Abstract. Heart failure is a grave and progressive illness. It is associated with multiple risk factors such as age and other possible factors. Accurately identifying and quantifying these risk factors is critical to developing personalized prevention and treatment strategies. However, it is difficult for common patients to predict heart failure without the professional diagnosis of doctors, and relying on human resources to predict heart failure is more subjective. This study proposes a method to evaluate and predict the factors related to heart failure based on multiple body indicators by using random forest algorithm. The random forest prediction model was applied to assess the correlation between multiple medical indicators such as creatinine phosphokinase (CPK), serum creatinine (SCR), ejection fraction (EF), age and heart failure. CPK was found to be the most associated with heart failure. In addition, increasing follow-up period can also effectively monitor heart failure progression in patients. In this high dimension prediction problem, the prediction effect of random forest is better, and the overall accuracy is higher. The approach used in this research is important for forecasting the risk of heart failure, enhancing the survival rates of patients, and alleviating the burden on healthcare systems.
Read full abstract