An autonomic nervous regulation event detection method based on body surface physiological signal decoupling

Abstract

This invention proposes a method for detecting autonomic nervous system-mediated events based on the decoupling of surface physiological signals. The method includes the following steps: acquiring surface electrophysiological activity signals from the chest using a surface electrophysiological acquisition device; separating the surface electrophysiological activity signals using a rapid independent component analysis method to obtain a reference electrocardiogram (ECG) signal and a mixed residual signal; using the R-wave sequence of the reference ECG signal as a time reference, excluding the high-energy ECG window centered on the R-wave from the mixed residual signal, and extracting neural-related electrical activity feature vectors from the low cardiac electrical activity intervals between heartbeats; constructing a sequence of cardiac electrical and neural potential activity states, establishing a sequence model, and outputting the continuous time trajectory of the autonomic nervous system-mediated state. This invention uses the R-wave peak value of the reference ECG as a time reference to locate and exclude high-energy ECG windows, and extracts neural-related features from the low cardiac electrical activity intervals between heartbeats, significantly reducing the interference of QRS residues on the assessment of autonomic nervous system-mediated regulation and improving feature stability and repeatability.

Description

Technical Field

[0001] This invention relates to the field of physiological signal analysis and health detection technology, specifically to a method for detecting autonomic nervous system-mediated events based on the decoupling of physiological signals from the body surface. Background Technology

[0002] The autonomic nervous system plays a crucial role in regulating physiological processes such as cardiovascular function, fluid balance, and stress response. Abnormalities in its regulatory state are often closely related to various clinical and sub-health conditions. Therefore, continuous and objective monitoring of autonomic nervous system regulation is of great significance for assessing changes in individual physiological status, assisting clinical decision-making, and health management. Existing autonomic nervous system assessment methods mostly rely on indirect indicators such as heart rate variability or employ invasive neural recording techniques. The former is easily affected by factors such as arrhythmias, exercise, and respiration, making it difficult to accurately reflect the dynamic changes in autonomic nervous system regulation. The latter suffers from problems such as high invasiveness and difficulty in long-term application.

[0003] In recent years, skin nerve-related electrical activity (SKNA) monitoring based on surface electrical signals has provided a new technical approach for autonomic nervous system research. However, under chest patch acquisition conditions, the raw surface electrical activity signal simultaneously contains cardiac electrophysiological activity (ECG), electromyography (EMG), respiratory / body movement artifacts, and weak autonomic nervous system-related electrical activity, with significant energy differences among different components. Existing techniques mostly rely on simple filtering or fixed threshold determination, making it difficult to stably suppress interference and obtain autonomic nervous system modulation-related characterizations suitable for continuous monitoring against a background of strong cardiac electrical activity. Existing techniques often focus on the direct separation of specific signal components or rely on fixed thresholds for event determination, making it difficult to balance individual differences and temporal continuity, and lacking the ability to systematically model changes in autonomic nervous system modulation states under complex physiological scenarios. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a method for detecting autonomic nervous system modulation events based on the decoupling of surface physiological signals. This method enables continuous modeling and change analysis of the autonomic nervous system modulation state based on surface electrical activity signals without directly separating neural discharge components, thus solving the problems mentioned in the background section. The technical solution provided by this invention is as follows:

[0005] A method for detecting autonomic nervous system-mediated events based on decoupling from surface physiological signals includes the following steps:

[0006] S1, a body surface electrophysiology acquisition device is used to acquire the electrical activity signal of the chest surface. The acquisition device includes at least one set of surface electrodes, a front-end analog amplification and filtering circuit, an analog-to-digital converter and a data recording / transmission module.

[0007] S2, the collected body surface electrical activity signals are separated by a rapid independent component analysis method to obtain a reference ECG signal and a mixed residual signal. The reference ECG signal is used for R-peak detection, heartbeat type detection and RR interval calculation.

[0008] S3 uses the R-peak sequence of the reference ECG signal as the time reference, excludes the high-energy ECG window centered on the R wave from the mixed residual signal, and extracts the neural-related electrical activity feature vector in the low cardiac electrical activity interval between heartbeats.

[0009] S4. Construct a sequence of cardiac electrical activity and neural potential activity as dual-channel inputs, establish a sequence model, and output the continuous time trajectory of the autonomic nervous system regulation state.

[0010] Preferably, the body surface electrical activity signal in S1 includes cardiac electrophysiological activity, chest wall electromyographic activity, respiratory-related modulation, and body movement artifacts.

[0011] Preferably, the body surface electrical activity signal described in S1 is digitized by an analog-to-digital converter after front-end conditioning, and the analog-to-digital conversion resolution is 12 bits, 16 bits or higher.

[0012] Preferably, the specific steps of S2 are as follows:

[0013] S2.1, for body surface electrical activity signals Bandpass filtering was performed with a frequency band of [0,50] Hz to obtain a reference signal that includes the coupling between ECG and respiration. ;

[0014] S2.2, based on the independent component analysis method, signal component separation is performed to obtain a mixed residual signal containing autonomic nerve activity and thoracic respiratory electromyography. ;

[0015] S2.3, based on reference signal Perform R-peak detection, heart rate category detection, and RR interval sequence calculation:

[0016]

[0017]

[0018]

[0019] in The time when the k-th R-peak occurs. The R-peak sequence represents the category label of the k-th heartbeat. , heart beat event sequence and RR interval sequences Used in subsequent steps.

[0020] Preferably, the specific steps of S3 are as follows:

[0021] S3.1, at each R peak Nearby definition of ECG high-energy window ; in each cardiac cycle Inside, remove the high-energy windows at both ends. and And select the low cardiac electrical activity interval from the remaining interval. ;

[0022] S3.2, in each Inside, from Extract mixed features of neural electrical activity to form a feature vector. ,in Logarithmic energy characteristics, For quantile amplitude characteristics, The number of neural electrical activities, This represents the total duration of neural electrical activity. Neural electrical activity energy.

[0023] The preferred method for determining neural electrical activity is:

[0024] First, estimate the local scale using the median and the median absolute difference (MAD):

[0025]

[0026] Then define the threshold. For coefficients:

[0027]

[0028] Will satisfy The continuous segments were identified as neural electrical activity.

[0029] Preferably, the specific steps of S4 are as follows:

[0030] S4.1, Constructing a sequence of cardiac electrical activity states at the beat level , ;

[0031] S4.2, Constructing a sequence of cardiac beat-level neural potential activity states ;

[0032] S4.3, the cardiac electrical activity state sequence C and the neural potential activity state sequence N are used as dual-channel inputs and aligned on each heartbeat k to obtain a joint input pair: This results in a dual-channel timing input: ;

[0033] S4.4, Constructing a Sequence Model , input sequence of dual channels Mapped to autonomic nervous system regulatory state sequence :

[0034]

[0035] in Indicates the first The values ​​of autonomic nervous system regulation corresponding to each time segment. This represents the number of time segments.

[0036] Preferably, the values ​​of the autonomic nervous system regulation state are smoothed. This is to reduce the impact of noise and short-term artifacts on trajectory estimation.

[0037] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0038] To address the issue that ECG energy is much higher than nerve potential energy in the anterior chest surface electrical signal, and that residual R waves can easily lead to misjudgment, this invention uses the peak value of the R wave in the reference ECG as a time reference to locate and exclude the high-energy window of the ECG. It extracts nerve-related features in the low cardiac electrical activity interval between heartbeats, significantly reducing the interference of residual QRS on the assessment of autonomic nervous system regulation, and improving the stability and repeatability of the features.

[0039] This invention constructs a sequence of cardiac electrical activity states (RR interval, heartbeat type, etc.) and a sequence of neural potential activity states (burst energy, suddenness, irregularity, etc.) based on heartbeats as dual-channel inputs. This allows the model to learn the state of autonomic neural regulation under explicit cardiac context constraints, effectively distinguishing premature beats, rhythm fluctuations and changes in neural regulation, and reducing false positives and scene transfer errors.

[0040] Unlike existing technologies that rely on threshold triggering or single statistical indicators for output, this invention uses a sequence model to map heartbeat-level inputs into segment-level autonomic nervous system regulation trajectory, enabling continuous characterization of the autonomic nervous system regulation process, quantification of trend changes, and localization of stage-specific abnormalities. This approach better meets the monitoring needs of process-oriented physiological scenarios. Attached Figure Description

[0041] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0042] Figure 1 This is the main flowchart of the present invention;

[0043] Figure 2 This is an illustration of the S2 signal separation in this invention;

[0044] Figure 3 This is an illustration of the high-energy and low-energy regions of cardiac electrical activity in S3 of the present invention;

[0045] Figure 4 This is a schematic diagram of the S4 neural regulation state sequence modeling of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] To make the above-mentioned objectives, features and effects of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] Example 1: Refer to Figures 1 to 4 A method for detecting autonomic nervous system-mediated events based on the decoupling of physiological signals from the body surface includes the following steps:

[0049] S1, Acquiring surface electrical activity signals. In this embodiment, a surface electrophysiological acquisition device is used to acquire the surface electrical activity signals of the subject's chest. The acquisition device includes at least one set of surface electrodes, a front-end analog amplification and filtering circuit, an analog-to-digital converter, and a data recording / transmission module. The electrodes are preferably disposable medical Ag / AgCl adhesive electrodes or dry / wet electrodes with conductive gel. The electrodes are attached to the surface of the subject's chest skin to acquire the chest surface potential change signal. To reduce contact impedance and improve the signal-to-noise ratio, the skin preparation area can be treated before acquisition, including cleaning skin oils and sweat, and if necessary, lightly exfoliating / shaving local body hair, and ensuring that the electrodes are fully attached to the skin and firmly fixed to reduce motion artifacts and baseline drift caused by wire pulling.

[0050] The electrode arrangement can employ a single-lead or multi-lead configuration. In a single-lead configuration, differential measurement with two frontal electrodes and a reference / ground electrode (or an equivalent right leg drive electrode) can be used to improve common-mode rejection capability. In a multi-lead configuration, a multi-point electrode array can be used to enhance spatial resolution and improve robustness against local electromyographic interference. The acquisition channel preferably uses a high-input-impedance, low-noise instrumentation amplifier as the front-end amplification unit, possessing a high common-mode rejection ratio (e.g., not less than 80dB or higher) to suppress power frequency and common-mode interference. To avoid front-end saturation and suppress DC bias, the front-end circuit can include DC isolation and anti-saturation design, and can optionally be equipped with an anti-aliasing analog low-pass filter. The body surface electrical activity signal is digitized by an analog-to-digital converter after front-end conditioning; in this embodiment, the sampling rate is set to 4000Hz to cover the rapidly changing components of cardiac electrical activity and provide sufficient temporal resolution for subsequent location of high-energy regions on a beat scale and extraction of low-energy region features between beats. The analog-to-digital conversion resolution can be 12-bit, 16-bit, or higher, preferably 16-bit or higher, to improve the quantization accuracy of low-amplitude neural-related electrical activity.

[0051] During the acquisition process, the data recording / transmission module can store the digitized signal in real time on a local storage medium or transmit it to an external computing device for online processing via wired / wireless communication. To ensure long-term monitoring stability, the system can further record acquisition quality-related information, including electrode contact impedance or equivalent contact quality indicators, saturation / detachment markers, and synchronization timestamps. When an abnormal increase in contact impedance or signal saturation is detected, a quality alarm can be output or quality-weighted processing can be applied to subsequent analysis segments. The surface electrical activity signal obtained through the above acquisition scheme can simultaneously include cardiac electrophysiological activity, chest wall electromyographic activity, respiratory-related modulation and body movement artifacts, as well as skin nerve-related electrical activity components related to autonomic nervous system regulation, providing a raw data foundation for subsequent steps such as heartbeat detection, high-energy region localization of heartbeat-scale ECG, and autonomic nervous system regulation state modeling.

[0052] S2, Cardiac Reference Signal Separation and Heartbeat Detection. In this embodiment, to achieve subsequent localization of the high-energy region of the ECG at the heartbeat scale and suppression of residual ECG based on R-peak detection, it is necessary to construct a reference signal containing ECG and respiratory coupling from the continuously acquired raw SKNA signals from the anterior chest surface. Based on this, a Fast Independent Component Analysis (FastICA) method is used to separate the reference ECG component for R-peak detection, while simultaneously completing heartbeat category detection and adjacent RR interval calculation.

[0053] S2.1, Construct a reference signal. This involves analyzing the original body surface electrical activity signal. Bandpass filtering was performed with a frequency band of [0, 50] Hz to obtain a relatively clean reference signal that includes respiratory and electrocardiographic coupling. :

[0054]

[0055] in, This represents the bandpass filter operator for the 0-50Hz range. This reference signal preserves the principal components of cardiac electrical activity and their coupling information influenced by respiratory modulation, which is used for subsequent independent component separation and R-peak detection.

[0056] S2.2 Signal component separation based on independent component analysis. In this embodiment, the observation input for independent component analysis consists of only two channels:

[0057] Channel 1: Reference ECG-respiratory coupling signal The original SKNA signal is obtained by performing a [0,50]Hz bandpass filter on the original signal in step S2.1, which is used to provide a stable reference for cardiac electrical activity morphology and respiratory modulation.

[0058] Channel 2: Mixed residual signal It contains autonomic nervous system-related electrical activity and other non-cardiac-dominant components such as thoracic respiratory electromyography, and may also contain a small amount of residual cardiac electrical activity. It can be obtained from the residue after detrending / baseline removal of the original signal and appropriate suppression of the cardiac dominant component, or directly from the second acquisition channel provided by the device.

[0059] Therefore, a two-channel observation vector is constructed:

[0060]

[0061] Stacking them at discrete sampling times t yields the observation matrix:

[0062]

[0063] Where N represents the number of sampling points within the analysis segment. Under this two-channel configuration, independent component analysis (ICA) yields two independent components. , One component is usually more correlated with cardiac electrical activity (and its respiratory coupling modulation), while the other component usually contains autonomic nervous system-related electrical activity and thoracic respiratory electromyography, and may contain a small amount of residual cardiac electrical activity, which is used to locate the high-energy region of ECG on the beat scale and suppress residual effects.

[0064] S2.3, R-peak detection, heartbeat type detection, and RR interval calculation based on reference ECG signals. This is performed after obtaining the electrocardiogram-respiratory coupling signal. Then, R-peak detection was performed to obtain the R-peak sequence:

[0065]

[0066] in This represents the time (or sampling point index) at which the k-th R-peak occurs. R-peak detection is validated by the following constraints: heart rate range constraint, RR interval continuity constraint, and QRS morphological consistency constraint (maintaining similarity between adjacent heartbeats).

[0067] Subsequently, a heartbeat category detection is performed on each heartbeat to form a heartbeat event sequence:

[0068]

[0069] in This represents the label for the k-th heartbeat category, including sinus beats, atrial premature beats (PACs), or ventricular premature beats (PVCs). Heartbeat category detection can be based on QRS morphological features, RR interval patterns (such as compensatory pauses), and similarity to local templates.

[0070] Finally, the adjacent RR interval sequence is calculated:

[0071]

[0072] The obtained R peak sequence , heart beat event sequence and RR interval sequences This will be used in subsequent steps to locate the high-energy and low-energy regions of ECG using the R peak, and to construct a heartbeat-level state vector containing the RR and heartbeat category context.

[0073] S3, based on the identification of low cardiac electrical activity regions and the construction of neural-related features using heartbeat sequences. After completing the two-channel separation in S2, a mixed residual signal containing components such as autonomic neural activity and thoracic respiratory electromyography is obtained. Because the R-wave pulse energy of ECG is significantly higher than that of autonomic nerve potentials and electromyography, pulse remnants synchronized with the R-wave may still be retained even in the mixed residual signal; while the other sub-waves of ECG (such as P-waves and T-waves) and respiratory-related waveform components have been stripped away as much as possible during the reference construction and separation process. Therefore, this step uses the R-peak sequence of the reference ECG signal as the time reference, excludes the high-energy ECG window centered on the R-wave in the mixed residual signal, and extracts the feature vector of neural-related electrical activity (a mixture of autonomic nerve activity and electromyography activity) in the low cardiac electrical activity interval between heartbeats for subsequent autonomic nervous system regulation state modeling and event correlation analysis.

[0074] S3.1, identify the high-energy ECG window centered on the R wave and the interval of low cardiac electrical activity between heartbeats. To suppress the residual effects of the R wave on the ECG, [the following is done] at each R wave. Nearby definition of ECG high-energy window This is used for masking / exclusion in subsequent feature calculations. Let the window half-widths be... and (This can be a fixed value or adaptive with RR), then:

[0075]

[0076] In the case of discrete sampling (sampling rate) ), corresponding to the sample index window:

[0077]

[0078] in for The corresponding sample index. To ensure that the exclusion window covers the residual main energy of the R-wave, and The QRS width or signal energy distribution of the reference ECG can be determined; alternatively, an adaptive method can be used: calculating within the neighborhood of each heartbeat. The local energy peak range is defined by taking the smallest continuous window covering that peak as the range. In each cardiac cycle Inside, remove the high-energy windows at both ends. and And select the low cardiac electrical activity interval from the remaining interval. Used to extract neurally relevant features. (Selection) One preferred method is the RR scaling window:

[0079]

[0080] in and satisfy , .

[0081] S3.2, construct a mixed feature vector of neural-related electrical activity. In each Inside, from Extract mixed features of neural electrical activity to form a feature vector. The "neural electrical activity state" (containing mixed information of electromyography and autonomic nerve activity) used to characterize the k-th cardiac interval includes logarithmic energy characteristics, quantile amplitude characteristics, and sudden / irregular characteristics.

[0082] Logarithmic energy characteristics: ,in To prevent small constants from having a value of zero.

[0083] Quantile amplitude characteristics: , This represents the p-th quantile (e.g., the 95th quantile).

[0084] Sudden / Irregular Characteristics: Considering the periodicity of low-energy active signals generated by precordial respiration, while the neural electrical activity (burst) generated by autonomic nervous activity is irregular and its energy is generally higher than electromyography but lower than ECG residuals, this embodiment... Internal structural burst indicators. An example method is to detect bursts using a locally robust threshold:

[0085] First, estimate the local scale using the median and the median absolute difference (MAD):

[0086]

[0087] Define threshold:

[0088]

[0089] and will satisfy Continuous segments are identified as bursts, and the number of bursts is counted. Total duration and burst energy To reflect the difference between "periodicity vs. irregularity," this embodiment further introduces a consistency / periodicity measure for cross-heartbeat sequences. or Calculate the autocorrelation peak position or the main peak of the spectrum to identify regular components of respiratory cycle modulation (which will be used as control variables or exclusion terms in subsequent modeling).

[0090] Finally, the neural-related mixed feature vector is obtained: .

[0091] S4 outputs sequence modeling and autonomic nervous system regulation trajectory based on dual-channel heartbeat sequences. After completing S2 to obtain the R-peak sequence and heartbeat event sequence, and S3 to extract neural-related mixed features in the low cardiac electrical activity interval between heartbeats, this step uses the heartbeat sequence as a unified time reference to construct cardiac electrical activity state sequences and neural potential activity state sequences as dual-channel inputs, establishes a sequence model, and outputs the continuous time trajectory of the autonomic nervous system regulation state, thereby objectively depicting the autonomic nervous system regulation process and its changing trends in continuous monitoring scenarios.

[0092] S4.1, Construct a heartbeat-level sequence of cardiac electrical activity states. For each heartbeat k, the sequence is derived from the reference ECG components. and its R-peak sequence Constructing the cardiac electrical activity state vector The aforementioned It should contain at least one or more of the following cardiac electrical activity characteristics:

[0093] Adjacent RR intervals: ;

[0094] Heartbeat category labels (sinus / atrial premature beats / ventricular premature beats): ;

[0095] Therefore, the state vector of cardiac electrical activity is represented as: ;

[0096] Take photos of each heart. Arranged chronologically, the sequence of cardiac electrical activity states is obtained: .

[0097] S4.2, constructing a sequence of cardiac-beat-level neural potential activity. The high-energy window of the R peak has already been excluded within each cardiac cycle in S3. and in the low cardiac electrical activity zone between heartbeats For mixed residual signals Extracting neural-related mixed feature vectors .in It includes statistical representations such as logarithmic energy, quantile amplitude, duration and energy of neural electrical activity, spectral entropy, or complexity, used to describe the mixed state of autonomic nervous activity and electromyographic activity. Arranging n_k of each heartbeat in chronological order yields a sequence of neural potential activity states. .

[0098] S4.3, Construct a dual-channel state sequence input. The cardiac electrical activity state sequence C and the neural potential activity state sequence N are used as dual-channel inputs and aligned at each heartbeat k to obtain a joint input pair: This results in a dual-channel timing input: .

[0099] In one implementation, the implicit representation can be obtained by encoding the two channels separately:

[0100]

[0101]

[0102] Then through the fusion operator Forming a fusion representation:

[0103]

[0104] The fusion operator can be a splicing, weighted summation, gating mechanism, or cross-attention mechanism.

[0105] S4.4, the sequence model outputs the trajectory of autonomic nervous system regulation states. Constructing the sequence model. , input sequence of dual channels Mapped to autonomic nervous system regulatory state sequence :

[0106]

[0107] in Indicates the first The autonomic nervous system regulation state value corresponding to each time segment (e.g., 10 seconds). This represents the number of time segments.

[0108] The aggregation relationship between heartbeat-level input and segment-level output can be achieved through time index mapping: Let the first... The set of heartbeat indices corresponding to each segment is: The model can learn from arrive The mapping; or first obtaining the fragment representation through attention pooling / robust aggregation. Then output by the decoder :

[0109]

[0110] In another implementation, the model can simultaneously output the uncertainty of the state estimate. or event risk This is used to characterize the confidence level of state estimation and the probability of the occurrence of autonomic nervous system regulatory events.

[0111]

[0112] To adapt to continuous monitoring of autonomic nervous activity in clinical scenarios, where autonomic nervous regulation may exhibit unidirectional drift or smooth changes, state evolution constraints can be introduced during model training or inference to ensure that the output trajectory satisfies continuous change and trend priors. For example:

[0113]

[0114] And through the Constraints are imposed on the sign consistency or smoothness to reduce the impact of noise and short-term artifacts on trajectory estimation.

[0115] Example 2: The computer-readable storage medium of this example stores a computer program that, when executed by a processor, implements the steps in the method for detecting autonomic nervous system modulation events based on decoupling of body surface physiological signals in Example 1.

[0116] The computer-readable storage medium in this embodiment can be an internal storage unit of the terminal, such as the terminal's hard disk or memory; the computer-readable storage medium in this embodiment can also be an external storage device of the terminal, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc. equipped on the terminal; furthermore, the computer-readable storage medium can include both the terminal's internal storage unit and external storage devices.

[0117] The computer-readable storage medium of this embodiment is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0118] Example 3: The computer device of this example includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the method for detecting autonomic nervous system modulation events based on decoupling of body surface physiological signals in Example 1.

[0119] In this embodiment, the processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The memory can include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0120] Those skilled in the art will clearly understand that each implementation can be achieved using software plus the necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0121] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting autonomic nervous system-mediated events based on decoupling of physiological signals from the body surface, characterized in that, Includes the following steps: S1, a body surface electrophysiology acquisition device is used to acquire the electrical activity signal of the chest surface. The acquisition device includes at least one set of surface electrodes, a front-end analog amplification and filtering circuit, an analog-to-digital converter and a data recording / transmission module. S2, the collected body surface electrical activity signals are separated by a rapid independent component analysis method to obtain a reference ECG signal and a mixed residual signal. The reference ECG signal is used for R-peak detection, heartbeat type detection and RR interval calculation. S3 uses the R-peak sequence of the reference ECG signal as the time reference, excludes the high-energy ECG window centered on the R wave from the mixed residual signal, and extracts the neural-related electrical activity feature vector in the low cardiac electrical activity interval between heartbeats. S4. Construct a sequence of cardiac electrical activity and neural potential activity as dual-channel inputs, establish a sequence model, and output the continuous time trajectory of the autonomic nervous system regulation state.

2. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 1, characterized in that, The surface electrical activity signals described in S1 include cardiac electrophysiological activity, chest wall electromyographic activity, respiratory-related modulation, and body movement artifacts.

3. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 2, characterized in that, The surface electrical activity signal described in S1 is digitized by an analog-to-digital converter after being conditioned at the front end. The analog-to-digital conversion resolution is 12 bits, 16 bits or higher.

4. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 3, characterized in that, The specific steps for S2 are as follows: S2.1, for body surface electrical activity signals Bandpass filtering was performed with a frequency band of [0,50] Hz to obtain a reference signal that includes the coupling between ECG and respiration. ; S2.2, based on the independent component analysis method, signal component separation is performed to obtain a mixed residual signal containing autonomic nerve activity and thoracic respiratory electromyography. ; S2.3, based on reference signal Perform R-peak detection, heart rate category detection, and RR interval sequence calculation: ； ； ； in The time when the k-th R-peak occurs. The R-peak sequence represents the category label of the k-th heartbeat. , heart beat event sequence and RR interval sequences Used in subsequent steps.

5. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 4, characterized in that, The specific steps for S3 are as follows: S3.1, at each R peak Nearby definition of ECG high-energy window ; in each cardiac cycle Inside, remove the high-energy windows at both ends. and And select the low cardiac electrical activity interval from the remaining interval. ; S3.2, in each Inside, from Extract mixed features of neural electrical activity to form a feature vector. ,in Logarithmic energy characteristics, For quantile amplitude characteristics, The number of neural electrical activities, This represents the total duration of neural electrical activity. Neural electrical activity energy.

6. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 5, characterized in that, The method for determining neural electrical activity is as follows: First, estimate the local scale using the median and the median absolute difference (MAD): ； Then define the threshold. For coefficients: ； Will satisfy The continuous segments were identified as neural electrical activity.

7. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 5, characterized in that, The specific steps for S4 are as follows: S4.1, Constructing a sequence of cardiac electrical activity states at the beat level , ; S4.2, Constructing a sequence of cardiac beat-level neural potential activity states ; S4.3, the cardiac electrical activity state sequence C and the neural potential activity state sequence N are used as dual-channel inputs and aligned on each heartbeat k to obtain a joint input pair: This results in a dual-channel timing input: ; S4.4, Constructing a Sequence Model , input sequence of dual channels Mapped to autonomic nervous system regulatory state sequence : ； in Indicates the first The values ​​of autonomic nervous system regulation corresponding to each time segment. This represents the number of time segments.

8. The method for detecting autonomic nervous system-mediated events based on decoupling of body surface physiological signals according to claim 7, characterized in that, Smoothing the values ​​of autonomic nervous system regulation This is to reduce the impact of noise and short-term artifacts on trajectory estimation.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the method for detecting autonomic nervous system modulation events based on the decoupling of body surface physiological signals as described in any one of claims 1-8.

10. A computer device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the method for detecting autonomic nervous system modulation events based on the decoupling of body surface physiological signals as described in any one of claims 1-8.

Images