A Novel Nontargeted Screening Strategy for New Psychoactive Substances: A Case Study of Synthetic Cannabinoids Based on Electron-Activated Dissociation High-Resolution Mass Spectrometry and Intelligent Elucidation

Nontargeted screening of new psychoactive
substances (NPSs) has
always been a challenging task, typically involving data acquisition,
the extraction of suspicious peaks, and mass spectrometry elucidation
in the screening process. The ongoing advancement of instrument acquisition
technology and data analysis methods has resulted in an increasing
amount of sample data requiring manual elucidation, significantly
reducing the efficiency of forensic identification work and leading
to issues such as missed detections and false positives. This study
proposed a novel nontargeted screening strategy that is capable of
automatically elucidating the NPS classes and chemical structures
of unknown designer drugs. For practical use, we applied electron-activated
dissociation (EAD) technology to analyze 181 synthetic cannabinoids
(SCs) and developed novel mass spectrometry intelligent elucidation
(MSIE) software to achieve the nontargeted screening of NPSs and automated
structural elucidation of SCs. MSIE software comprises an NPS nontargeted
screening model, an SC subclass classification model, and a mass spectrometry
intelligent elucidation algorithm. The NPS nontargeted screening model
was trained on CID data from 505 NPSs, achieving the classification
of 8 NPS classes, with the highest F1 score reaching 93.3%. The SC
subclass classification model was trained on EAD data from 181 SCs,
achieving the classification of 7 SC parent structures, with the highest
F1 score reaching 95.3%. The mass spectrometry intelligent elucidation
algorithm includes functionalities such as candidate chemical structure
generation, spectral prediction, candidate structure scoring, and
fragment ion peak matching, all without any manual intervention throughout
the entire process.