Research On Sleep Staging Method Based On BCG And Respiratory Signal

In today’s fast-developing society,more and more adults are troubled by sleep problems at night,and the state of each stage during sleep can provide an effective reference for doctors to evaluate sleep quality and diagnose sleep-related diseases.Ballistocardiography is a human body physiological information monitoring technology that does not need to be contacted and is not restricted.During sleep,the BCG signal can be used to collect the force of the human body surface due to the heartbeat and respiration,so as to collect the physiological parameter information during the sleep period.Effectively extracting the available heartbeat signal components and breathing signal components from the BCG signal is the prerequisite for its use in staging the various stages of sleep.This article mainly decomposes and reconstructs BCG to accurately extract and recover the heartbeat signal and breathing signal,and carry out corresponding research on the technology of sleep stage staging.First of all,this article introduces the physiological basis and time-domain waveforms of BCG signals and ECG,and analyzes the criteria for staging each stage of sleep.Screen the available signals for the original BCG signals in different states,use wavelet decomposition denoising method to denoise the interference that is easy to appear in the heartbeat signal,and use the double slope R peak detection algorithm to complete the location of the R peak position in the heartbeat signal extract.Then,the BCG signal after state screening is decomposed and reconstructed.Since the BCG signal contains multiple components such as heartbeat,respiration,and environmental interference,the selection of components after decomposition will affect the reconstruction result,so this paper proposes improved empirical wavelet transform algorithm.When directly using empirical wavelet transform to process BCG signals,add the step of boundary optimization and integration of the spectrum after adaptively dividing the boundary to obtain accurate respiratory signals and heartbeat signals.At the same time,an improved depth volume is proposed.The product generates a confrontation network to establish the connection between the two to construct the mapping model between the heartbeat signal and the original ECG signal.Finally,analyze and extract the heart rate variability and respiratory variability in the heartbeat signal and respiratory signal,use particle swarm algorithm and genetic algorithm to train the support vector machine to optimize the support vector machine,analyze and compare the sleep staging based on the heart rate variability Support vector machine model and sleep staging support vector machine model based on heart rate and respiratory variability,combined with parameter optimization algorithm to obtain the optimal support vector machine model,the results show that this article uses heart rate,respiratory variability characteristics to train the optimal support vector machine model The accuracy rate of sleep stage four stages can reach 76.38%,and the accuracy rate of sleep stage six stages can reach 71.11%.