Cardiovascular diagnostic detection using multilayer perceptron approach

Cardiovascular disease (CVD) affects the heart and blood vessels, causing symptoms like chest pain, shortness of breath, and dizziness. CVD is worsened by unhealthy habits and conditions such as high blood pressure, high cholesterol, and diabetes. Early identification and treatment are crucial. Recent advancements in machine learning (ML) and deep learning (DL) offer new methods for predicting CVD and improving patient outcomes. Current CVD prediction methods are limited by small sample sizes, outdated techniques, and lack of diverse data, hindering their reliability and effectiveness. Using a publicly available cardiovascular dataset comprising 70 000 patient records and 13 features, seven machine-learning (Random Forest, KNN, Naive Bayes, Gradient Boosting, SVM, SGD-SVM, Logistic Regression) and two deep-learning (MLP, CNN) models were evaluated under 5-fold cross-validation. The proposed MLP (3 hidden layers × 300 neurons, tanh activation, α = 0 . 001 ) achieved 94.5 % accuracy, 0.95 precision and 0.95 recall, outperforming all other methods. This study aims to evaluate the performance of various ML and DL algorithms for predicting CVD. We conducted an exploratory data analysis during the feature engineering phase. We applied seven ML algorithms, including Random Forest, Logistic Regression, KNN, Gradient Boosting, Naive Bayes, SVM and SVM with SGD classifier, and two DL models, including MLP and CNN, using tools like Matplotlib, Scikit-learn (2021), and Tensorflow (2024). The MLP model achieved the highest accuracy at 94.5%, outperforming previous studies and demonstrating the potential of advanced algorithms in CVD prediction. This study provides a comprehensive comparison of ML, DL algorithms, highlighting their effectiveness in predicting CVD. Our findings indicate that ML and DL algorithms, particularly the MLP model, can significantly enhance CVD prediction, offering valuable insights for future research and clinical applications.