak time of the two-dimensional pulse signal within a single heartbeat cycle is extracted. The time difference between the initial time and the peak time is calculated to obtain the time when the pulse wave arrives at the radial artery, which is taken as the pulse arrival time.
[0010] Furthermore, the analysis of the correlation between the heart rate intensity and pulse arrival time of the subjects, and the determination of the autonomic nervous system regulation efficiency index, includes: Calculate the correlation coefficient between the time series composed of heart rate intensity and the time series composed of pulse arrival time; perform negative correlation mapping on the correlation coefficient to obtain the autonomic nervous system regulation efficiency index.
[0011] Furthermore, the rate of change of the arterial vasospasm index and the autonomic nervous system regulation efficiency index is used to determine the neurovascular decompensation index, including: The change in the arterial vasospasm index over adjacent unit time intervals is used as the first change. Obtain the change in the autonomic nervous system regulation efficiency index over adjacent unit time intervals as the second change value; By combining the first change, the second change, and the trend of the first change, the unit decompensation index per unit time is determined. By combining the decompensation indices for all units of time, the neurovascular decompensation index is obtained.
[0012] Furthermore, the analysis of the time-series correlation between the decompensation index and the arterial vasospasm index to determine cardiovascular function risk factors for depression includes: The correlation between the decompensation index and the arterial vasospasm index was analyzed to obtain the time-series correlation index; By combining decompensation indicators, arterial vasospasm index, and time-series correlation index, cardiovascular function risk factors for depression were identified; among them, decompensation indicators, arterial vasospasm index, and time-series correlation index were all positively correlated with cardiovascular function risk factors.
[0013] Furthermore, after analyzing the time-series correlation between the decompensation index and the arterial vasospasm index to determine the cardiovascular function risk factors for depression, the method further includes: The machine learning model was trained using the subjects' electrocardiogram signals, two-dimensional pulse signals at various radial artery pulse reference points, and cardiovascular function risk factors as the training set.
[0014] Secondly, a pulse feature extraction system for assessing cardiovascular function in patients with depression is provided, the system comprising the following modules: The acquisition module is used to acquire the subject's electrocardiogram signal and two-dimensional pulse wave signals at various radial artery pulse reference points; The pulse propagation analysis module is used to determine the spatial misalignment index based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point. The spasm analysis module is used to extract the envelope phase difference of the two-dimensional pulse signal during diastole, and combine it with the spatial disharmony index to determine the arterial vascular spasm index; The neural regulation module is used to analyze the correlation between the subject's heart rate intensity and pulse arrival time, and to determine the autonomic nervous system regulation efficiency index. The neurovascular analysis module is used to determine neurovascular decompensation indicators by combining the rate of change of the arterial vascular spasm index and the autonomic nervous system regulation efficiency index. The risk analysis module is used to analyze the temporal correlation between the decompensation index and the arterial spasm index to determine the cardiovascular function risk factors of depression.
[0015] The embodiments of the present invention have at least the following beneficial effects: To address the shortcomings of traditional pulse feature extraction methods, such as missing pathological correlations and failure to capture micro-features, this invention, in its core process, firstly constructs a spatial disharmony index to quantify abnormal vascular reactivity based on the differences in pulse wave propagation delay and distal fluctuation attenuation characteristics at the radial artery pulse reference point. Secondly, it analyzes rhythm dysregulation through envelope phase difference analysis of diastolic pulse signals, fusing these to generate an arterial vasospasm index to characterize vasospasm caused by sympathetic nerve activation. Thirdly, by combining temporal synchronization analysis of ECG R waves and pulse peak values, it quantifies the missing correlation between pulse wave propagation time and heart rate fluctuations, constructing an autonomic nervous system regulation efficiency index. Further analysis of changes in autonomic nervous system regulation efficiency and vasospasm worsening within adjacent time periods generates and outputs a neurovascular decompensation index. Finally, it integrates the arterial vasospasm index, decompensation index, and temporal correlation to construct a risk factor for depressive cardiovascular dysfunction. This invention fully maps the pathological chain of "neural activation → vasospasm → compensation failure," improving the accuracy of early screening and assessment of depressive cardiovascular dysfunction. Attached Figure Description
[0016] Figure 1 This is a flowchart of a method for extracting pulse features for assessing cardiovascular function in patients with depression, provided in one embodiment of the present invention. Figure 2 This is a flowchart illustrating a method for obtaining the spatial misalignment index according to an embodiment of the present invention. Figure 3 This is a system block diagram of a pulse feature extraction system for assessing cardiovascular function in patients with depression, provided as an embodiment of the present invention. Detailed Implementation
[0017] The following describes in detail, with reference to the accompanying drawings, a specific scheme for pulse feature extraction for assessing cardiovascular function in patients with depression, provided by the present invention.
[0018] Please see Figure 1 The diagram illustrates a flowchart of a pulse feature extraction method for assessing cardiovascular function in depressive patients, according to an embodiment of the present invention. The method includes the following steps: Step S100: Acquire the subject's electrocardiogram signal and two-dimensional pulse image signals at each radial artery pulse reference point.
[0019] In a soundproof electromagnetic shielded room that meets medical physiological signal acquisition standards, the subject assumes a standard supine position and an integrated device containing the following sensors is deployed: High-precision pulse acquisition unit: A 16-channel array-type micro-piezoresistive sensor is fixed at the point of strongest radial artery pulsation, which in this embodiment is recorded as the radial artery pulse reference points, including: the cun point, the guan point, and the chi point. It records the two-dimensional pulsation pressure signal in real time, i.e., the two-dimensional pulse signal. In this embodiment, the 16-channel array-type micro-piezoresistive sensor has a range of 0–50 kPa, a linearity error ≤0.3% FS (full-scale percentage), and a resolution of 0.05 kPa.
[0020] ECG synchronization unit: Configured with a standard three-lead ECG, it achieves hardware-level time synchronization with the pulse signal via a GPS synchronization clock module. The ECG R-wave is marked as the hemodynamic cycle reference, and the two-dimensional pulse signal is segmented, with each heartbeat cycle defined as the time window from the first R-wave to the next. In this embodiment, the standard three-lead ECG has a sampling rate of 500Hz, and the GPS synchronization clock module achieves an error of 10ms between itself and the pulse signal hardware level.
[0021] Finally, the data collected by the integrated device is preprocessed by using a Gaussian filter for noise reduction to ensure the accuracy of the collected data.
[0022] Step S200: Determine the spatial misalignment index based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point.
[0023] Two-dimensional pulse signal recordings were obtained from various radial artery pulse reference points of the subject within a single heartbeat cycle. In patients with depression, hyperactivity of the hypothalamic-pituitary-adrenal (HPA) axis leads to sustained activation of the sympathetic nervous system, resulting in dysregulation of vascular smooth muscle tone and spatial nonuniformity of radial artery pulse propagation.
[0024] Based on this, the spatial discordance index is determined according to the pulse propagation delay and pulse propagation attenuation between each pulse reference point. Please refer to [link to relevant documentation]. Figure 2 , Figure 2 Flowchart of the method for obtaining the spatial disharmony index: Step S210: Determine the pulse propagation time difference between adjacent pulse reference points based on the propagation order between each pulse reference point.
[0025] First, the propagation delay time of the two-dimensional pulse signal at different pulse reference points in the radial artery of the subject was obtained by using a cross-correlation function; Taking the pulse propagation between the Cun and Guan points as an example, calculate the pulse propagation time difference between the Cun and Guan points. The corresponding calculation formula is: ; where argmax is the argmax function, which makes The value of the variable when it reaches its maximum value; max is the function that takes the maximum value; This represents the two-dimensional pulse information corresponding to the inch point at time t; For the key point at Two-dimensional pulse information corresponding to the moment; The time difference of the pulse wave signal between the cun point and the guan point; This represents the total duration of a single heartbeat cycle.
[0026] Furthermore, taking the pulse propagation between the guan (hand) and ulnar (inch) points as an example, the pulse propagation time difference between the guan and ulnar points is calculated. The corresponding calculation formula is: ;in, This refers to the two-dimensional pulse image information corresponding to the control point at time t; For the scale point at Two-dimensional pulse information corresponding to a given moment.
[0027] Step S220: Determine the pulse wave fluctuation value of each pulse reference point based on the fluctuation of the two-dimensional pulse wave signal at each pulse reference point.
[0028] Under normal circumstances, the pulse in the radial artery of a person weakens towards the distal end. However, in patients with depression, the distal pulse is strengthened due to hardening of the distal blood vessels. Therefore, the pulse of a patient with depression does not show a gradual weakening trend along the propagation sequence.
[0029] The variance of the two-dimensional pulse signal at each pulse reference point is obtained as the pulse fluctuation value of each pulse reference point. The variance can reflect the degree of fluctuation of the two-dimensional pulse signal within a single heartbeat cycle.
[0030] Step S230: Determine the vascular sclerosis characteristic value of the subject based on the attenuation of pulse wave values at different pulse reference points.
[0031] Based on the propagation order among pulse reference points, the average ratio of the pulse wave undulation value of the subsequent pulse reference point to that of the preceding pulse reference point is calculated and used as the vascular sclerosis characteristic value. The larger this vascular sclerosis characteristic value, the more significant the vascular sclerosis in the distal radial artery of the subject.
[0032] Step S240: Combine the vascular sclerosis characteristic value and the pulse propagation time difference to determine the spatial disharmony index.
[0033] Calculate the difference in pulse propagation delay between adjacent pulse reference points in the radial artery of the subject. : The difference in pulse propagation delay reflects the difference in vascular reactivity at different sites in the radial artery. A larger delay difference indicates decreased vascular compliance or dysfunction, corresponding to neuroregulatory disorders caused by depression.
[0034] The spatial disharmony index was determined by combining the pulse delay difference and the characteristic value of arteriosclerosis; both the pulse delay difference and the characteristic value of arteriosclerosis were positively correlated with the spatial disharmony index.
[0035] In some embodiments, the formula for calculating the spatial disharmony index is: , where ln is a function of the natural logarithm. A larger spatial discrepancy index indicates significant spatial discrepancy in the propagation of the pulse in the subject.
[0036] Step S300: Extract the envelope phase difference of the two-dimensional pulse signal during diastole, and determine the arterial vasospasm index by combining it with the spatial disharmony index.
[0037] The core pathological chain of cardiovascular dysfunction in depression is: hyperactivity of the HPA axis - excessive activation of the sympathetic nervous system - dysregulation of vascular smooth muscle tone - vasospasm. The spatial disharmony index (SDI) of the pulse signal of the subjects can only reflect the "basic manifestations" of vascular dysfunction. Further analysis is needed to analyze the "spastic deterioration" after the aggravation of cardiovascular dysfunction.
[0038] For each pulse reference point, the maximum and minimum phases of the two-dimensional pulse signal during diastole are obtained. More specifically: for each pulse reference point, the received two-dimensional pulse signal is first averaged across each channel of the sensor array, outputting and reducing the dimensionality to a one-dimensional time-domain pulse wave sequence. Then, the one-dimensional time-domain pulse wave sequence is processed using Hilbert transform to calculate the instantaneous phase data stream reflecting the sequence fluctuations. Curve fitting is performed on the instantaneous phase data stream to obtain the instantaneous phase curve. Finally, the maximum instantaneous phase in the upper envelope curve corresponding to the instantaneous phase curve during diastole is obtained as the maximum phase, and the minimum instantaneous phase in the lower envelope curve corresponding to the instantaneous phase curve is obtained as the minimum phase.
[0039] The maximum and minimum phases are compared to determine the pulse rhythm desynchronization index of the pulse reference point in the radial artery of the subject. Specifically, the maximum and minimum phases are normalized using the cosine function in the trigonometric function. The absolute value of the difference between the maximum and minimum phases after normalization is used as the pulse rhythm desynchronization index.
[0040] The pulse rhythm asynchrony index of each pulse reference point within the same heartbeat cycle is used to determine the pulse rhythm asynchrony coefficient. Specifically, the average value of the pulse rhythm asynchrony index of each pulse reference point within the same heartbeat cycle is calculated as the pulse rhythm asynchrony coefficient. A larger value of this pulse rhythm asynchrony coefficient reflects arterial pulsation rhythm disorder, which is directly related to spasmodic contraction caused by sympathetic nerve activation in depression.
[0041] The arterial vasospasm index is determined by combining the pulse rhythm disorder coefficient and the spatial disharmony index; both the pulse rhythm disorder coefficient and the spatial disharmony index are positively correlated with the arterial vasospasm index.
[0042] In some embodiments, the arterial vasospasm index is obtained by multiplying the pulse rhythm disorder coefficient and the spatial disharmony index as the arterial vasospasm index. The higher the value of the arterial vasospasm index, the more it indicates that the subject's arteries are experiencing significant spasmodic contractions due to excessive activation of the sympathetic nervous system caused by depression.
[0043] Step S400: Analyze the correlation between the subject's heart rate intensity and pulse arrival time to determine the autonomic nervous system regulation efficiency index.
[0044] Vasospasm leads to decreased hemodynamic efficiency and obstructed cardiac energy transmission, which in turn forces the autonomic nervous system to exert stronger regulation to maintain circulatory stability. Since the autonomic nervous system of patients with depression is already disordered, this compensatory need further exposes its regulatory failure, revealing the significant impact of the subjects' current cardiovascular dysfunction.
[0045] Simultaneously acquire the subject's electrocardiogram (ECG) signal and two-dimensional pulse signal within a single heartbeat cycle, and record the time corresponding to the R wave in the ECG signal. In this embodiment of the invention, the start time of the R wave in the ECG signal can be taken as the time corresponding to the R wave.
[0046] The initial time was taken as the time corresponding to the R wave in the ECG signal of the subject; the peak time of the two-dimensional pulse signal within a single heartbeat cycle was extracted. The R wave corresponds to the start of cardiac contraction. As the heart begins to contract and pump blood, the pulse wave begins to propagate through the vascular system until it reaches peripheral arteries, such as the radial artery. The peak of the pulse signal typically corresponds to the peak of the pulse wave at a particular location.
[0047] The time difference between the initial time and the peak time is calculated to obtain the time when the pulse wave arrives at the radial artery, which is taken as the pulse arrival time.
[0048] In ECG signal recordings, the R wave marks the start of cardiac electrical activity. The time it takes for the pulse wave to reach the radial artery is regulated by the tension of the sympathetic and vagus nerves, which is essentially a regulation of heart rate. However, in patients with depression, the efficiency of cardiac ejection energy transmission decreases due to autonomic nervous system disorder and imbalance, thus disrupting the physiological correlation between the time it takes for the pulse wave to reach the radial artery and heart rate fluctuations.
[0049] Based on this, a sliding window approach was used for local analysis of the ECG and two-dimensional pulse signals. Specifically, the sliding window was constructed using the synchronous signal data at the onset of the R wave as the endpoint, based on data collected within a 30-second historical period, with a window overlap rate of 75%. It should be noted that when data is insufficient, all remaining data are considered as a single sliding window. This can be understood as acquiring the subject's two-dimensional pulse signal and synchronous ECG signal, and dividing them into a series of sliding windows. .
[0050] The heart rate intensity within a heartbeat cycle is calculated as the time difference between adjacent R waves in the ECG signal within a sliding window; a smaller time difference indicates more intense heart rate fluctuations within the cycle. Each heartbeat cycle is acquired within the sliding window, and a heart rate intensity sequence is constructed according to the temporal order of the heartbeat cycles, resulting in a time-series sequence of heart rate intensities corresponding to one sliding window. For each sliding window, a pulse arrival time sequence is constructed according to the heartbeat cycle, resulting in a time-series sequence of pulse arrival times corresponding to one sliding window.
[0051] The correlation coefficient between the time series composed of heart rate intensity and the time series composed of pulse arrival time is calculated; the correlation coefficient is then negatively correlated to obtain the autonomic nervous system regulation efficiency index. In this embodiment of the invention, the correlation coefficient between the two time series is the Pearson correlation coefficient.
[0052] The correlation coefficient is used to reflect the relationship between heart rate intensity and pulse propagation time.
[0053] Normally, when the sympathetic or vagus nerves are excited, the heart rate increases while the time it takes for the pulse wave to reach the radial artery decreases. Therefore, the correlation between heart rate intensity and pulse arrival time is usually negative. However, in patients with depression, the data for both decrease simultaneously or are not significantly correlated due to the failure of neural regulation.
[0054] In some embodiments, the difference between the constant 1 and the normalized correlation coefficient can be calculated as the autonomic nervous system regulation efficiency index, achieving a negative correlation mapping to the correlation coefficient. This autonomic nervous system regulation efficiency index reflects the efficiency of the subject's autonomic nervous system regulation within the sliding window. The smaller the value, the more likely the subject has an autonomic nervous system regulation imbalance, which may be related to cardiovascular health problems caused by mood disorders. It should be noted that each sliding window corresponds to one autonomic nervous system regulation efficiency index.
[0055] In step S500, the rate of change of the arterial vasospasm index and the autonomic nervous system regulation efficiency index is combined to determine the neurovascular decompensation index.
[0056] Extract the autonomic nervous regulation efficiency index corresponding to each sliding window, and obtain the average value of the arterial vasospasm index for all heartbeat cycles within each sliding window, which is denoted as the mean arterial vasospasm index. In each pair of adjacent windows ( In this study, the changes in the mean values of the autonomic nervous system regulation efficiency index and the arterial vasospasm index were calculated. The change in the average value of the arterial vasospasm index over adjacent unit time intervals is recorded as the first change. This first change reflects the change in the arterial vasospasm index and is used to assess whether the vasomotor function has changed or whether the degree of vasospasm has worsened. For the k-th sliding window, the first change... for: ;in, The mean value of the arterial spasm index for the (k+1)th sliding window; is the average value of the arterial spasm index in the k-th sliding window.
[0057] The change in the autonomic nervous system regulation efficiency index over adjacent unit time intervals is obtained as a second change value. This second change value reflects the change in autonomic nervous system regulation efficiency, indicating whether the regulatory capacity of the autonomic nervous system changes between adjacent windows, and whether the neural regulatory function is maintained or weakened. For the k-th sliding window, its second change value... for: ;in, The autonomic neural regulation efficiency index is the (k+1)th sliding window. is the autonomic nervous system regulation efficiency index for the k-th sliding window.
[0058] In this embodiment of the invention, the time corresponding to each sliding window is taken as the unit time.
[0059] By combining the first change, the second change, and the trend of the first change, the unit decompensation index per unit time is determined.
[0060] In this embodiment of the invention, the unit decompensation index corresponding to the k-th sliding window The calculation formula is: ;in, This is the second change in the k-th sliding window; This represents the first change in the k-th sliding window; To prevent division by zero constant, it can be taken as 0.01 in this embodiment of the invention; sign is the sign function. When greater than 0, The function takes the value 1; otherwise, When less than 0, The function value is -1; when When equal to 0, The function value is 0; max is the function that takes the maximum value.
[0061] Among them, the unit decompensation index represents the intensity of the neural regulatory response triggered by vasospasm per unit time. The larger the value of the unit decompensation index, the weaker the autonomic nervous system's response to vasospasm. For example, when the spasm worsens, the compensatory decline can reflect the weakening of neural regulatory ability and indicate that the subject's depressive symptoms are relatively significant.
[0062] By combining the unit decompensation indices of all units within a time period, the neurovascular decompensation index is obtained. Specifically, the average value of the unit decompensation index corresponding to each sliding window of the subject is used as the subject's neurovascular decompensation index.
[0063] Step S600: Analyze the temporal correlation between the decompensation index and the arterial vasospasm index to determine the cardiovascular function risk factors of depression.
[0064] The total acquisition time is divided into m equal-length sub-intervals (e.g., each interval is 120 seconds), and the mean value of the arterial vasospasm index corresponding to each sub-interval is calculated. Building a dataset Calculate the mean value of the neurovascular decompensation index corresponding to each sub-interval. And construct a dataset The data in the dataset is sorted according to time sequence.
[0065] Analyzing the correlation between the decompensation index and the arterial vasospasm index yields a time-series correlation index, more specifically: Get data set , The Pearson correlation coefficient between them is used as a time-series correlation index. The closer the value of this time-series correlation index is to 1, the stronger the correlation between the arterial vasospasm index and the decompensation index. The decompensation index will exacerbate depressive cardiovascular dysfunction caused by arterial vasospasm.
[0066] By combining decompensation indicators, arterial vasospasm index, and time-series correlation index, cardiovascular function risk factors for depression were identified; among them, decompensation indicators, arterial vasospasm index, and time-series correlation index were all positively correlated with cardiovascular function risk factors.
[0067] In some embodiments, the mean of the arterial vasospasm index across all heartbeat cycles during the subject's acquisition timeline is used. And the mean of decompensation indicators for all cardiac cycles. Comprehensive quantification of cardiovascular function risk factors in current subjects : This functional risk factor characterizes the subject's risk of depressive cardiovascular dysfunction.
[0068] Ultimately, cardiovascular function risk factors obtained through pathology-driven pulse deep feature analysis will be used to... One of the key spatiotemporal dynamic features is input into the pre-trained machine learning model. This involves using cardiovascular function risk factors as the training set to train the machine learning model. The training set also includes the subject's electrocardiogram (ECG) signals and two-dimensional pulse images at various radial artery pulse reference points. The machine learning model can be a gradient boosting decision tree, a temporal convolutional network, or a graph neural network, etc. This model, trained offline, learns the complex nonlinear relationship between cardiovascular function risk factors and two-dimensional pulse signals from electrocardiogram (ECG) and radial artery pulse reference points. It can also integrate basic physiological parameters from the patient's clinical history as auxiliary features. By inputting real-time ECG signals and two-dimensional pulse signals from radial artery reference points, it can generate cardiovascular function risk factors for the subject within the current time window. It should be noted that the machine learning training process is well-known to those skilled in the art and will not be elaborated upon here.
[0069] Please see Figure 3 , Figure 3 This invention provides a system block diagram for a pulse feature extraction system for assessing cardiovascular function in patients with depression. The system includes the following modules: The acquisition module is used to acquire the subject's electrocardiogram signal and two-dimensional pulse wave signals at various radial artery pulse reference points; The pulse propagation analysis module is used to determine the spatial misalignment index based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point. The spasm analysis module is used to extract the envelope phase difference of the two-dimensional pulse signal during diastole, and combine it with the spatial disharmony index to determine the arterial vascular spasm index; The neural regulation module is used to analyze the correlation between the subject's heart rate intensity and pulse arrival time, and to determine the autonomic nervous system regulation efficiency index. The neurovascular analysis module is used to determine neurovascular decompensation indicators by combining the rate of change of the arterial vascular spasm index and the autonomic nervous system regulation efficiency index. The risk analysis module is used to analyze the temporal correlation between the decompensation index and the arterial spasm index to determine the cardiovascular function risk factors of depression.
Claims
1. A method for extracting pulse features for assessing cardiovascular function in patients with depression, characterized in that, The method includes the following steps: Acquire the subject's electrocardiogram signals and two-dimensional pulse signals at various radial artery pulse reference points; The spatial misalignment index is determined based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point; The envelope phase difference of the two-dimensional pulse signal during diastole is extracted and combined with the spatial disharmony index to determine the arterial spasm index. The correlation between heart rate intensity and pulse arrival time in subjects was analyzed to determine the autonomic nervous system regulation efficiency index; By combining the rate of change of the arterial vasospasm index and the autonomic nervous system regulation efficiency index, the neurovascular decompensation index was determined. By analyzing the temporal correlation between the decompensation index and the arterial vasospasm index, cardiovascular function risk factors for depression can be identified.
2. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The pulse reference points include: the cun point, the guan point, and the chi point; the two-dimensional pulse signal is a two-dimensional pulse pressure signal.
3. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The determination of the spatial misalignment index based on the pulse propagation delay and pulse propagation attenuation between each pulse reference point includes: Based on the propagation order between each pulse reference point, the pulse propagation time difference between adjacent pulse reference points is determined; Based on the fluctuation of the two-dimensional pulse signal at each pulse reference point, the pulse fluctuation value of each pulse reference point is determined. Based on the attenuation of pulse wave values at different pulse reference points, the vascular sclerosis characteristic values of the subjects were determined. The spatial disharmony index is determined by combining the vascular sclerosis characteristic value and the pulse propagation time difference.
4. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 3, characterized in that, The determination of the subject's arteriosclerosis characteristic values based on the attenuation of pulse wave values at different pulse reference points includes: Based on the propagation order between each pulse reference point, the average value of the ratio of the pulse wave fluctuation value of the next pulse reference point to the previous pulse reference point is calculated among all adjacent pulse reference points, and used as the characteristic value of arteriosclerosis.
5. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The extraction of the envelope phase difference of the diastolic two-dimensional pulse signal, combined with the spatial discocoordination index, to determine the arterial vasospasm index includes: For each pulse reference point, obtain the maximum and minimum phases of the two-dimensional pulse signal during diastole; By comparing the maximum and minimum phases, the pulse rhythm desynchronization index at the pulse reference point in the radial artery of the subject was determined; By combining the pulse rhythm asynchrony index of each pulse reference point within the same heartbeat cycle, the pulse rhythm asynchrony coefficient for each heartbeat cycle is determined. The arterial vasospasm index is determined by combining the pulse rhythm disorder coefficient and the spatial disharmony index; both the pulse rhythm disorder coefficient and the spatial disharmony index are positively correlated with the arterial vasospasm index.
6. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The method for obtaining the pulse arrival time is as follows: The time corresponding to the R wave in the ECG signal of the subject is obtained as the initial time; the peak time of the two-dimensional pulse signal within a single heartbeat cycle is extracted. The time difference between the initial time and the peak time is calculated to obtain the time when the pulse wave arrives at the radial artery, which is taken as the pulse arrival time.
7. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The analysis of the correlation between heart rate intensity and pulse arrival time in the subjects determined the autonomic nervous system regulation efficiency index, including: Calculate the correlation coefficient between the time series composed of heart rate intensity and the time series composed of pulse arrival time; perform negative correlation mapping on the correlation coefficient to obtain the autonomic nervous system regulation efficiency index.
8. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The rate of change of the arterial vasospasm index and the autonomic nervous system regulation efficiency index is used to determine the neurovascular decompensation indicators, including: The change in the arterial vasospasm index over adjacent unit time intervals is used as the first change. Obtain the change in the autonomic nervous system regulation efficiency index over adjacent unit time intervals as the second change value; By combining the first change, the second change, and the trend of the first change, the unit decompensation index per unit time is determined. By combining the decompensation indices for all units of time, the neurovascular decompensation index is obtained.
9. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, The analysis of the time-series correlation between the decompensation index and the arterial vasospasm index identifies cardiovascular function risk factors for depression, including: The correlation between the decompensation index and the arterial vasospasm index was analyzed to obtain the time-series correlation index; By combining decompensation indicators, arterial vasospasm index, and time-series correlation index, cardiovascular function risk factors for depression were identified; among them, decompensation indicators, arterial vasospasm index, and time-series correlation index were all positively correlated with cardiovascular function risk factors.
10. The method for extracting pulse features for assessing cardiovascular function in patients with depression according to claim 1, characterized in that, After analyzing the time-series correlation between the decompensation index and the arterial vasospasm index to determine the cardiovascular function risk factors for depression, the method further includes: The machine learning model was trained using the subjects' electrocardiogram signals, two-dimensional pulse signals at various radial artery pulse reference points, and cardiovascular function risk factors as the training set.
