Rapid and robust on‐scene detection of cocaine in street samples using a handheld near‐infrared spectrometer and machine learning algorithms

Abstract

On-scene drug detection is an increasingly significant challenge due to the fastchanging drug market as well as the risk of exposure to potent drug substances. Conventional colorimetric cocaine tests involve handling of the unknown material and are prone to false-positive reactions on common pharmaceuticals used as cutting agents. This study demonstrates the novel application of 740-1070 nm small-wavelength-range near-infrared (NIR) spectroscopy to confidently detect cocaine in case samples. Multistage machine learning algorithms are used to exploit the limited spectral features and predict not only the presence of cocaine but also the concentration and sample composition. A model based on more than 10,000 spectra from case samples yielded 97% true-positive and 98% true-negative results. The practical applicability is shown in more than 100 case samples not included in the model design. One of the most exciting aspects of this on-scene approach is that the model can almost instantly adapt to changes in the illicit-drug market by updating metadata with results from subsequent confirmatory laboratory analyses. These results demonstrate that advanced machine learning strategies applied on limited-range NIR spectra from economic handheld sensors can be a valuable procedure for rapid on-site detection of illicit substances by investigating officers. In addition to forensics, this interesting approach could be beneficial for screening and classification applications in the pharmaceutical, food-safety, and environmental domains.

Performance evaluation of handheld Raman spectroscopy for cocaine detection in forensic case samples

“…For a regression model, it is unlikely that a discrete 0.0% cocaine concentration is predicted for all non‐cocaine drugs and cutting agents. An adjustable cut‐off threshold on the predicted concentration is therefore applied in a similar fashion as earlier work using a near infrared (NIR)‐based model 7 . In this way, the number of FP results can be reduced by accepting a certain degree of FN results for low‐level (i.e., below threshold) cocaine samples.…”

“…Full details of these sets are reported elsewhere. 7 Thirteen case samples provided by the Amsterdam Police Laboratory were included in this study that were identified to 1).…”

“…The general characteristics of these sets are as follows: 58 cocaine HCl samples (average cocaine content 64.4 wt%, range 85.5–19.1 wt%); 32 cocaine base samples (average cocaine content 75.4 wt%, range 99.4–31.5 wt%); 20 pure adulterants, cutting agents or uncontrolled pharmaceuticals; 10 mixtures of common cutting agents; 10 common controlled substances with a white powdery appearance (e.g., amphetamine, mephedrone, MDMA, and methamphetamine). Full details of these sets are reported elsewhere 7 …”

“…An adjustable cut-off threshold on the predicted concentration is therefore applied in a similar fashion as earlier work using a near infrared (NIR)-based model. 7 In this way, the number of FP results can be reduced by accepting a certain degree of FN results for low-level (i.e., below threshold) cocaine samples. From a legal point of view, even low-level cocaine samples are controlled substances, and excluding these from subsequent further legal action is unwanted.…”

“…This may lead to false negative (FN) results when the incorrect test is performed or a controlled substance is encountered for which no test is available. 3 Other indicative testing strategies for on-site cocaine detection using handheld devices include electrochemical analysis, 6 (near) infrared spectroscopy, [7][8][9][10] and Raman spectroscopy [11][12][13][14] where the latter two offer possibilities to analyze substances directly through packaging material (when this material is sufficiently transparent). Over the last decade, Raman spectroscopy has become a viable tool in forensic chemistry, and its applications have been reviewed multiple times.…”