An efficient Heart-Disease Prediction System using Machine Learning and Deep Learning Techniques

One of the leading causes of death globally is stroke and it is highly essential to develop solutions to identify stroke. Though early detection is helpful in saving human lives, hence it is required to collect many physiological data such as blood pressure, ECG, glucose level and cholesterol level using wearable devices. Fog computing assisted processing is initially used to store and analyse large amounts of data. But in this approach, initial quality assessment, many preprocessing tasks, and feature extraction are the challenges in the fog computing layer, and the decisions are obtained at increased computation time. The main objective of this paper is to develop efficient disease prediction systems capable of accurately predicting stoke and heart related diseases, as well as determining the severity level of the diseases with less computation time. Initially we use Fog computing framework for classifying the parameters an individual patient. An Enhanced Grey-Wolf Optimization-based Feature Selection Algorithm (EGWO-FSA) and Deep Belief Network (DBN) disease prediction system is used to predict heart, stroke and diseases. According to the proposed approach, quantitative analysis is done to determine parameters like specificity, accuracy, as well as sensitivity.