Rapid Screening of Consumer Products by GCxGC-HRT and Machine Learning Assisted Data Processing

• DOI: 10.1021/jasms.3c00107
• Publication Date: 2023-07-06T14:31:45Z
• Authors: Michael J. Hartnett; William D. Watson; Jake A. Janssen; Jenna Hua; Jarod Grossman; Qingchu Peng; Prativa Hartnett; Kristin A. Favela
• Keywords: Irritants; Humans; Cosmetics; Software; Algorithms; Gas Chromatography-Mass Spectrometry; 3. Good health

This work demonstrates high-throughput screening of personal care products to provide an overview potential exposure. Sixty-seven from five categories (body/fragrance oil, cleaning product, hair care, hand/body wash, lotion, sunscreen) were rapidly extracted and then analyzed using suspect by two-dimensional gas chromatography (GCxGC) high-resolution mass spectrometry (GCxGC-HRT). Initial peak finding integration performed commercial software, followed batch processing the machine learning program Highlight. Highlight automatically performs background subtraction, chromatographic alignment, signal quality review, multidilution aggregation, grouping, iterative integration. data set resulted in 2,195 compound groups 43,713 individual detections. Compounds concern (101) downselected classified as mild irritants (29%), environmental toxicants/severe (51%) endocrine disrupting chemicals/carcinogens (20%). High risk compounds such phthalates, parabens, avobenzone detected 46 out 67 (69%), only 5 (7%) listed these on their ingredient labels. The results for compared software (ChromaTOF) 5.3% detections discerned Highlight, demonstrating strength algorithm effectively discover low-level signatures. provides a significant labor advantage, requiring 2.6% time estimated largely manual workflow software. In order address needed postprocessing assignment identification confidence, new machine-learning-based was developed assess assigned library matches, balanced accuracy 79% achieved.