Predicting Heart Disease Risk with Machine Learning Integrating Family History and Medical Data

Heart disease ranks as a primary cause of illness and death globally, influenced by multiple factors. Early detection of heart disease risk is vital for successful intervention and prevention efforts. This project investigates various machine learning methods for predicting heart disease, focusing particularly on including family history as a significant predictor. Although conventional models typically consider factors like age, cholesterol, blood pressure, and lifestyle, our approach underscores the importance of family history—a known yet frequently overlooked element in heart disease risk assessments. The dataset used in this research comprises patient records featuring both medical history and family lineage information, offering a holistic perspective on genetic risk factors as well as clinical indicators. By integrating family history into machine learning frameworks such as logistic regression, random forests, and naïve Bayes, we aspire to enhance the precision and reliability of heart disease predictions. These models were subjected to a range of performance metrics, such as accuracy, precision, recall, and AUC-ROC scores. Throughout this study, we emphasized evaluating how the inclusion of family history, alongside traditional risk factors, affects the predictive power of various models. Findings suggest that models utilizing family history yield superior results compared to those that depend exclusively on clinical aspects, thereby providing more tailored risk evaluations. This improved predictive ability has promising implications for early diagnosis, personalized medicine, and effective healthcare strategies.