nps aimex: AI for forensic drug analysis | NPS AImEx

Analytical Challenge

The unequivocal identification of an unknown psychoactive substance typically requires:

Multiple independent analytical techniques,
Expert interpretation,
Significant time and specialised infrastructure.

For novel substances, database-based approaches alone are insufficient.

Benchtop NMR spectrometer (Spinsolve 80 Ultra) with data analysis monitors in laboratory

Benchtop NMR spectrometer (Spinsolve 80 Ultra) with data analysis monitors in laboratory

Scientific Concept

NPS AImEx develops a fully integrated, AI-assisted workflow that combines:

Automated interpretation of GC–EI–MS data,
AI-based generation of molecular structure candidates,
Independent validation using predicted NMR, IR, and Raman spectra,
Expert-rule-based filtering and statistical confidence assessment.

background pattern

background pattern

AI workflow for NPS identification: generation from EI-MS, verification with NMR, IR, Raman spectra

AI workflow for NPS identification: generation from EI-MS, verification with NMR, IR, Raman spectra

Workflow Structure

Step 1 – Initial Assignment

GC–MS spectral comparison (where reference data exist),
AI-supported structure proposal for unknown compounds.

Step 2 – Validation

Cross-validation using spectroscopic methods,
Consistency checks across independent techniques,
Assignment of confidence metrics.

Quality Targets

Automated structure elucidation without manual expert intervention in >70% of cases,
At least twofold reduction in analysis time,
Increased robustness and traceability of identification results