An intelligent heart disease prediction framework by optimal feature integration with multi serial cascaded adaptive dilated network using heuristic improvement

Heart disease remains a leading cause of death globally, and early detection is crucial for effective intervention and improved patient outcomes. The challenge in existing approach limits the accuracy and efficiency, leading to missed diagnoses. This research work developed an adaptive system to predict cardiac illness using a heuristic assisted deep learning approach which can accurately and efficiently predict the risk of heart disease from patient data and improved patient outcomes. Initially the data was gathered and deep features from gathered data are extracted by auto encoder, then it is given to the suggested Improved Uniform Random Variable based Equilibrium Optimizer for choosing the optimal features and optimizes the weights of the autoencoder. The optimally selected features are combined for the formation of the optimal weighted features. These features are provided as the input to the Multi Serial Cascaded Adaptive Dilated Network. This network is formed by combining a Deep Temporal Convolution Network, Long Short-Term Memory, and One-Dimensional Convolution Neural Network. For enhancing the prediction accuracy, the parameters in the suggested Multi Serial Cascaded Adaptive Dilated Network are tuned by Improved Uniform Random Variable based Equilibrium Optimizer. The result demonstrated that the accuracy of the developed framework is 94.40 %, which is enhanced than other existing approaches. Thus, the designed system can detect heart disease and improving patient outcomes. By analyzing patient data and predicting potential outcomes, the developed framework can help clinicians for personalized treatment plans that are tailored to individual needs and risk factors.