This study presents an in-depth experimental analysis of a proposed cardiovascular disease (CVD) detection system, developed using Matlab2022b on a high-performance computational setup. A critical preprocessing phase involved outlier removal from the Coronary Heart Disease (CHD) dataset, enhancing data reliability and enabling effective partitioning into training, validation, and testing subsets (70:15:15). The core of the system is a multilayer perceptron artificial neural network (MLP-ANN) trained with Bayesian Regularization (BR), selected for its superior generalization and adaptability. Through rigorous training over 1,000 epochs with 20 hidden layers, the model achieved a notable accuracy of 84.62%. Visual analyses, including training progress graphs, gradient descent behavior, and error histograms, demonstrate the model’s efficiency in learning complex patterns and minimizing prediction errors. Comparative performance evaluation against state-of-the-art methods highlights the proposed system’s competitive edge, underscoring its robustness and potential for practical application in CVD diagnosis.