Predicting Heart Failure Survival with Machine Learning: Assessing My Risk

Abstract

This study investigates the application of machine learning techniques for heart disease prediction using a comprehensive dataset of 918 patients. The research employs multiple algorithms, including Logistic Regression, Random Forest, Support Vector Machine (SVM), and Neural Networks, to develop predictive models based on 11 clinical features. The dataset, compiled from five independent sources, underwent thorough preprocessing and was split into training (70%) and test (30%) sets. Model performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC metrics. Results demonstrate consistently high performance across all models, with the SVM achieving the highest overall performance (accuracy: 88.41%, precision: 89.76%, recall: 90.85%, F1-score: 90.30%, ROC-AUC: 94.97%). Key predictors identified include age, maximum heart rate, and ST depression (Oldpeak). The study's findings have significant implications for clinical practice, offering the potential for rapid, objective heart disease risk assessment. The consistent performance across different model architectures provides flexibility for implementation in various healthcare settings. Limitations include potential data collection variability and gender imbalance in the dataset. Future research directions include developing more sophisticated neural networks, incorporating additional data types, and conducting prospective studies to validate model performance in real-world clinical settings. This research contributes to the growing body of evidence supporting the use of machine learning in medical diagnostics. The developed models enhance early detection and risk stratification of heart disease, potentially improving patient outcomes through timely interventions.