A Device for the Rapid Detection of Benzodiazepines and Synthetic Cannabinoids via Fluorescence Spectroscopy and Machine Learning

Abstract

Introduction: Drug abuse is a worsening societal issue across the globe. In the UK, the abuse of Benzodiazepines and Synthetic Cannabinoids is particularly prevalent, especially in healthcare and custodial settings, and there is currently no solution to quickly detect these substances for harm reduction. Methods: We are developing a portable and rapid device that utilizes Fluorescence Spectroscopy and Machine Learning to detect Benzodiazepines and Synthetic Cannabinoids in a variety of media, including saliva. The device will be able to distinguish between variants of a given drug to provide an informative output to the end user. Results: Development of the first prototype of the device is nearing completion, and lab data has been collected for training the device's drug-detecting predictive model. Current experiments with established supervised-learning algorithms show favourable results in distinguishing Synthetic Cannabinoids. Trials of the device in UK drug hotspots are imminent and will result in a significant data collection of the scans performed and the predictions that the model made per scan. This will provide us with an unprecedented insight into the pervasiveness of illegal drug use in the UK and drive improvements for future iterations of the device. Conclusions: We feel that this is a crucial and promising technology for harm reduction to stem the flow of drugs to the most vulnerable in society.