Application Research Of Human Blood Pressure Prediction And Abnormal Heart Rhythm Classification Based On Machine Learning

Blood pressure and electrocardiogram are two important indicators of human health in clinical testing.Wrist blood pressure measurement and electrocardiogram observation are the routine diagnostic methods.The diagnostic process depends on professional medical staff.The controllability of the measurement time is poor,and it is easy to exacerbate the shortage of medical resources.This paper uses machine learning method to predict blood pressure and classify arrhythmia,which propose a prediction model of blood pressure based on support vector machine and a classification model of arrhythmia based on adaptive boosting algorithm.The main research contents are as follows.Firstly,the physiological signals of electrocardiogram and pulse wave as well as their feature processing methods are deeply analyzed in this paper.As an important physiological signal,electrocardiogram and pulse wave have periodic and typical waveform characteristics.Different feature selection and extraction methods analyze human physiological data from different perspectives.Secondly,this paper proposes a blood pressure prediction model based on support vector machine regression algorithm.According to the sampling frequency of data,a reasonable expansion method of data is designed,and the correlation between physiological characteristics and blood pressure is analyzed by unsupervised and supervised feature selection methods.Feature selection and combination are carried out.The mapping relationship between physiological characteristics and blood pressure is studied by using support vector machine of non-linear kernel function to achieve accurate blood pressure prediction.Thirdly,a classification model of arrhythmia based on adaptive boosting algorithm is proposed.Data set of arrhythmia is defined according to the standards of the Association for the Advancement of Medical Devices and inter-patient scheme.Five electrocardiogram features are extracted from different angles of time domain and transform domain after electrocardiogram signal is filtered and segmented.An adaptive boosting algorithm using classification tree as base classifier is used to learn the differences among various cardiac arrhythmias and achieve accurate multi-classification of cardiac arrhythmias.Finally,the specific experimental environment and evaluation indexes of blood pressure prediction model and arrhythmia classification model are introduced,and the experimental results are analyzed to verify the validity and accuracy of the model.