43 results about "ECG feature" patented technology

The invention belongs to the technical field of medical signal processing, and discloses a sleep apnea detection method and system, electronic equipment and a storage medium, and the method comprises the following steps: obtaining synchronous ECG signals and respiratory signals, and processing the signals into bimodal data fragments containing a plurality of time scales; a parallel neural network architecture is constructed, ECG features and respiratory signal features are extracted from the data segments of all scales, and cross-modal feature pairs are formed; designing a cross-scale dynamic weight correction attention mechanism, and carrying out dynamic weighting and information fusion on cross-modal feature pairs among different scales by introducing a correction factor based on an Euclidean distance among feature vectors so as to enhance the relevance between local details and a global context; and inputting the fused multi-scale features into a classifier, and outputting the probability of the sleep apnea event. Through multi-scale collaborative analysis and a dynamic weight correction attention mechanism, physiological information of different time dimensions is effectively fused, and the detection accuracy is improved.

The application belongs to the technical field of medical signal processing, and discloses a sleep apnea detection method and system, an electronic device and a storage medium. The method comprises the following steps: acquiring a synchronous ECG signal and a breathing signal, and processing the synchronous ECG signal and the breathing signal into a bimodal data segment containing multiple time scales; constructing a parallel neural network architecture, extracting ECG features and breathing signal features from each scale data segment respectively, and forming a cross-modal feature pair; designing a cross-scale dynamic weight correction attention mechanism, introducing a correction factor based on the Euclidean distance between feature vectors, dynamically weighting and information fusion on the cross-modal feature pair between different scales, so as to enhance the relevance of local details and global context; inputting the fused multi-scale features into a classifier, and outputting the probability of a sleep apnea event. Through multi-scale collaborative analysis and a dynamic weight correction attention mechanism, the application effectively fuses physiological information of different time dimensions, and improves the detection accuracy.

The invention relates to an ECG arrhythmia classification method based on time-frequency attention enhancement and heterogeneous fusion driving. The method comprises the following steps: acquiring an original ECG signal; de-noising and overlapping slicing are carried out on the original ECG signal to obtain an ECG signal segment with a fixed length; performing time-frequency attention enhancement on the ECG signal segments to obtain time-frequency fusion ECG features; constructing a heterogeneous fusion network, and fusing channel information and position information into time-frequency fusion ECG features; constructing a full-connection network layer driven by heterogeneous fusion, fusing the key features extracted from different scales, and generating a prediction probability distribution vector; an arrhythmia recognition model is constructed based on time-frequency attention enhancement, a heterogeneous fusion network and a full-connection network layer driven by heterogeneous fusion, a loss function is designed, and ECG signal arrhythmia classification is achieved. According to the method, accurate ECG signal classification in a complex scene is realized, and the robustness of the model to individual difference and noise interference is enhanced.

The invention discloses an emotion prediction method and system based on electrocardiographic and computerized electrical coupling. The method comprises the following steps: acquiring electrocardiosignal data and electroencephalogram data of a patient to be predicted; on the basis of a feature analysis algorithm, according to the electrocardiosignal data and the electroencephalogram signal data, determining electrocardiosignal feature data and electroencephalogram feature data; based on a coupling analysis algorithm, determining coupling characteristic data according to the electrocardiosignal data and the electroencephalogram signal data; and inputting the electrocardio feature data, the electroencephalogram feature data and the coupling feature data into a trained emotion prediction model to obtain an emotion prediction result of the to-be-predicted patient. Therefore, the accuracy and the stability of emotion prediction can be improved, accurate evaluation of the emotion state of the patient is realized, and a reliable basis is provided for emotion management and related medical intervention.

The invention discloses an electrocardio feature point positioning method based on double U-Net fused long and short range context information, comprising: preprocessing an electrocardio signal, including re-sampling to a unified sampling rate, and removing noise interference; segmenting the electrocardiosignal, wherein the electrocardiosignal is segmented into a single-cardiac-beat segment and a neighbor multi-cardiac-beat segment based on the R peak position; data enhancement operation is applied to the segmented cardiac beat segments; using double U-Net to extract features, including a first U-Net to extract local semantic features of a single cardiac beat segment and a second U-Net to extract global semantic features of a multi-cardiac beat segment; fusing the local semantic features and the global semantic features through a cross-level dual gating module to generate purified fused features; and positioning feature points of the electrocardiosignal based on the fusion features, wherein the feature points comprise the positions of P waves, QRS wave groups and T waves. According to the invention, the precision and robustness of feature point positioning are improved.

The invention discloses a portable state analysis method and system based on multi-mode electroencephalogram and electrocardio, and belongs to the field of state analysis. The portable state analysis method comprises the steps that electroencephalogram signals and electrocardio signals of a subject are synchronously collected; performing preprocessing and feature extraction on the signals to obtain electroencephalogram features and electrocardio features; according to the values of the electrocardio characteristics in different physiological states, constructing difference value characteristics and ratio characteristics representing differences between the states; the electroencephalogram features, the electrocardio features and the new construction features are fused, and multi-modal features are obtained; and finally, inputting the fusion features into a pre-trained table priori data fitting network model, and outputting a classification result or a quantitative prediction value for representing the state of the subject. According to the method, through systematic state comparison feature engineering and an efficient table data model, the accuracy, generalization ability and automation level of multi-modal physiological signal analysis are improved, and the whole scheme is realized based on portable equipment and is suitable for state evaluation of various non-laboratory scenes.

A deep neural network pre-training method for classifying electrocardiogram (ECG) data and a device for the same are disclosed. A method for training an ECG feature extraction model may include receiving a ECG signal, extracting one or more first features related to the ECG signal by inputting the ECG signal to a rule-based feature extractor or a neural network model, extracting at least one second feature corresponding to the at least one first feature by inputting the ECG signal to an encoder, and pre-training the ECG feature extraction model by inputting the at least one second feature into at least one of a regression function and a classification function to calculate at least one output value. The pre-training of the ECG feature extraction model may include training the encoder to minimize a loss function that is determined based on the at least one output value and the at least one first feature.

The invention discloses an SNN learning accelerator for ECG monitoring, which comprises a register configuration module, an asynchronous global control module, an asynchronous neural network control module, an asynchronous binary classification CNN module, an asynchronous quadruple classification SNN module, an asynchronous weight updating module and a memory management module, and is characterized in that firstly, a circuit corresponding to the accelerator constructs a double-layer dynamic network; layered dynamic power consumption management: selectively activating a high-precision secondary anomaly detection four-classification SNN network through a primary anomaly detection two-classification CNN network; secondly, the accelerator supports on-chip reasoning and efficient learning, and effectively eliminates ECG feature differences of different individuals on the premise of ensuring privacy security of user data; and finally, the circuit is controlled by using a pulse neural network and an asynchronous logic circuit, so that compared with a synchronous network, the power consumption is greatly reduced.

This disclosure relates generally to a method and system for analyzing an effect of age-related variations on the electrocardiogram (ECG). State-of-the-art methods have delved into supervised deep-learning approaches based on regression that does not reflect the relation between a chronological age (CA) and a biological age (BA) of the subject. The disclosed method involves an unsupervised learning approach with a three-step training strategy combining model training and deep ECG features clustering with a controlled initialization. A deep learning model is trained to obtain an age-informed convolutional autoencoder network by identifying one or more errors while processing the reconstructed ECG. Combining the CA and ECG, the trained deep learning model reveals the BA of the heart as a contributing biomarker for estimating the overall BA of the body.

A diagnostic electrocardiogram system employs an electrode lead system (40) to generate one or more electrode signals indicative of electrical activity of a subject's heart (10). The diagnostic electrocardiogram system also employs a diagnostic electrocardiograph (50) coupled to the electrode lead system (40) to communicate (e.g., list, display, and / or print) a subject electrocardiogram (20) and one or more diagnostic electrocardiograms (30) designated as morphologically matching the subject electrocardiogram (20), involving determining a probability that the diagnostic electrocardiogram(s) (30) represent an accurate diagnostic assessment of the subject electrocardiogram (20) based on a similarity between a morphology of the subject electrocardiogram (20) and a morphology of the at least one diagnostic electrocardiogram (30). The subject electrocardiogram (20) provides one or more interpreted information of ECG features derived from the electrical activity of the subject's heart (10) indicated by the electrode signal(s). The diagnostic electrocardiogram(s) provide one or more diagnosed information of ECG features derived from recorded electrical activity of a diagnostic heart (11).

This disclosure provides a model training method and an atrial fibrillation heartbeat detection method. The method includes: inputting a target electrocardiogram (ECG) segment into a feature extraction network to output ECG features; inputting the ECG features and a first positional encoding into an encoder to output ECG encoded features; repeating the following operations until the number of iterations matches the number of decoder layers to obtain an atrial fibrillation heartbeat detection model: inputting the ECG encoded features, the first positional encoding, the decoder embedding, and the second positional encoding into the y-th decoding layer to obtain the output of the y-th decoding layer, using this output as the input to the bounding box prediction head to output the offset, and using the offset to update the predicted bounding box; inputting the classification prediction head to output the heartbeat classification prediction result; performing loss calculations on the category label and bounding box label of the target ECG segment with the heartbeat classification prediction result and predicted bounding box to obtain classification loss values ​​and bounding box loss values, thereby adjusting the model parameters of the model to be trained to obtain the adjusted model.

The invention provides a small-sample multi-label electrocardiogram classification method which comprises the steps that a support set and a query set of an electrocardiogram are obtained, and each of the support set and the query set comprises a plurality of electrocardiogram signals; for each electrocardiogram signal in the support set and the query set, performing feature extraction in various different scales to obtain a plurality of extraction features corresponding to each electrocardiogram signal; carrying out attention enhancement on each extracted feature to obtain a corresponding enhanced feature, and then carrying out cross-scale fusion on all the enhanced features to obtain a fused feature map corresponding to each electrocardiogram signal; and aggregating the fusion feature maps corresponding to the support set into a class prototype, then calculating a similarity matrix between the class prototype and the fusion feature maps corresponding to the query set, and determining the class of each electrocardiogram signal in the query set according to the similarity matrix. The problem that electrocardiogram features cannot be fully extracted and accurately classified in the prior art is solved.

The present invention discloses an electrocardiogram (ECG) classification method, system, electronic device, and storage medium. The method includes: performing preprocessing operations on multiple acquired electrocardiograms to obtain a source domain dataset and a target domain dataset; inputting the source domain dataset into a preset class weighted width learning model to perform weight calculation to obtain a source domain connection weight; inputting the target domain dataset and the source domain connection weight into a preset domain adaptation class weighted width learning model to perform knowledge transfer to obtain a target domain connection weight; training the domain adaptation class weighted width learning model based on the source domain connection weight and the target domain connection weight to obtain an ECG classification model; inputting the target domain dataset into the ECG classification model to perform ECG classification to obtain a classification label. In an embodiment of the present invention, accurate and rapid recognition of various heartbeats of different subjects can be achieved, solving the problem of significant differences in ECG features between subjects.

The invention relates to an ECG synchronous feature recognition algorithm and device for four-electrode ICG impedance measurement and a medium, and the algorithm comprises the steps: employing a standard quadrupole method for configuration, and synchronously collecting an ICG impedance change signal and an ECG impedance electrocardiosignal of a human body; a multi-stage multi-order filtering processing method is adopted, the collected ICG impedance change signals and the collected ECG impedance electrocardiosignals are preprocessed, and then ICG de-noised signals and ECG de-noised signals are obtained; drawing an impedance change diagram and an impedance electrocardiogram on the basis of the obtained ICG de-noised signal and the ECG de-noised signal, and identifying main feature points of ICG and ECG in the same period and extracting feature parameters on the basis of the drawn impedance change diagram and impedance electrocardiogram; through the synergistic effect of signal synchronous acquisition, multi-stage filtering and same-period feature recognition, the adaptive capacity of the algorithm to the physiological signal time-varying characteristics is enhanced, the precision and stability of cardiac function evaluation are greatly improved, and the technical defects of an existing ICG measurement algorithm are effectively overcome.

The invention discloses a heart failure feature prediction system based on time-space fusion of heart sound and electrocardiosignals, which relates to the technical field of heart failure prediction and comprises a signal acquisition unit, a time-space fusion unit and a heart failure feature prediction unit. The signal collecting unit is used for collecting heart sound signals of a patient through a heart sound sensor making contact with the chest wall of the patient, collecting electrocardiosignals of the patient through an electrocardioelectrode attached to the skin of the patient, arranging the electrocardiosignals into current patient data and sending the current patient data to the time-space fusion unit, and the time-space fusion unit is used for obtaining and processing the current patient data. Performing multi-dimensional feature extraction on the current patient data to form a heart sound feature group and an electrocardio feature group, and performing space-time fusion processing by adopting a weighted average method to obtain a time fusion feature and a space fusion feature; the method can help to find potential risks in the early stage of diseases, so that doctors can formulate intervention measures in advance according to prediction results, thereby delaying disease progression, reducing the occurrence rate of heart failure, and improving life quality and prognosis of patients.

The present application belongs to the field of information security, and particularly relates to a two-stage WBAN authentication method based on iris features and ECG features, which comprises the following steps: collecting an iris image through an iris collector, and performing first-stage identification and authentication on the identity of a user through four stages of iris image preprocessing, iris image feature extraction and quantization, iris template making and iris template matching; after the first-stage authentication, collecting an electrocardiogram signal of the user using a sensor, and performing second-stage authentication on the identity of the user through four stages of ECG signal preprocessing, ECG feature extraction, ECG feature quantization and a fuzzy commitment-based key agreement mechanism. Only the user who passes the two-stage authentication is considered to have passed the authentication. The method adopts two-stage identity authentication, improves the accuracy of identity identification, and is beneficial to protecting the security of the WBAN and protecting the privacy of the user.

The invention discloses an ECG hash coding method and system based on sample similarity guidance, and belongs to the technical field of biological signal data machine learning, and the method comprises the steps: obtaining a data set containing a plurality of ECG signals, extracting features, and constructing an ECG feature matrix; for unsupervised learning, a feature space measurement matrix and a Hash space modulation matrix are introduced, and a generalized similarity preserving equation for aligning an Euclidean space and a Hamming space is constructed in combination with an ECG feature matrix and a to-be-solved Hash coding matrix; for supervised learning, introducing a semantic space measurement matrix on the basis of unsupervised learning, and constructing generalized similarity preserving equations which simultaneously align the Euclidean space and the Hamming space and align the semantic space and the Hamming space; converting the equation into a matrix optimization problem, and solving to obtain a Hash coding matrix; and the Hash coding matrix is used for coding the ECG signal, so that the Hash coding quality and the auxiliary diagnosis retrieval efficiency can be effectively improved.

This invention relates to an ECG arrhythmia classification method based on time-frequency attention enhancement and heterogeneous fusion, comprising: acquiring the original ECG signal; denoising and overlapping slicing the original ECG signal to obtain fixed-length ECG signal segments; performing time-frequency attention enhancement on the ECG signal segments to obtain time-frequency fused ECG features; constructing a heterogeneous fusion network to integrate channel information and location information into the time-frequency fused ECG features; constructing a fully connected network layer driven by heterogeneous fusion to fuse key features extracted at different scales and generate a predicted probability distribution vector; constructing an arrhythmia recognition model based on time-frequency attention enhancement, the heterogeneous fusion network, and the fully connected network layer driven by heterogeneous fusion, and designing a loss function to achieve ECG signal arrhythmia classification. This invention achieves accurate ECG signal classification in complex scenarios and enhances the model's robustness to individual differences and noise interference.

The present application relates to a kind of ECG feature wave segmentation method and FS-Net model, ECG feature wave segmentation method specifically includes the following steps: S1.ECG signal denoising;S2.Using Pan-Tompkins algorithm demarcation ECG signal R peak information, calculate RR interval information, then using sliding window to data normalization processing;S3.ECG signal waveform correction and feature enhancement: first, ECG signal is carried out smooth processing, then the QRS wave of ECG signal, P wave and T wave are corrected, finally feature enhancement is carried out;S4.FS-Net model is constructed;S5.ECG signal after processing in step S1-S3 is input into FS-Net model and is trained, realizes the segmentation to feature wave in ECG signal;S6.post-processing algorithm;S7.ECG signal segmentation: using the FS-Net model after training to the measured ECG signal is segmented.The present application is corrected and enhanced to ECG signal, effectively enhances the performance of P wave and T wave in electrocardiogram, improves the accuracy of feature wave detection.

A system and method for guiding medical device placement. The system includes a console, a portable acquisition unit for receiving intracavitary and surface electrocardiogram (ECG) signals, and connection hardware, including a sterile ECG adapter and cable for fluid-filled catheters or a sterile clip cable for guidewires. Processors display real-time ECG waveforms and, in response to user input, capture static snapshots at different insertion depths. These snapshots are displayed chronologically on a graphical user interface, forming a visual trajectory of the device. The system determines the device's location by detecting ECG features from each snapshot and comparing them to predetermined boundaries. A location indication is displayed for each snapshot, providing objective guidance for placement and confirmation.

The application discloses an ECG monitoring-oriented SNN learning accelerator, which comprises a register configuration module, an asynchronous global control module, an asynchronous neural network control module, an asynchronous binary classification CNN module, an asynchronous four-classification SNN module, an asynchronous weight update module and a memory management module. Firstly, the accelerator corresponds to a circuit to build a double-layer dynamic network, hierarchical dynamic power management is performed, and a four-classification SNN network of high-precision secondary abnormality detection is selectively activated through a binary classification CNN network of primary abnormality detection. Secondly, the accelerator supports on-chip reasoning and efficient learning, effectively eliminates the ECG feature differences of different individuals under the premise of guaranteeing the privacy and security of user data. Finally, the circuit uses a pulse neural network and an asynchronous logic circuit for control, and the power consumption is greatly reduced compared with a synchronous network.

The invention belongs to the field of machine learning, and discloses a class incremental learning method for ECG identity recognition. Comprising the following steps: preprocessing electrocardiosignals; constructing a memory bank, and introducing a similarity penalty term in sample screening based on class center approximation to ensure that finite memory samples cover the diversity of electrocardiogram waveforms; inheriting old model parameters to construct a new model, and performing joint training by using a memory bank and new data; a double constraint mechanism is introduced in the training process: 1, subspace distillation constraint: strictly aligning geometric structures of new and old model feature subspaces through SVD (singular value decomposition), Grassmann distance and KL divergence; and 2, new and old category separation constraint: utilizing self-attention to enhance prototype features and maximize the minimum distance from a new category to an old category prototype set. According to the method, the problems of disastrous forgetting and category confusion caused by ECG feature drift are effectively solved, and the recognition accuracy in a continuous learning scene is remarkably improved.

The invention relates to a transfer learning driving electrocardio double-state judgment and personalized course optimization system. According to the system, accurate judgment of the attention state in a learning scene is achieved by constructing a specific mapping model of electrocardio rhythm features and the attention state; based on attention early warning features in the electrocardiogram time sequence data, an attention load pre-judgment model is established, and prospective adaptation of a course is achieved; meanwhile, the system is optimized in real time by using an individual electrocardio baseline template library and an edge end, an attention judgment threshold is dynamically calibrated, and individual difference and scene interference are avoided. According to the technical scheme, through electrocardiosignal real-time monitoring and course dynamic generation, accurate perception of learning attention and personalized course optimization are achieved, the learning efficiency and adaptability are effectively improved, and the problems that a traditional course adjustment mode is single, and electrocardiosignal features are not fully used for learning cognitive assessment are solved.

This application proposes a classification model, construction method, and application of myocardial ischemia based on 12-lead ECG. By detecting the ST-T segment beat-by-beat changes using entropy domain, frequency domain, and Lyapunov domain analysis methods, spatiotemporal ECG feature parameters related to myocardial ischemia are obtained. A machine learning model for predicting myocardial ischemia is established, ultimately achieving optimal ECG feature selection related to myocardial ischemia. This addresses the problem that conventional 12-lead ECGs often lack clear ST segment or T wave features related to myocardial ischemia, severely affecting the sensitivity and accuracy of myocardial ischemia diagnosis. The method includes: acquiring a 12-lead ECG; calculating the ST-T segment sample entropy of the 12-lead ECG; converting the 12-lead ECG into a 3-lead ECG vector map; extracting the ST-T segment from the 3-lead ECG vector map and calculating its spatial and temporal feature values; and using a grid search method to select the optimal ECG features reflecting myocardial ischemia. Therefore, the sensitivity and accuracy of myocardial ischemia diagnosis are improved.

The invention discloses a fatigue recognition method based on brain-heart adaptive coupling learning, and belongs to the technical field of fatigue recognition, and the method comprises the steps: obtaining marked EEG features and ECG features; performing multi-source depth alignment on the EEG features and the ECG features to obtain multi-source information after feature alignment; performing dynamic coupling learning of local functions on the local EEG features and the ECG features to obtain a coupling mode of the local EEG features and the ECG features; acquiring a multi-source global coupling mode of the EEG features and the ECG features by adopting a coupling strength guided self-adaptive order raising strategy; and decoding the fatigue state of the multi-source global coupling mode by adopting a linear layer. According to the method, a multi-source self-adaptive coupling learning strategy is provided, efficient decoding of fatigue features is achieved, multi-level and high-level fatigue states can be accurately recognized, brain-heart cognitive interaction modes in different fatigue states are disclosed, and a new thought is provided for development of the multi-source coupling technology.

The invention provides an electrocardio detection method and device, electronic equipment and a storage medium. The method comprises the steps that an electrocardio signal of a target user and case text information of the target user are obtained; determining an electrocardio feature vector of the electrocardio signal; performing coding processing on the case text information to obtain a semantic vector; and according to the semantic vector and the electrocardio feature vector, obtaining a detection result of the electrocardio signal.

The invention discloses a biomedical signal noise reduction method and device based on multi-dimensional information fusion and a medium, and relates to the technical field of biomedical signal processing.The method comprises the steps that an original ECG signal, an original ACC signal and an original GYRO signal of a target user in a dynamic electrocardiogram monitoring scene are preprocessed; performing layered noise reduction on the preprocessed first ECG signal; and according to the preprocessed first ACC signal and the first GYRO signal, carrying out artifact suppression on ECG features extracted from the second ECG signal subjected to layered noise reduction, inversely mapping the target ECG features subjected to artifact suppression to the space of the second ECG signal, and determining a target ECG signal which is high in signal-to-noise ratio and contains a complete P-QRS-T wave group. According to the method and the device, the cooperative suppression of various noise types in dynamic electrocardiogram monitoring is realized, the clinical diagnosis value of the ECG signal is guaranteed, and the noise reduction effect and the signal availability rate of the ECG signal in a dynamic scene are improved.

The present disclosure provides a cuffless continuous blood pressure signal reconstruction method and system, relating to the technical field of blood pressure monitoring, comprising: acquiring synchronous PPG signals and ECG signals; using a trained attention time sequence network, extracting PPG feature vectors and ECG feature vectors from the PPG signals and ECG signals respectively, and performing multi-modal fusion on the PPG feature vectors and ECG feature vectors, and based on the fused feature vectors, reconstructing a cuffless continuous blood pressure signal; wherein the feature vector extraction increases an information screening machine and a time sequence learning machine between the encoder and the decoder, which are respectively used for learning a channel probability distribution vector and a time sequence weight vector, guiding the decoder to decode high-level semantic features and the time sequence relationship of PPG and ECG before and after; the present disclosure proposes an attention time sequence network based on multi-modal fusion, and realizes the reconstruction of continuous blood pressure signals by analyzing the photoplethysmogram signal and the electrocardiogram signal.

The present application relates to the technical field of computer vision deep learning and medical data processing, in particular to an ECG data classification method based on RNN and Mamba sequence model, first, the original ECG data is encoded using an encoder to extract its preliminary features, and position coding is added.Next, on the one hand, based on the time sequence of ECG data, the time domain features of ECG data are extracted through RNN and Mamba module; on the other hand, based on the characteristics of ECG data wave, the frequency-time information and amplitude-frequency two kinds of frequency domain features are obtained from the ECG features through continuous wavelet transform CWT and discrete Fourier transform DFT; then the two kinds of frequency domain features are added to the time domain features to realize the enhancement effect of the time domain features.After that, a residual network ResNetECG is used to classify the enhanced time domain features.The present application improves the accuracy of the classification result.

This disclosure relates generally to a method and system for analyzing an effect of age-related variations on the electrocardiogram (ECG). State-of-the-art methods have delved into supervised deep-learning approaches based on regression that does not reflect the relation between a chronological age (CA) and a biological age (BA) of the subject. The disclosed method involves an unsupervised learning approach with a three-step training strategy combining model training and deep ECG features clustering with a controlled initialization. A deep learning model is trained to obtain an age-informed convolutional autoencoder network by identifying one or more errors while processing the reconstructed ECG. Combining the CA and ECG, the trained deep learning model reveals the BA of the heart as a contributing biomarker for estimating the overall BA of the body.