Development of a muscle activated switch for the severely debilitated

Abstract

Patients with Locked-in Syndrome are unable to move, apart from certain eye movements and perhaps one or two other muscles; this may be due to Amyotrophic Lateral Sclerosis, nerve cell damage or a stroke, among other reasons. The condition normally results in the patient losing the ability of verbal communication. The task at hand was to develop an EMG switch that would generate outputs based on the detected muscle activity. The existing switch achieved substantial success, giving binary YES/NO outputs. The investigation sought to consider three limitations of the existing system: the sensitivity of the switch to the smallest possible signal amplitude; the ability of the switch to detect signals of differing magnitudes; the ability of the switch to detect signals of differing velocities. Significant progress was made regarding the first two matters. By using an adaptive threshold based on moving averages and upper bounds defined by data spread, the algorithm was able to detect large and small motions to an average of 91.8% accuracy, 95.8% precision and 95.9% recall-building on the previous algorithm. Regarding detecting signals of differing magnitudes, set threshold bands were show on screen as visual prompters to appear at different levels, which the patient/user would use to aim for different threshold bands to generate a number of different outputs. With training, a user would be able to communicate more effectively. Future research should seek to continue improved the solutions to the first two challenges, and make progress on signal velocity detection.