PLMPFS: Predictive Learning with Polynomial Features and SMOTETomek Balancing based Heart Disease Prediction

Accurate and reliable heart disease prediction remains a difficulty in numerous in medicine, where the ability to correctly identify and act upon predictive insights can significantly influence outcomes and efficiencies. Traditional predictive models often struggle with class imbalance, leading to biased predictions that favor the majority class. To address these challenges, we developed a Predictive Learning Model (PLM) that incorporates several advanced techniques, including polynomial feature engineering and SMOTETomek for data balancing. The model is built using a deep learning framework with multiple layers to fully understand the data's intricate nonlinear relationships. Model performance was evaluated using key classification metrics, including confusion matrix analysis and Receiver Operating Characteristic (ROC) curve analysis. The PLM demonstrated significant improvements in classification accuracy, achieving an area under the ROC curve (AUC) of 0.97. It also showed robust performance across various metrics, with a precision of up to 98%, and sensitivity and specificity rates approaching 99%. These results highlight an improvement in the model's outcome prediction capabilities accurately, even in the presence of class imbalance.