Research On Association Analysis Between Human Gait And ECG With Machine Learning

Orderly movement of human body’s gait posture changing with time,which is the most essential and ubiquitous movement in people’s daily life.The coordinated movement patterns of various parts of the human body are combined to form the behavioral characteristics of walking.The heart is also a vital organ of the human body,and the ECG signal reflects the details of the heart’s periodic beats.Both are closely related to health and are widely used in various fields,especially in the field of sports and medical rehabilitation.Therefore,the study of the relationship between gait characteristics and ECG signals is particularly important.At the same time,however,the versatility of gait and ECG brings great challenge to the study of the relationship between them.Based on the study of gait characteristics and ECG information,a new regression model is established.The main research contents of this paper are summarized as follows:(1)Design 5 kinds of scene experiments,walk or run in different states,capture the three-dimensional coordinates of each part of the body while walking through the motion acquisition system,extract features,take each gait cycle as time point,and extract gait features.Finally,among the extracted features,ten gait cycles are one segment,and take average of ten gait one segment.(2)The position of the R wave in the ECG signal is detected by the method of automatically adjusting the threshold,and the accuracy of the positioning is ensured by the backtracking search and the refractory period detection algorithm.In this paper,the initialization threshold of the algorithm is improved to prevent the influence of the interference signal on the threshold and make the threshold more reasonable.At the same time,when the ECG refractory period is set,the refractory period is automatically adjusted with the change of the RR interval,so that the algorithm becomes more adaptable.(3)A regression model based on regularized limit learning machine(RELM)is proposed,which analyzes the correlation between ECG information and human gait,and analyzes the eight aspects of the collected real ECG signals and gait characteristics.A data set is constructed with a small subset of every ten gait cycles.Experimental results show that human gait does have potential relevance,and gait features based on RELM regression model can be used for RR interval prediction.In addition,increasing the number of gait features can significantly improve predictive performance;which we have found demonstrates that the faster the gait speed,the better the predictive performance.The study found that the predictions of ECG information for walking and running were slightly different.