Efficient Hybrid Prediction Model for Optimized Gait Analysis

Abstract

The analysis of human gait aids in identifying and tracking people at a distance newlinedepending on their walking pattern. This human gait analysis is used in video newlinesurveillance, defense, medical applications, etc. Emergence of IoT (Internet of Things), newlinehuman-assistive technologies in healthcare services have reached the peak of their newlineapplication in various fields, particularly in disease diagnosis and its treatment process. newlineThese IoT devices require awareness of human movements to provide better assistance newlinein clinical applications and the daily activities of users. Hence, real-time gait analysis newlineremains the key catalyst for the development of intelligent assistive devices. Gait newlinerecognition systems have substantially advanced in terms of high accuracy recognition newlinewith the integration of machine learning and deep learning algorithms. Many existing newlinemodels are focused on improvement of gait recognition for better performance in realtime newlineapplications. To properly reconcile these medical profession developments, a newlinesystem must determine the most essential body traits that influence accurate diagnosis newlineand classification. The main goal and purpose of this research work on gait recognition newlineare to provide the best way for the diagnosis of the disorders seen in patients with newlineParkinson s disease. The objective of this work has been achieved by primarily newlinedesigning a smart Wearable IoT (WIoT) based system for the collection of the gait newlinesignals. The collected signals were examined by various classification models for newlineenhancement the treatment process in the medical field through reduced features of gait newlineanalysis. A hybrid novel Firefly Optimized Whale (FOW) algorithm has been proposed newlinefor identification of best features of the gait signals. An optimized new hybrid algorithm newlinethat ensembles the Convolutional Neural Network (CNN) layers with Gated Recurrent newlineUnit (GRU) and Bat-Inspired classification networks with improved performance were newlinecarried out and the results were analyzed. These novel CNN and GRU networks were newlineemployed for obtai