Adaptive Bio Inspired Deep Neural Network based Heart Disease Prediction Model using Propagation Measures

The problem of heart disease prediction has been identified as key challenge in medical science. There exist numerous techniques towards predicting heart disease using various data sets. However, the accuracy of predicting heart disease is still below the mark. To handle this issue, an Adaptive Bio Inspired DNN based Heart Disease Prediction Model (ABDNN) model is presented in this article. The method utilizes different data sets containing health records of various patients which includes clinical and diagnosis features. The procedure begins with preprocessing the dataset using a feature-level normalization scheme. The normalized dataset is then used for feature selection through the Walrus Optimization Algorithm (WaOA), where the objective function is designed to measure the Walrus Position Support (WPS). The WPS indicates the position of the walrus and guides the feature selection process. As a result, the selected features are used to train a deep neural network, with neurons designed to evaluate the Heart Disease Diffusion Support Scale (HDSM) for classification. The proposed model enhances the accuracy of heart disease prediction across various datasets.