Related Experiment Video
Updated: Jun 17, 2026
07:10
Fluorescent Paper Strips for the Detection of Diesel Adulteration with Smartphone Read-out
Published on: November 9, 2018
Field-Portable Technology for Illicit Drug Discrimination via Deep Learning of Hybridized Reflectance/Fluorescence
Alexander Power1, Matthew Gardner2, Rachael Andrews2
1Department of Computer Science, University of Bath, Bath BA2 7AY, U.K.
Analytical Chemistry
|May 7, 2025
Summary
A new portable device uses hybrid spectroscopy and deep learning to accurately identify novel psychoactive substances (NPS), including benzodiazepines and nitazenes, in street drugs. This technology aids harm reduction efforts by enabling rapid, on-site drug analysis.
Area of Science:
- Analytical Chemistry
- Forensic Science
- Machine Learning
Background:
- Novel psychoactive substances (NPS) present a significant challenge due to their potency and rapid evolution.
- Existing field-portable technologies struggle to detect NPS, hindering harm reduction services.
- Benzodiazepines, synthetic opioids, synthetic cannabinoids, and cathinones are major contributors to NPS-related harm.
Purpose of the Study:
- To develop and validate a portable spectroscopic method for identifying NPS.
- To provide concentration information for detected NPS.
- To support community-based harm reduction by enabling rapid, on-site drug analysis.
Main Methods:
- Hybridizing fluorescence and reflectance spectroscopies.
- Utilizing a deep learning algorithm trained on preprocessed spectral data.
- Employing a low-cost, portable device requiring minimal user training.
Main Results:
- Accurate identification of 11 benzodiazepines in street tablets, including excipients.
- Detection of complex drug mixtures, such as nitazene + benzodiazepine, fentanyl + xylazine, and heroin + nitazene.
- Demonstrated potential for generalization to other drug classes beyond NPS.
Conclusions:
- Hybrid spectroscopy combined with deep learning offers a viable solution for NPS detection.
- The portable device has immediate potential to enhance community-based harm reduction services.
- The developed approach shows promise for broader applications in drug analysis.
