The Efficacy Of Prognostic Prediction Model In Patients With Chronic Heart Failure

Chronic Heart Failure(CHF)is the last stage in the development of various heart diseases.CHF is considered to be a major public health challenge in the cardiovascular field in the 21 st century.This paper combines the classic survival analysis method with the popular machine learning algorithm to build a predictive model for the prognosis of CHF patients.Compared with the traditional survival analysis method,the research results in this field are enriched by the application of research on the survival problems of CHF patients,and the prediction model with high accuracy is constructed.In this paper,the CHARM project is used for the data set of patients with chronic heart failure.First,the data set is cleaned and have a feature engineering to complete the variable screening work,and then the traditional survival analysis methods such as Kaplan-Meier curve and Cox proportional hazards regression model are used to analyze the risk factors.Then,the follow-up time is processed separately and the reliability of the follow-up time is analyzed by using the random forest algorithm.Finally,four machine learning algorithms are used to build a classification model and a regression model to predict the survival outcome and survival time of patients and then compare the model performance.After survival analysis,we finally get 30 characteristics that have a greater impact on the outcome.Body weight is a protective factor,higher body weight is more conducive to patient survival.The overall survival rate of women is slightly higher than that of men.Older patients are at higher risk,and those under 60 have a much higher survival rate than those over 60.The machine learning algorithm is used to predict the prognosis of patients.The final experiment shows that the follow-up time has a significant impact on the prediction accuracy of the survival outcome.The AUC value of the XGBoost and Light GBM algorithms for the prediction of the survival outcome is as high as 0.96,and the prediction efficiency is satisfactory.Besides,patient’s weight、age and ejection fraction are the top three influencing factors.And the Light GBM algorithm for predicting patient survival time obtaines the best predictive performance with an error of 5 days.The predictive model constructed in this paper can provide a highly reliable prediction for the prognosis of patients with chronic heart failure,and the conclusions obtained from the analysis can provide valuable information for clinical treatment,which is of great significance for clinical research.