Citation

Coppey F, Bécue A, Sacré PY, Ziemons EM, Hubert P, Esseiva P. Forensic Sci. Int. 2020; 317: e110498.

Copyright

(Copyright © 2020, Elsevier Publishing)

DOI

10.1016/j.forsciint.2020.110498

PMID

33017781

Abstract

The analysis of illicit drugs faces many challenges, mainly regarding the production of timely and reliable results and the production of added value from the generated data. It is essential to rethink the way this analysis is operationalised, in order to cope with the trend toward the decentralization of forensic applications. This paper describes the deployment of an ultra-portable near-infrared detector connected to a mobile application. This allows analysis and display of results to end users within 5s. The development of prediction models and their validation, as well as strategies for deployment within law enforcement organizations and forensic laboratories are discussed.

Language: en

Keywords

Machine learning; Big data; Validation; Cannabis; Heroin; Cocaine; Forensic science; Near infrared; Statistical model