Heart disease, a leading cause of death worldwide, necessitates early detection and treatment, which can be facilitated by accurate predictive models. However, the accuracy and generalizability of predictions are impacted by the difficulties of high dimensionality and data imbalance.