Predicting Cardiovascular Disease Risk using Artificial Neural Networks Models

The rise in Heart related disease medically known as cardiovascular disease (CVD) is cause of concern for medical practitioners. It is one of the reasons of sizeable number of deaths worldwide, emphasizing the urgent need for early and accurate risk prediction models. In recent years, Artificial Neural Networks (ANNs), which simulate the structure and behavior of the human brain, have become popular because they can handle complex and intricate non-linear relationship medical data. In this study the performance of various ANN architectures is explored and compared for predictive accuracy. The models included in this investigation are convolutional neural networks (CNN), deep belief networks (DBN), radial basis functions (RBF), long short-term memory networks (LSTM), and multilayer neural networks (MNN). The metrics precision, accuracy, F1-score, AUC-ROC and recall are used to evaluate the models to ensure a comprehensive assessment of performance. It was found that the LSTM model showed the maximum accuracy (90%) and AUC-ROC score (89.24%), showing that it may be better for CVD prediction. While the Convolutional Neural Network demonstrated lower predictive power with an accuracy of 76%, the multilayer Neural Network models also fared well, with an accuracy of 86.67%. The study finds that more sophisticated architectures, such as LSTM and MNN, greatly improve CVD risk prediction overall. In conclusion, the comparative analysis of ANN models shows that more advanced architectures like LSTM and Multilayer networks significantly improve cardiovascular disease risk prediction.