Research On Diagnosis Model Of Essential Hypertension Complicated By Coronary Heart Disease Based On Machine Learning Algorithm

Objectives: Coronary atherosclerotic heart disease is the most common cardiovascular disease in clinical practice,and essential hypertension is one of the most important risk factors for coronary heart disease.If it is not controlled early and timely,it will easily progress to heart failure.Early diagnosis of coronary heart disease is conducive to timely treatment and improved prognosis.The conventional clinical diagnosis methods include: coronary artery contrast agent,electrocardiogram,cardiac ultrasound,etc.,but these diagnostic methods cannot be used the early diagnosis of coronary heart disease,the cost is relatively high.At present,the situation of misdiagnosis and missed diagnosis of coronary heart disease in China is still serious.Machine learning is an emerging artificial intelligence discipline that has been widely used to assist doctors in making objective predictions and judgments.Starting from "medical data + machine learning",it provides an auxiliary diagnosis method for essential hypertension complicated by coronary heart disease.Methods: Taking the patients with essential hypertension,coronary heart disease and essential hypertension in the medical big data platform of Chongqing Medical University as the research object,the general information,vital signs,past history,and laboratory examination indicators of the patients are collected from the platform.Use single factor analysis,LASSO,Logistic regression analysis to filter the difference indicators,use the grid search algorithm to search for the optimal parameters of the machine learning algorithm,and select the four machine learning algorithms of RF,XGBoost,CART,and BPNN to construct the classification model.Evaluate model performance with a series of indicators such as accuracy,sensitivity,specificity,area under the curve,etc.Results: 2487 patients with essential hypertension complicated by coronary heart disease were selected as the study group,and 3904 patients with essential hypertension were selected as the control group.The study group and the control group included a total of 58 indicators.The univariate analysis screened out 45 indicators.The 45 indicators entered the LASSO regression to screen out 23 indicators.The Logistic analysis screened out18 indicators from the 23 indicators and included 18 indicators.4 kinds of machine learning models,and the Logistic model as a control at the same time.In the training set,the AUC of RF,XGBoost,CART,BPNN,and Logistic models are 0.969,0.948,0.801,0.848,0.784.In the test set,the RF,XGBoost,CART,BPNN,and Logistic models have AUCs of 0.885,0.892,0.781,0.788,0.771.The performance of the RF and XGBoost models in the training set and the test set is better than the other three models,and the XGBoost model is slightly better than the RF model.The auxiliary diagnostic tools developed based on the better-performing XGBoost and RF models can meet the needs of assisting doctors in the early diagnosis of essential hypertension complicated by coronary heart disease.Conclusion: The machine learning diagnosis model established by using medical data and machine learning algorithms has a good auxiliary diagnosis function for essential hypertension complicated by coronary heart disease,and can provide certain help to the early diagnosis of essential hypertension complicated by coronary heart disease.