Comparative Study on Prediction of Survival Event of Heart Failure Patients Using Machine Learning and Statistical Algorithms

In a pressing global health concern with substantial morbidity and mortality rates, accurate survival prediction is paramount for informed decision-making and enhanced patient well-being. This study presented a comparative investigation aimed at predicting the survival events of heart failure (HF) patients through the utilization of both machine learning and statistical algorithms. A comprehensive dataset drawn from Allied Hospital and the Faisalabad Institute of Cardiology, Faisalabad, Pakistan, was used. The Synthetic Minority Over-Sampling Technique (SMOTE) was employed on the data to rectify the imbalance, and a notable improvement was observed. To ascertain significant variables, statistical methods (Mann-Whitney and Chi-Square) were compared with machine learning-based feature selection to identify pivotal features for survival prediction, namely ejection fraction and serum creatinine. Remarkably, on final training with these features, the Random Forest Classifier emerges as the top-performing model, boasting an accuracy exceeding 90%. These findings hold the potential to substantially enhance patient prognosis, management, and outcomes, consequently alleviating the strain on healthcare systems. Doi: 10.28991/SciMedJ-2023-05-02-01 Full Text: PDF