Detection And Analysis Of Hyperlipidemia Based On Machine Learning

Hyperlipidemia is an important inducement leading to a variety of cardiovascular and cerebrovascular diseases,and there are many complications resulting from it.Because of its insidiousness,most diseases do not have typical symptoms,and it is easy to delay the most effective prevention and treatment time,causing great harm to people’s physiological functions.Among them,high cholesterol concentration is manifested as hypercholesterolemia,which leads to overall abnormal blood lipid levels.Accordingly,cholesterol concentration value is the index that detects hyperlipidemia very important.At present,most of the commonly used blood lipid detection methods are measured by hospital biochemical detection,which not only causes the feeling of pain to the tested subjects,but also increases the risk of infection of the acupuncture site,with poor real-time performance.Bioelectrical impedance technology,as a non-invasive detection technology,can well distinguish different electrical signals reflected by different tissues and organs of the tested object,providing a new idea for non-invasive lipid component detection research.Existing studies have shown that there is a certain linear relationship between cholesterol concentration and electrical impedance,but the experiment was carried out in vitro with a mixture of cholesterol solution,which has a certain gap in complexity and anisotropy compared with human body.In this paper,a novel non-invasive continuous monitoring scheme of cholesterol concentration based on bioelectrical impedance spectroscopy is proposed to assist medical treatment in the detection of hyperlipidemia.According to the principle of bioelectrical impedance measurement,a human body impedance spectrum detection platform based on embedded system was designed to detect non-invasive cholesterol concentration.The platform is mainly composed of signal excitation module,signal conditioning module,AD acquisition module,etc.The signal excitation module is controlled by a microprocessor to generate a sinusoidal current excitation signal of 100 k Hz-300 k Hz.The conditioning circuit carries out differential amplification and bandpass filtering on the output signal of the detecting electrode.The amplitude-phase detection module gets the amplitude-phase and phase-angle parameters of the electrical impedance of the human arm,and then converts them into digital signals that can be recognized by the microcontroller.The schematic diagram of each module is designed experimentally,and the simulation is carried out to prove the correctness of the detection platform.Machine learning algorithms in disease risk prediction of advantage than traditional forecasting method,to reduce the individual difference,of impedance spectrum relation model is proposed in this paper to join and body composition analyzer correlation between higher body composition index optimization method,the prediction model of hyperlipemia comprehensive data set for data preprocessing,feature selection,and the characteristics of the principal component(PCA)feature dimension reduction engineering processing,get a characteristic variables linear independent of each other between data sets,using the Lasso regression algorithm of machine learning to build relationship model of bioelectrical impedance spectroscopy and cholesterol concentration,Implementation of cholesterol concentration prediction,prediction results and the testing results of the biochemical correlation was 0.79,the average relative error is 10.47%,the measurement system and the regression model to a certain extent,noninvasive detection of the human body cholesterol level,lipid and correctly judge the presence of abnormal,real-time monitoring of the patients with dyslipidemia and medical tests have very good auxiliary function.