Support Vector Machine Classification of Adulterated Illicit Opioids Using Paper-Spray Mass Spectrometry.

Area of Science:

• Forensic Chemistry
• Analytical Chemistry
• Data Science

Background:

• Illicit opioid supply in North America is complex and potent, posing risks to drug users.
• Current drug checking methods, like targeted paper-spray mass spectrometry (PS-MS), struggle to detect novel psychoactive substances.
• Emerging adulterants in illicit drugs necessitate advanced analytical techniques for timely identification.

Purpose of the Study:

• To develop and validate a machine learning classifier for detecting unknown adulterants in opioid samples.
• To utilize support vector machine (SVM) algorithms for identifying substances beyond routine targeted analysis.
• To improve drug checking capabilities for emerging threats in the illicit drug supply.

Main Methods:

• Employed a support vector machine (SVM) classifier with full-scan mass spectrometry data.
• Used paper-spray mass spectrometry (PS-MS) for rapid, on-site drug analysis.
• Applied Shapely additive explanations (SHAP) to interpret SVM model predictions.

Main Results:

• The SVM classifier successfully detected ortho-methylfentanyl, an emerging adulterant, in fentanyl and caffeine mixtures.
• Achieved a precision of 0.77 and a recall of 0.94 for ortho-methylfentanyl detection.
• SHAP analysis provided insights into the chemical features driving adulterant identification.

Conclusions:

• The developed SVM approach effectively identifies adulterants from mass spectrometry data during routine drug analysis.
• This method enhances the ability of drug checking programs to detect novel and emerging substances.
• The findings offer a new strategy for improving the safety of the illicit drug supply.