Breathing and electrocardio information integrated recording and displaying method and system based on overall spatial-temporal characteristics

Abstract

The invention discloses a breathing and electrocardio information integrated recording and displaying method and system based on overall spatial-temporal characteristics. The method comprises the steps that a high-precision sensor synchronously collects electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and breathing wave signals and preprocesses the signals based on different signal characteristics; carrying out multi-cycle electrocardiogram merging by utilizing an R-wave detection algorithm and introducing a variational mode decomposition algorithm; carrying out feature extraction from the multi-mode signal, and carrying out respiratory wave and electrocardiowave conjoint analysis; carrying out feature fusion through a spatial-temporal feature fusion network of a multi-branch attention mechanism, and generating and displaying a combined electrocardiograph distribution diagram; carrying out anomaly detection through an anomaly detection model based on a Transform architecture; synchronously displaying in an integrated display interface; according to the system and the method, efficient acquisition, analysis and visualization of the multi-modal physiological signals are realized by integrating synchronous acquisition, intelligent interpretation and integrated display of various physiological signals, and the efficiency and the accuracy of clinical diagnosis are favorably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical monitoring, and more particularly to a respiratory electrocardiogram information integrated recording display method and system based on overall spatiotemporal features. BACKGROUND

[0002] At present, with the development of medical technology, the monitoring of physiological signals of patients gradually extends from single indicators to comprehensive analysis of multi-modal data.

[0003] Traditional monitoring equipment usually collects electrocardiogram (ECG), phonocardiogram (PCG), arterial pulse waveform (ABP), venous pulse waveform (VP) and respiratory wave (RESP) respectively, and needs to rely on multiple devices for collection, and then analyzes and displays through independent algorithms.

[0004] However, the traditional monitoring method has the following disadvantages:

[0005] (1) Data isolation: the correlation between physiological signals is not fully utilized.

[0006] (2) Time synchronization problem: the difference in time resolution of different signals may cause analysis errors.

[0007] (3) Insufficient feature extraction and fusion: existing multi-modal data analysis methods mainly rely on simple superposition or split-screen display, and cannot fully utilize advanced technologies such as deep learning to efficiently extract and fuse the spatiotemporal features of multi-source signals.

[0008] (4) Low accuracy of abnormality detection: traditional interpretation models are often based on single signals or fixed rules, which are difficult to adapt to complex clinical scenarios, especially in multi-modal signal joint analysis.

[0009] (5) Poor visualization effect: the complexity of multi-modal signals makes it difficult to intuitively present them, and doctors need to spend a lot of time interpreting the data, affecting the efficiency of diagnosis.

[0010] Therefore, how to propose a respiratory electrocardiogram information integrated recording display method based on overall spatiotemporal features to realize efficient collection, analysis and visualization of multi-modal physiological signals is a problem that needs to be solved by those skilled in the art. SUMMARY

[0011] Therefore, the present application provides a respiratory electrocardiogram information integrated recording display method based on overall spatiotemporal features to solve some of the technical problems mentioned in the background art.

[0012] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0013] A method for integrated recording and display of respiratory and electrocardiographic information based on overall spatiotemporal characteristics includes the following steps:

[0014] S1. Synchronously acquire electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave signals through high-precision sensors, and preprocess each acquired signal based on different signal characteristics;

[0015] S2. Use the R-wave detection algorithm and introduce the variational mode decomposition (VMD) algorithm to merge multi-cycle electrocardiograms;

[0016] S3. Extract time-frequency and morphological features from multimodal signals of electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave, and perform joint analysis of respiratory wave and electrocardiogram;

[0017] S4. Feature fusion is performed through a spatiotemporal feature fusion network with a multi-branch attention mechanism. Based on the fusion feature results and the merging results of multi-cycle electrocardiograms, a joint electrocardiogram distribution map is generated and displayed.

[0018] S5. An anomaly detection model based on the Transformer architecture receives the results of spatiotemporal feature extraction and fusion as input, and performs anomaly detection in combination with preset judgment conditions;

[0019] S6. Display all signals synchronously in the integrated display interface and mark the anomaly detection results.

[0020] Preferably, the preprocessing of each acquired signal based on its different signal characteristics includes:

[0021] Power frequency interference removal, electromyographic noise removal, and baseline drift correction are performed on the electrocardiogram signal;

[0022] Low-frequency noise removal and high-frequency noise removal are performed on the heart sound signal;

[0023] Motion artifact removal is performed on arterial and venous pulse signals;

[0024] Baseline drift correction was performed on the respiratory signal.

[0025] Preferably, the specific content of step S2 is as follows:

[0026] S21. Use the R-wave detection algorithm to segment each cardiac cycle;

[0027] S22. Introduce the Variational Mode Decomposition (VMD) algorithm to extract the high-frequency and low-frequency components of each cardiac cycle;

[0028] S23. Align the time axes of all cardiac cycles and calculate the weighted average to generate a long-cycle electrocardiogram;

[0029] S24. Automatically identify and remove abnormal cardiac cycles.

[0030] Preferably, in step S22, the specific steps of the variational mode decomposition VMD algorithm include:

[0031] Initialize parameters K and penalty parameter a, where K represents the number of decomposed IMFs, and a controls the strictness of the bandwidth constraint;

[0032] Construct a VMD optimization problem and solve it by alternating direction method of multipliers ADMM iteration;

[0033] Extract high-frequency and low-frequency components for subsequent analysis.

[0034] Preferably, the content of time-frequency and morphological feature extraction in step S3 is:

[0035] The ECG signal features include QRS complex position, R-R interval, ST segment offset, T wave amplitude, heart rate variability; the heart sound features include heart sound spectrum energy distribution and abnormal heart sound features; the arterial pulse waveform features include rise time, fall time, peak amplitude and spectrum energy distribution; the venous pulse waveform features include waveform period, amplitude, rise slope and spectrum energy distribution; the respiratory wave features include respiratory period, amplitude, frequency and respiratory variability index.

[0036] Preferably, the content of joint analysis of respiratory wave and ECG wave is:

[0037] Extract the period and amplitude features of the respiratory wave;

[0038] Calculate the change of R-R interval within the respiratory period;

[0039] Identify respiratory-related arrhythmias by assessing the correlation between respiration and ECG signals using the Pearson correlation coefficient.

[0040] Preferably, the specific content of step S4 is:

[0041] S41. Construct a spatio-temporal feature fusion network MBASTFFN using a multi-branch attention mechanism:

[0042] S42. The multi-branch input layer receives the time-frequency and morphological features extracted from the multi-modal signals, the spatio-temporal feature encoder combines convolutional neural network CNN and bidirectional long short-term memory network Bi-LSTM to capture spatial and temporal features, the cross-modal attention mechanism enhances the correlation of different modal features through multi-head attention, and the fully connected layer and the output layer generate joint feature representation;

[0043] S43. Spatially align the generated joint features with the multi-cycle ECG merging results, introduce an attention mechanism to dynamically adjust the weight of each signal, and highlight the abnormal area;

[0044] S44. The spatial distribution of the combined ECG waves is generated using three-dimensional visualization technology to visually display the spatial relationship of different signals.

[0045] Preferably, the specific content of step S5 is:

[0046] S51. An abnormality detection model TADM based on a Transformer architecture is constructed;

[0047] S52. The input layer receives the results of spatio-temporal feature extraction and fusion;

[0048] S53. The encoder layer uses a multi-layer Transformer encoder to extract global context features;

[0049] S54. A set of preset interpretation conditions is defined, the abnormality probability is output through the decoder layer, the attention mechanism is introduced to dynamically assign weights to important features, and the model is continuously optimized through online learning.

[0050] A respiratory ECG information integrated recording and display system based on overall spatio-temporal features, based on the respiratory ECG information integrated recording and display method based on overall spatio-temporal features, comprising: a multi-source signal acquisition and preprocessing module, a multi-cycle ECG merging module, a spatio-temporal feature extraction module, a feature fusion module, an abnormality detection and interpretation model, and an integrated display interface;

[0051] The multi-source signal acquisition and preprocessing module is used to synchronously acquire ECG, PCG, arterial pulse waveform, venous pulse waveform and respiratory wave signals through high-precision sensors, and preprocess each signal based on different signal characteristics;

[0052] The multi-cycle ECG merging module is used to merge multi-cycle ECG using an R-wave detection algorithm and introducing a variational mode decomposition VMD algorithm;

[0053] The spatio-temporal feature extraction module is used to extract time-frequency and morphological features from ECG, PCG, arterial pulse waveform, venous pulse waveform and respiratory wave multi-modal signals, and perform respiratory wave and ECG wave joint analysis;

[0054] The feature fusion module is used to perform feature fusion through a multi-branch attention mechanism spatio-temporal feature fusion network, generate a combined ECG wave distribution map based on the fusion feature results and the multi-cycle ECG merging results, and display the combined ECG wave distribution map;

[0055] The abnormality detection and interpretation model is used to receive the results of spatio-temporal feature extraction and fusion as input based on a Transformer architecture abnormality detection model, and perform abnormality detection in combination with preset interpretation conditions;

[0056] The integrated display interface is used to synchronously display all signals and mark the abnormality detection results.

[0057] A processing terminal comprises a memory and a processor, the memory stores a computer program executable on the processor, and the processor implements the method when executing the computer program.

[0058] According to the technical solution, compared with the prior art, the application provides a respiratory electrocardiogram information integrated recording display method and system based on overall space-time characteristics, which integrates respiratory, electrocardiogram, heart sound and pulse waveform and other physiological signals, realizes integrated recording and display based on overall space-time characteristics, and helps to improve the accuracy and efficiency of clinical diagnosis; the variational mode decomposition algorithm is introduced to enhance the signal quality after merging through multi-cycle electrocardiogram merging technology, and solves the problem that short cycle signals cannot reflect global characteristics in traditional methods; the multi-branch attention mechanism space-time feature fusion network realizes efficient feature extraction and fusion of multi-modal signals; the abnormal detection model based on Transformer dynamically allocates weights by using the attention mechanism, and improves the abnormal detection accuracy; the attention mechanism highlights the abnormal area, and performs integrated display of multi-modal signals, and provides more intuitive visual analysis tools. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.

[0060] Figure 1 A respiratory electrocardiogram information integrated recording display method based on overall space-time characteristics provided by the present application is shown in the figure;

[0061] Figure 2 A multi-cycle electrocardiogram merging method provided by the present application is shown in the figure;

[0062] Figure 3 A space-time feature fusion method provided by the present application is shown in the figure;

[0063] Figure 4 A Transformer-based abnormal detection method provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0064] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of the present application.

[0065] The embodiment of the present application discloses a respiratory electrocardiogram information integrated recording display method based on overall spatio-temporal features, as shown in the figure, comprising the following steps: Figure 1

[0066] S1. Synchronously collecting electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave signal through high-precision sensor, and pre-processing each collected signal based on different signal characteristics;

[0067] S2. Using R-wave detection algorithm and introducing variational mode decomposition (VMD) algorithm to perform multi-cycle electrocardiogram merging;

[0068] S3. Extracting time-frequency and morphological features from electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave multi-modal signal, and performing respiratory wave and electrocardiogram wave joint analysis;

[0069] S4. Performing feature fusion through multi-branch attention mechanism spatio-temporal feature fusion network, generating joint electrocardiogram wave distribution map based on fusion feature result and multi-cycle electrocardiogram merging result, and displaying the map;

[0070] S5. Receiving the result of spatio-temporal feature extraction and fusion as input through an abnormality detection model based on Transformer architecture, and performing abnormality detection in combination with preset interpretation conditions;

[0071] S6. Synchronously displaying all signals and marking abnormality detection result in the integrated display interface.

[0072] In order to further implement the above technical solutions, the content of pre-processing each collected signal based on different signal characteristics is as follows:

[0073] Remove power frequency interference, electromyographic noise and baseline drift correction from the electrocardiogram signal;

[0074] Specifically, use a 50Hz notch filter to remove power frequency interference:

[0075]

[0076] Wherein, ω0=2π×50 / fs, fs is the sampling frequency, and r is the attenuation coefficient of the filter;

[0077] ​The electromyographic noise is removed by using wavelet transform, and a Daubechies wavelet base function is selected for decomposition and reconstruction.

[0078] The baseline drift is removed by using a high-pass filter (cut-off frequency 0.5 Hz).

[0079] The heart sound signal is subjected to low-frequency noise removal and high-frequency noise removal.

[0080] Specifically, low-frequency noise is removed by using a low-pass filter (cut-off frequency 500 Hz), and high-frequency noise is removed by using wavelet transform.

[0081] The arterial pulse and venous pulse signals are subjected to motion artifact removal.

[0082] Specifically:

[0083] Motion interference is removed by using an adaptive filtering algorithm:

[0084]

[0085] wherein x(n) is an input signal, w k (n) is an adaptive filtering coefficient.

[0086] The respiratory signal is subjected to baseline drift correction.

[0087] The baseline drift is removed by using a moving average filter:

[0088]

[0089] wherein N is a window size.

[0090] In order to further implement the above technical solutions, as shown in Figure 2 the specific content of step S2 is:

[0091] S21. Each cardiac cycle is segmented by using an R-wave detection algorithm.

[0092] S22. A variational mode decomposition (VMD) algorithm is introduced to extract high-frequency and low-frequency components of each cardiac cycle.

[0093] S23. The time axes of all cardiac cycles are aligned, and a weighted average value is calculated to generate a long-cycle electrocardiogram.

[0094] S24. Abnormal cardiac cycles are automatically identified and removed.

[0095] In order to further implement the above technical solutions, in step S22, the specific steps of the variational mode decomposition (VMD) algorithm include:

[0096] Initialize parameters K and penalty parameter α, wherein K represents the number of decomposed IMFs, and α controls the strictness of bandwidth constraint.

[0097] The VMD optimization problem is constructed and solved by the alternating direction method of multipliers (ADMM) iteratively:

[0098]

[0099] where u k (t) is the kth modal component, ω k is the center frequency of the kth mode, δ(t) is the unit impulse function, f(t) is the input cardiac cycle signal, and λ is the balance term weight;

[0100] The high-frequency and low-frequency components are extracted for subsequent analysis.

[0101] The algorithm formula of step S23 is:

[0102]

[0103] where x L (t) is the low-frequency component signal time series, x H (t) is the high-frequency component signal time series, w L is the weight of the low-frequency component, w H is the weight of the high-frequency component, and y(t) is the final weighted average result time series.

[0104] To further implement the above technical solutions, the content of time-frequency and morphological feature extraction in step S3 is:

[0105] The electrocardiogram features include QRS complex position, R-R interval, ST segment offset, and T wave amplitude, and heart rate variability.

[0106] The Pan-Tompkins algorithm is used to detect the QRS complex, and the ST segment offset and T wave amplitude are calculated.

[0107] The heart sound features include heart sound spectrum energy distribution and abnormal heart sound features.

[0108] The short-time Fourier transform is used to extract the heart sound spectrum features, and the frequency spectrum energy distribution is used to identify the murmur and split sound.

[0109] The arterial pulse waveform features include rise time, fall time, peak amplitude, and frequency spectrum energy distribution.

[0110] The venous pulse waveform features include waveform period, amplitude, rise slope, and frequency spectrum energy distribution.

[0111] The time domain features of arterial and venous pulses are extracted by calculating the rise time, fall time, and peak value of the waveform, and the fast Fourier transform (FFT) is used to calculate the frequency spectrum energy.

[0112] The respiratory wave features include respiratory cycle, amplitude, frequency, and respiratory variability index.

[0113] To further implement the above technical solutions, the content of the joint analysis of the respiratory wave and the electrocardio wave is:

[0114] extracting the cycle and amplitude features of the respiratory wave;

[0115] calculating the change of R-R interval in the respiratory cycle;

[0116] identifying the respiratory-related arrhythmia by evaluating the correlation between the respiratory wave and the electrocardio signal through the Pearson correlation coefficient;

[0117]

[0118] wherein x i is the respiratory wave signal, and y i is the electrocardio signal; if the correlation coefficient r exceeds a threshold value (such as 0.7), it is determined as the respiratory-related arrhythmia.

[0119] To further implement the above technical solutions, as shown in Figure 3 , the specific content of step S4 is:

[0120] S41. Constructing a spatio-temporal feature fusion network MBASTFFN using a multi-branch attention mechanism:

[0121] S42. The multi-branch input layer receives the time-frequency and morphological features extracted from the multi-modal signals, the spatio-temporal feature encoder combines the convolutional neural network CNN and the bidirectional long short-term memory network Bi-LSTM to capture the spatial and temporal features, the cross-modal attention mechanism enhances the correlation of different modal features through multi-head attention, and the fully connected layer and the output layer generate joint feature representation;

[0122]

[0123] wherein Q, K, and V are the query (Query), key (Key), and value (Value) matrices, d k is the dimension of the key;

[0124] S43. Spatio-temporal alignment of the generated joint features and the multi-cycle electrocardiogram merging results, introduction of attention mechanism to dynamically adjust the weight of each signal, and highlighting of the abnormal area;

[0125]

[0126] wherein M is the number of modalities, w i is the weight of the i-th modality, and F i is the feature of the i-th modality;

[0127] S44. The spatial distribution of the combined electrocardio wave is generated using three-dimensional visualization technology to intuitively show the spatial relationship of different signals.

[0128] To further implement the above technical solutions, as shown in the figure, the specific content of step S5 is: Figure 4

[0129] S51. An abnormality detection model TADM based on a Transformer architecture is constructed;

[0130] S52. The input layer receives the results of spatio-temporal feature extraction and fusion;

[0131] S53. The encoder layer uses a multi-layer Transformer encoder to extract global context features;

[0132] H t =Transformer(X)

[0133] Where X is an input feature matrix, H t is a hidden state;

[0134] S54. A set of preset interpretation conditions is defined, the abnormality probability is output through the decoder layer, the attention mechanism is introduced to dynamically assign weights to important features, and the model is continuously optimized through online learning.

[0135] In actual application, the preset interpretation conditions are, for example: heart rate anomaly: detecting a heart rate lower than 60 beats per minute or higher than 100 beats per minute; ST segment deviation: detecting an ST segment elevation or depression exceeding a set threshold; apnea: detecting a continuous decrease in respiratory amplitude to zero.

[0136] A respiratory electrocardio information integrated recording and display system based on overall spatio-temporal features, based on a respiratory electrocardio information integrated recording and display method based on overall spatio-temporal features, comprising: a multi-source signal acquisition and preprocessing module, a multi-cycle electrocardiogram merging module, a spatio-temporal feature extraction module, a feature fusion module, an abnormality detection and interpretation model, and an integrated display interface;

[0137] The multi-source signal acquisition and preprocessing module is used to synchronously acquire electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform, and respiratory wave signals through high-precision sensors, and preprocess each signal based on different signal characteristics;

[0138] In actual application, 12-lead electrocardiogram electrodes, heart sound microphones, photoelectric pulse sensors, and respiratory impedance sensors can be used for signal acquisition, and high-precision crystal oscillators (error <1 μs) are provided to realize time synchronization;

[0139] ​A multi-cycle electrocardiogram merging module is configured to utilize an R-wave detection algorithm and introduce a variational mode decomposition (VMD) algorithm to perform multi-cycle electrocardiogram merging.

[0140] A spatio-temporal feature extraction module is configured to extract time-frequency and morphological features from multi-modal signals including electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform, and respiratory waveform, and perform joint analysis of respiratory waveform and electrocardiogram waveform.

[0141] A feature fusion module is configured to perform feature fusion through a multi-branch attention mechanism spatio-temporal feature fusion network, generate a joint electrocardiogram waveform distribution map based on the fusion feature result and the multi-cycle electrocardiogram merging result, and display the joint electrocardiogram waveform distribution map.

[0142] An abnormality detection and interpretation model is configured to be based on a Transformer architecture abnormality detection model, receive the spatio-temporal feature extraction and fusion result as input, and perform abnormality detection in combination with a preset interpretation condition.

[0143] An integrated display interface is configured to synchronously display all signals and mark abnormality detection results.

[0144] A processing terminal includes a memory and a processor, the memory stores a computer program executable on the processor, and the processor implements a respiratory electrocardiogram information integrated recording and display method based on overall spatio-temporal features when executing the computer program.

[0145] The embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts of each embodiment can be referred to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the related parts can be referred to the method part.

[0146] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to the embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for integrated recording and display of respiratory electrocardiographic information based on overall spatiotemporal features, characterized by, Comprise the following steps: S1. Synchronously collect electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave signals by high-precision sensors, and preprocess each signal collected based on different signal characteristics; S2. Use R-wave detection algorithm and introduce variational mode decomposition VMD algorithm for multi-cycle electrocardiogram merging; S3. Extract time-frequency and morphological features from electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform and respiratory wave multi-modal signals, and perform respiratory wave and electrocardiogram wave joint analysis; S4. Perform feature fusion through a multi-branch attention mechanism spatio-temporal feature fusion network, generate joint electrocardiogram wave distribution map based on fusion feature results and multi-cycle electrocardiogram merging results, and display; S5. Through an abnormality detection model based on a Transformer architecture, receive the results of spatio-temporal feature extraction and fusion as input, and perform abnormality detection in combination with preset interpretation conditions; S6. Synchronously display all signals and mark abnormality detection results in an integrated display interface.

2. The method according to claim 1, wherein, The content of preprocessing each signal collected based on different signal characteristics is: Remove power frequency interference, electromyographic noise and baseline drift from the electrocardiogram signal; Remove low-frequency noise and high-frequency noise from the heart sound signal; Remove motion artifacts from the arterial pulse and venous pulse signals; Baseline drift correction is performed on the respiratory signal.

3. The method according to claim 1, wherein, The specific content of step S2 is: S21. Use R-wave detection algorithm to segment each cardiac cycle; S22. Introduce variational mode decomposition VMD algorithm to extract high-frequency and low-frequency components of each cardiac cycle; S23. Align the time axes of all cardiac cycles and calculate the weighted average to generate a long-cycle electrocardiogram; S24. Automatically identify and remove abnormal cardiac cycles.

4. The method according to claim 3, wherein, In step S22, the specific steps of the variational mode decomposition VMD algorithm include: Initialize parameters K and penalty parameter α, where K represents the number of decomposed IMFs, and α controls the strictness of bandwidth constraint; Construct a VMD optimization problem and solve it by alternating direction multiplier method ADMM iteration; Extract high-frequency and low-frequency components for subsequent analysis.

5. The method of claim 1, wherein the method is based on integrated recording and display of respiratory electrocardiographic information based on overall spatio-temporal features. The content of time-frequency and morphological feature extraction in step S3 is: Electrocardiogram signal features include QRS complex position, R-R interval, ST segment offset, T wave amplitude, and heart rate variability; heart sound features include heart sound spectrum energy distribution and abnormal heart sound features; Arterial pulse waveform features include rise time, fall time, peak amplitude and spectrum energy distribution; venous pulse waveform features include waveform period, amplitude, rise slope and spectrum energy distribution; respiratory wave features include respiratory period, amplitude, frequency and respiratory variability index.

6. The method of claim 1, wherein the method is based on integrated recording and display of respiratory electrocardiographic information based on overall spatio-temporal features. The content of respiratory wave and electrocardiogram wave joint analysis is: Extract the period and amplitude features of the respiratory wave; Calculate the change of R-R interval within the respiratory period; Identify respiratory-related arrhythmias by assessing the correlation between respiration and electrocardiogram signals using Pearson correlation coefficient.

7. The method according to claim 1, wherein, The specific content of step S4 is: S41. Construct a multi-branch attention mechanism spatio-temporal feature fusion network MBASTFFN: S42. The multi-branch input layer receives the time-frequency and morphological features extracted from the multi-modal signals. The spatio-temporal feature encoder captures spatial and temporal features using a convolutional neural network (CNN) and a bidirectional long short-term memory network (Bi-LSTM). The cross-modal attention mechanism enhances the correlation between different modal features through multi-head attention. The fully connected layer and the output layer generate a joint feature representation. S43. The generated joint features are spatio-temporally aligned with the multi-cycle electrocardiogram merging results. An attention mechanism is introduced to dynamically adjust the weights of each signal, highlighting the abnormal areas. S44. A three-dimensional visualization technique is used to generate a spatial distribution map of the joint electrocardiogram, visually displaying the spatial relationships between different signals.

8. The method of claim 1, wherein the method is based on integrated recording and display of respiratory electrocardiographic information based on overall spatio-temporal features. The specific content of step S5 is: S51. An abnormality detection model TADM based on the Transformer architecture is constructed. S52. The input layer receives the results of spatio-temporal feature extraction and fusion. S53. The encoder layer uses a multi-layer Transformer encoder to extract global context features. S54. A set of pre-defined interpretation conditions is defined. The decoder layer outputs an abnormality probability. An attention mechanism is introduced to dynamically assign weights to important features, and the model is continuously optimized through online learning.

9. A respiratory electrocardiographic information integrated recording display system based on overall spatiotemporal features, characterized by The method for recording and displaying integrated respiratory electrocardiogram information based on overall spatio-temporal features according to any one of claims 1-8, comprising: a multi-source signal acquisition and preprocessing module, a multi-cycle electrocardiogram merging module, a spatio-temporal feature extraction module, a feature fusion module, an abnormality detection and interpretation model, and an integrated display interface; The multi-source signal acquisition and preprocessing module is used to synchronously acquire electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform, and respiratory wave signals through high-precision sensors, and preprocess each signal based on different signal characteristics. The multi-cycle electrocardiogram merging module is used to merge multi-cycle electrocardiograms using an R-wave detection algorithm and introducing a variational mode decomposition (VMD) algorithm. The spatio-temporal feature extraction module is used to extract time-frequency and morphological features from electrocardiogram, phonocardiogram, arterial pulse waveform, venous pulse waveform, and respiratory wave multi-modal signals, and perform joint analysis of respiratory waves and electrocardiogram waves. The feature fusion module is used to perform feature fusion through a multi-branch attention mechanism spatio-temporal feature fusion network, generate a joint electrocardiogram distribution map based on the fusion feature results and the multi-cycle electrocardiogram merging results, and display the map. The abnormality detection and interpretation model is based on a Transformer architecture abnormality detection model, which receives the results of spatio-temporal feature extraction and fusion as input, and performs abnormality detection based on pre-defined interpretation conditions. The integrated display interface is used to synchronously display all signals and mark the abnormality detection results.

10. A processing terminal comprising a memory and a processor, the memory having stored therein a computer program executable on the processor, characterized in that, The processor executes the computer program to implement the method for recording and displaying integrated respiratory electrocardiogram information based on overall spatio-temporal features according to any one of claims 1-8.

Images