In a recent study out of the Shangdong University of Political Science and Law and the Shandong Energy Group, based in Shandong, China, headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS) was used alongside machine learning (ML) algorithms to detect the geographic origins of coal associated with larceny cases. Their findings were published in the Journal of Chromatography A (1).
Coal, which is formed through long-term accumulation and transformation of plant remains over millions of years in appropriate geological environments, plays an important role in the power and chemical industries as an abundant and inexpensive fossil fuel. According to the International Energy Agency’s 2023 Coal Market Report, worldwide coal demand topped 8.5 billion tons in 2023, a 1.4% rise over the previous year. Moreover, with rapid global industry and urbanization, the need for coal continues to expand. However, a side effect of this process has been a rise in theft of coal and similar products. In forensic science, collecting and analyzing trace evidence from coal samples is critical for confirming evidence, reconstructing cases, and noting forensic contributions to solving larceny crimes.
Read More: Deep Learning Framework for Peak Detection at the Intact Level of Therapeutic Proteins
Precisely identifying coal trace evidence can be challenging with current methods, due to its minute quantity, fine texture, and intricate composition. For this study, the scientists integrated ML with volatile identification to accurately differentiate coal geographical origins through applying headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS). This technique has quickly grown popular as an emerging type of analysis technology (2). When GC and IMS are combined, detection accuracy was enhanced compared to IMS alone. Further, because HS-GC-IMS measurements have a two-dimensional nature to them, the results would contain great quantities of data (which may contain up to 106 or even 107 data points). This would lead to appropriate chemometric processing being required when HS-GC-IMS analysis is carried out. This technique has been primarily used to evaluate volatile organic compounds (VOCs) and provide continuous injection functionality and fast analytical capability.
As part of using HS-GC-IMS, the topographic distribution of volatiles in coals was depicted visually, allowing the elucidation of subtle distinctions through spectra and fingerprint analysis. Further, four supervised ML algorithms were created to quantitatively predict the geographic origins of natural coals using the HS-GC-IMS dataset; from there, they were subsequently compared with unsupervised models.
Read More: Advances in Chromatography Using Artificial Intelligence and Machine Learning
Notable volatile compounds were identified through quantitative analysis and an optimal Random Forest model. Random Forest is a commonly used ML algorithm that combines the output of various decision trees to reach a result (3). It has been praised for its ease of use and flexibility, as it can handle both classification and regression problems. When combined with quantitative analysis, the model offered a rapid readout and achieved an average accuracy of 100% in coal identification. Overall, by combining ML and HS-GC-IMS, a robust protocol with high classification accuracy was supplied, which successfully overcame the limitations of traditional analysis with low precision and efficiency. This study proved the potential and value of ML-assisted HS-GC-IMS analysis for rapidly authenticating coal trace evidence in larceny cases based on their volatile composition.
(1) Lu, W.; Ding, C.; Zhu, M. Discrimination of Coal Geographical Origins Through HS-GC-IMS Assisted with Machine Learning Algorithms in Larceny Case. J. Chromatogr. A 2024, 1735, 465330. DOI: 10.1016/j.chroma.2024.465330
(2) Gu, S.; Zhang, J.; Wang, J.; Wang, X.; Du, D. Recent Development of HS-GC-IMS Technology in Rapid and Non-Destructive Detection of Quality and Contamination in Agri-Food Products. TrAC Trends Analyt. Chem. 2021, 144, 116435. DOI: 10.1016/j.trac.2021.116435
(3) What is random forest? IBM 2024. https://www.ibm.com/topics/random-forest (accessed 2024-9-23)
SFE-SFC-MS Used to Analyze Transferred Plastic Additives from Laboratory Materials
September 30th 2024Supercritical fluid extraction (SFE) was combined with supercritical fluid chromatography (SFC) hyphenated to mass spectrometry (MS) to analyze plastic additives that could be transferred into the environment from laboratory gloves.