Congestive Heart Failure Prediction Based On Feature Selection and Machine Learning Algorithms

Obesity is one of the main public health problems worldwide that present an association with many medical conditions. Therefore, obesity and its comorbidities such as congestive heart failure are a subject of investigation in many areas. One of these areas is the application of computer science for solving problems in the medical field. This paper shows the evaluation of machine learning algorithms for the prediction of congestive heart failure disease. A dataset of 412 medical summaries was used. It has 342 features with information of diseases, medications, treatments among others. Due to the extensive number of features, feature selection techniques such as Chi Square and Mutual Information were applied. The results showed that the highest precision metric of 0.94 was obtained with the Random Forest algorithm with 23 best features selected by the Mutual Information technique. Regarding AUC, the algorithm the highest value of 0.89 was obtained with the Naïve Bayes algorithm with 24 best features selected by the Chi Square technique.