Machine learning combined with tandem column technology for food and tobacco analysis and discrimination based on the biomarker of L-lactic acid

Excessive D-lactic acid in fermented foods can result in unpleasant flavors and is difficult for the human body to metabolize. This study is the first to combine tandem column technology (C18 and Poroshell chiral-T) with a machine learning strategy. The method achieved a resolution of 2.86, with a concentration range of 50.00 μg·L −1 -50.00 mg·L −1 and a limit of detection of 16.67 μg·L −1 . We aimed to classify different lactic acid products and tobacco, and the research showed that L-lactic acid was predominant in all food analyses , while D-lactic acid was only found in lactic acid products In addition, the machine learning algorithms were optimized, and the random forest model effectively distinguished samples with similar lactic acid content. Based on this collaborative model of tandem column and machine learning, lactic acid chiral biomarkers gained the ability to classify food samples. The accuracy of model reached 80 %. This strategy contributes to the chiral analysis of lactic acid in complex food matrices. It also demonstrates the practical value of classification in food research.