Multi-sensor Data Fusion based Medical Data Classification Model using Gorilla Troops Optimization with Deep Learning

Abstract

Wireless Body Sensor Network (BSN) comprises wearables with different sensing, processing, storing, and broadcast abilities. Once several devices acquire the data, multi-sensor fusion was needed for transforming erroneous sensor information into maximum quality fused data. Deep learning (DL) approaches are utilized in different application domains comprising e-health for applications like activity detection, and disease forecast. In recent times, it can be demonstrated that the accuracy of classification techniques is enhanced by the combination of feature selection (FS) approaches. This article develops a Multi-sensor Data Fusion based Medical Data Classification Model using Gorilla Troops Optimization with Deep Learning (MDFMDC-GTODL) algorithm. The proposed MDFMDC-GTODL method enables collection of various daily activity data using different sensors, which are then fused to produce high-quality activity data. In addition, the MDFMDC-GTODL technique applies optimal attention based bidirectional long short term memory (ABLSTM) for heart disease prediction. In this study, Gorilla Troops Optimization Algorithm based FS (GTOA-FS) technique is involved to improve the classification performance. The simulation outcome of the MDFMDC-GTODL technique are validated and the results are investigated in different prospects. A wide-ranging simulation analysis stated the better performance of the MDFMDC-GTODL method over other compared approaches.