Heart failure prediction based on random forest algorithm using genetic algorithm for feature selection

Abstract

A disorder or illness called heart failure results in the heart becoming weak or damaged. In order to avoid heart failure early on, it is crucial to understand the causes of heart failure. Based on validation, two experimental processing steps will be applied to the dataset of clinical records related to heart failure. Testing will be done in the first step utilizing six different classification algorithms, including K-nearest neighbor, neural network, random forest, decision tree, Naïve Bayes, and support vector machine (SVM). Cross-validation was employed to conduct the test. According to the results, the random forest algorithm performed better than the other five algorithms in tests employing the algorithm. Subsequent testing uses an algorithm with the best accuracy value, which will then be tested again using split validation with varying split ratios and genetic algorithms as a selection feature. The value generated from testing using the genetic algorithm selection feature is better than the random forest algorithm alone, which is recorded to produce an accuracy value of 93.36% in predicting the survival of heart failure patients.