Abstract
Real-time substance use detection is an ever-evolving field of modern science that could enable doctors to timely intervene substance-use in natural environments. Tools exist that aim to detect substance use in pre-recorded data streams, but few see implementations of real-time substance-use detection. This paper proposes a real-time machine- learning based algorithm for substance use detection on a platform that emulates real-time data stream transmissions at 5G frequencies for Wireless Body Area Networks. The Machine learning algorithms within this paper specifically focus on electrodermal activity patterns along with associated bio-physiological signals to detect Cocaine-use and utilize confirmed cases to compare the accuracy of the proposed algorithm against existing detection algorithms in an emulated natural environments. This paper claims that electrodermal activity can be used as a predictor for Cocaine use, using confirmed cases to validate the accuracy of our algorithm and establish a connection between electrodermal activity and Cocaine-use.