The World Health Organization attributes 31% of the world population mortality statistics to heart-related diseases making it one of the deadliest diseases in the world. Even though many heart conditions are curable, a timely and precise diagnosis significantly raises the chances of survival. However, procedures like blood tests, ECG, and angiography are incredibly costly and depend heavily on human precision. This study explores how feature selection impacts the accuracy of machine learning (ML) classifiers in diagnosing heart disease. Specifically, it evaluates the effectiveness of two widely used optimization techniques, Genetic Algorithms (GA) and Particle Swarm Optimization (PSO), in selecting the most relevant features. These methods are tested alongside various classifiers, including Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree (DT), Naïve Bayes (NB), Random Forest (RF), and Multi-Layer Perceptron (MLP). The research is based on datasets from the UCI repository, which include data from Cleveland, Hungarian, Switzerland, and Long-Beach-Va. Various performance metrics were used in the evaluation of the model including accuracy, sensitivity, specificity, and F-score. The results showed that with Accuracy and computational complexity, the GA faired slightly better or in some cases roughly on par with the PSO. The contribution of the study focuses on the integration of intelligent decision support systems with feature selection methods.