Multi-cascaded heart disease prediction using hybrid deep learning and optimization techniques

A novel deep learning based heart disease prediction model is proposed. Initially, the collected data is fed into the preprocessing phase using the NaN fill method. Then, the preprocessed data is given to the data transformation phase using data normalization approach. Further, the transformed data are fed into the optimal weighted feature selection process, which is selected by using the developed Mutated Iteration-based Fire Hawk with Coyote Optimization (MI-FHCO). Subsequently, heart disease is predicted by Multi-Cascaded Deep Learning Network (MDLNet). The best accuracy rate of the proposed approach is attained as 96.65% for dataset 4 to demonstrate its superior performance.