A Novel Chemometric Local Approach for Qualitative and Quantitative Analysis of Cocaine, MDMA, and THC-Related Products: Method Application within the Mossos d’Esquadra (Catalan Regional Police)

A novel chemometric protocol integrating near-infrared (NIR) spectroscopy with advanced classification and regression models for the qualitative and quantitative analysis of illicit drugs, implemented by the Catalan Regional Police, the Mossos d'Esquadra, in 2024-2025, is presented. The study addresses the increasing demand for efficient and reliable forensic analysis, precipitated by a marked rise in drug seizures and the complexity of judicial processes in Catalonia. The proposed workflow combines rapid, nondestructive NIR screening with optimized machine learning models, including linear and nonlinear approaches, to identify and quantify active compounds in seized samples of cocaine, MDMA, and cannabis derivatives. To enhance analytical accuracy, the methodology leverages local model calibration using real seized data and compares the model performance to commercial global approaches. Validation results demonstrate that the local model delivers robust classification, yielding a global accuracy of 97%. When it comes to quantitative regression, prediction errors and bias lower than those of global models for key compounds were found. These outcomes confirm the suitability of local-model-based NIR-chemometrics for routine forensic applications, facilitating rapid turnaround and reliable results while maintaining operational efficiency. The protocol offers an accessible and cost-effective alternative to proprietary solutions, underscoring its value in the forensic analysis landscape.