Identification Of Malignant Arrhythmias Based On Spatiotemporal Analytic Modeling Of Pulse Signal

Malignant arrhythmias are a serious cardiovascular disease.Rapid and accurate detection of malignant arrhythmias using information technology is very important for the treatment of patients.The recognition of malignant arrhythmias based on the quantitative analysis of pulse signals is the key research co ntent in the field of intelligent diagnosis of diseases.Because the shape and period of pulse signals are changeable,there are still some problems to realize the quantitative description of its waveform changes and the recognition of malignant arrhythmia s.Based on this,this study focuses on the modeling,quantitative description of pulse signal and its application in the identification of malignant arrhythmia.The main work is as follows:1.Aiming at the noise existing in the pulse signal,an improved adaptive noise set empirical mode decomposition method is used to filter it.After multi-layer decomposition of the original pulse signal,components containing the main components of the pulse signal are selected to reconstruct the signal to realize signa l filtering.The simulation results show that compared with the conventional filtering method,the improved adaptive noise set empirical mode decomposition method preserves the waveform characteristics of the original signal better,and effectively removes the noise in the signal.At the same time,the dynamic coefficient of variation method is proposed for the period segmentation of pulse wave.The experimental results show that the accuracy of period segmentation is more than 99%,and the proposed method can effectively realize the period segmentation of pulse signal.2.Aiming at the problem that the pulse signal has a large range of dynamic changes in shape and period,a pulse signal spatio-temporal analytical modeling technology is proposed.According to the forming mechanism and the time domain waveform of the pulse signal characteristics,build a quantitative analysis model,and use the conditional constraints are applied to solve the nonlinear programming method to the model parameters of the model pa rameters to solve a planning process,realize the quantitative analysis of pulse signal,and compares the different structure model,the modeling result with the international standard physiological signal pulse signal for experimental data in the database,different numbers of Gaussian and Lognormal basis functions are used to model the pulse wave data of healthy people,extreme bradycardia,extreme tachycardia,ventricular tachycardia and ventricular flutter fib.The results show that the pulse wave decom position based on three lognormal basis functions is the best.3.Aiming at the problem that the pulse rate and waveform characteristics of the interference segment are missing in the actual acquisition,a prediction method for the pulse rate and waveform of the interference segment based on the combination of spatiotemporal analytical modeling and neural networks is proposed.Combined with the pulsation characteristics of pulse waves,the pulse signal without interference in the database is used to construct the pulse rate sequence and waveform of the interference segment based on the characteristic parameters of pulse temporal and spatial analytic modeling.The pulse rate and waveform of the interference segment are predicted by combining the Elman neural network.However,the global search of the Elman neural network has the defect of blindness.The Sparrow search algorithm(SSA)is used to realize the adaptive optimal search of single neural network weights.The simulation results show that the spatio-temporal analytic model-the Elman neural network model-based on SSA optimization can effectively improve the accuracy of the pulse rate and waveform prediction in the interference segment.4.Aiming at the problem of accurate identification of malignant arrhy thmias,a malignant arrhythmia identification technique is proposed based on the mechanism of changes in pulse rhythm and hemodynamic information caused by malignant arrhythmias in humans.Firstly,the pulse signals of patients with different types of malignant arrhythmias were modeled to obtain the model parameters.Secondly,according to the significant change parameters in the analytic model of pulse signal,finally,probabilistic neural network and random forest machine learning algorithm are used to realize the intelligent recognition of malignant arrhythmia.The experimental results show that the performance of random forest classifier is the best in arrhythmia recognition,with an average accuracy of 97.083%.At the same time,compared with the deep learning intelligent recognition method directly using beat segmentation,the overall accuracy of recognition is only 83.20% by combining convolutional neural network and bidirectional long and short time memory network.Therefore,temporal and spatial anal ysis modeling of pulse is more effective for malignant arrhythmia recognition.