Clinical Decision Support using Machine Learning and Deep Learning for Heart Disease Risk Assessment

Abstract

Cardiovascular diseases remain the leading cause of death globally, underscoring the importance of early risk detection. A recent strategy leverages machine learning to estimate the likelihood of heart disease by analyzing structured medical data. This process begins with organizing raw data, then extracting meaningful features from it. Predictive models are subsequently trained using clear patterns found in the data, and their performance is rigorously evaluated to assess effectiveness. Several methods—Logistic Regression, Random Forest, SVM, XGBoost, and Neural Networks—were compared despite their differing methodologies. Performance was assessed across five key metrics: accuracy, precision, recall, F1-score, and ROC-AUC. While simpler models showed limitations, ensemble techniques and neural networks demonstrated superior capability in identifying complex indicators of heart disease risk. Such machine learning approaches hold potential for earlier diagnosis, serving as supportive tools for clinicians during patient assessments. Though not flawless, these systems could offer valuable assistance in practical healthcare environments.