Designing Communication System for Person with Locked in Syndrome Using Machine Learning Technique
S. Ramkumar, G. Emayavaramban, K. Sathesh Kumar, K. Shankar, M. Ilayaraja, Padmanaban Sriramakrishnan, J. Macklin Abraham Navamani · 2019
Electrroculogram (EOG) based Human Computer Interaction (HCI) research was growing day by day due to need of new handheld assistive devices for the persons with Locked in Syndrome (LIS). For that particular disability only the solution was to develop communication devices to overcome their problem. So we conducted our experiment with ten subjects using Randomized Hough Transform (RHT) and Random Forest Tree Classifier (RFT). During the study, all the subjects were male subjects and their age was ranged from 30 to 45. At the end of the study we obtained the average maximum accuracy of 94.48% and mean minimum accuracy of 88.90% and mean average accuracy of 92.77% for RFT classifier. Best classification accuracy of individual subjects was obtained for the subject S4 with average mean classification accuracy of 93.96%. The minimal classification accuracy of individual subjects was attained by the subject S8 with average mean classification accuracy of 91.70%. The study concluded that individual performances were presented in between 93.96% to 91.70% and finalized that expanding the command signals to handle the exterior machines like wheelchair, TV Remote, mouse, game controller, mobile phone controller, alarm controller, keyboard and home Appliances were possible for disbaled person in usual way.