Heart Disease Prediction using Fog Computing based Wireless Body Sensor Networks (WSNs)

Abstract

Wireless Body Sensor Network (BSNs) are devices that can be worn by human beings. They have sensors with transmission, computation, storage and varying sensing qualities. When there are multiple devices to obtain data from, it is necessary to merge these data to avoid errors from being transmitted, resulting in a high quality fused data. In this proposed work, we have designed a data fusion approach with the help of data obtained from the BSNs, using Fog computing. Everyday activities are gathered in the form of data using an array of sensors which are then merged together to form high quality data. The data so obtained is then given as the input to ensemble classifier to predict heart-related diseases at an early stage. Using a fog computing environment, the data collector is established and the computation process is done with a decentralised system. A final output is produced on combining the result of the nodes using the fog computing database. A novel kernel random data collector is used for classification purpose to result in an improved quality. Experimental analysis indicates an accuracy of 96% where the depth is about 10 with an estimator count of 45 along with 7 features parameters considered.