Chemometric Processing of Ecstasy Tablet Thermal Profiles Using Machine Learning Methods

Ecstasy is an amphetamine-type substance scientifically known as MDMA (3,4- methylenedioxymethamphetamine) that belongs to a group of illicit drugs called “club drugs” that are widespread and distributed illegally (Maione et al., 2016). Its ever-growing popularity and variety in multiple forms, shapes, and colors complicates the identification and tracking process to law enforcement and forensic analysts. The current study focused on integrating data analytics and advanced chemometric methods to Differential Scanning Calorimetry (DSC) to analyze the thermal profiles of ecstasy tablets. This approach aims to discriminate between tablets linked to different suspects and origins through thermal profiling, which will provide law enforcements with an advanced toolset to enhance the precision and efficacy of forensic investigations by identifying sources and distribution patterns to aid the disruption of this illegal drug. To enhance the reliability and accuracy of drug profiling, it is important to integrate data analysis techniques into forensic science. The current research used DSC thermal data for the analysis of 17 various ecstasy tablets. After cleaning the data, peak integration was performed using OriginPro, a specialized thermal analysis software, to conduct peak analysis and extract the additional key features. Thereafter, chemometric techniques and multivariate analyses such as Hierarchical Cluster Analysis, K- Means Cluster Analysis and Principal component analysis (PCA) were applied to the extracted values to compare the various ecstasy tablets and further explore the data. The findings answered the research questions and confirmed the chemometric techniques used are effective tools for differentiating between samples, with some challenges noted in achieving distinct separation in certain cases. Despite these challenges, the analysis provided valuable insights, reinforcing the applicability of these methods in drug profiling. Future samples will undergo the same analysis, results will be compared to the current PCA values of this study by using distance metrics to either find similarities or differences. The similar values will be considered as a match to the same drug profile linked to a suspect or origin, or the differences will add a new drug profile to the library of this analysis.