Kernel-lasso feature expansion method: boosting the prediction ability of machine learning in heart attack

Abstract Heart attack needs to cause a high degree in modern society, which is now at the top of the list of most common diseases. One-third of all deaths in the world are caused by heart disease, and in our country, hundreds of thousands of people die of heart disease every year. If people can predict heart attack at an early period, they may prevent and cure the potential risk. Therefore, one novel and efficient prediction model is needed. Based on machine learning, we design one novel method in feature selection called kernel-lasso feature expansion to increase the prediction ability. This method considers the potential influence by the curse of dimension and lack of features, which can increase the dimension of the existed data and select the effective features by lasso regression. Compared with other feature selection methods such as Lasso and step regression, the novel method increases the predicted ability of gradient boosting machine, DNN, and SVM models, while SVM with kernel-lasso feature expansion achieves the best performances in Accuracy (0.84(%95Cl:0.78-0.89)).