Early Detection Of Coronary Heart Disease Based On Ensemble Learning Algorithm

Cardiovascular disease is the significant harm to the global public health problem,as a kind of common chronic cardiovascular disease,coronary heart disease the incidence,mortality is rising year by year,and showed a trend of getting younger over,rising to the harm of human society.Inhibition of coronary heart disease morbidity and even death is the basic way of early warning and diagnosis,research how to in the early clinical noninvasive,nondestructive identification of individuals at risk of coronary heart disease has important clinical significance and significant social and economic benefits,the application of machine learning algorithms provide may solve the above problem.The purpose of this paper is to study the potential application value of integrated learning algorithm for the diagnosis of coronary heart disease without invasive and noninvasive screening.For this purpose,this paper systematically studies the development trend of the risk early warning model of coronary heart disease and the application of machine learning algorithm to diagnose coronary heart disease.Relying on the national natural science fund project-based on the electrical and mechanical activity on the surface of variability of coronary heart disease(CHD)distribution entropy characteristics research funding,in patients with established coronary heart disease(CHD)electronic medical record large data sets.System science research integrated learning algorithm model process programming,algorithm implementation,for early screening for diagnosis of coronary heart disease the individual approach and analysis,a new algorithm in this paper,the main work done is as follows:(1)To participate in the establishment of an electronic medical record medical data set for patients with coronary heart disease in shandong province.Coronary heart disease risk factors based on previous studies,combined with the practice of cardiology,clinical expert experience and advice,field acquisition in jinan above hospital subjects ecg,heart sounds,such as pulse wave waveform data,the integration of basic information,clinical symptoms and physiological and biochemical indexes,such as data.(2)An ensemble learning algorithm model for early screening diagnosis of CHD was proposed.Compared with the traditional machine learning algorithm,the accuracy,sensitivity and specificity were 93.92%,96.40%and 85.71%,respectively,based on the risk factor database of CHD.By using SMOTE algorithm,the sample size of normal people was matched to the sample size of CHD group,and the heterogeneous integration learning algorithm was constructed,and the accuracy,sensitivity and specificity were 96.28%,97.81%and 94.65 respectively.(3)Through statistical analysis of the clinical data sets of CHD risk factors,the characteristics of significant differences were found.At the same time,the use of three kinds of homogeneous RF integrated learning algorithms,GB,XGB do importance 'ratings of the characteristics of the data set,with the importance of hypertension gradeas a baseline,were selected for a total of 16,22,18 characteristics.In the two methods,hypertension,age,total cholesterol,glucose,high-density lipoprotein,triglyceride and other characteristics are common characteristics,and are common risk factors for coronary heart disease.