Machine learning-assisted classification and adulteration detection of fatty oils using fatty acid profiles obtained via supercritical fluid chromatography

Fatty oils are essential in the pharmaceutical field for enhancing the solubility and oral bioavailability of drugs, as well as reducing the risk of cardiovascular diseases. Consequently, detecting adulteration in specific fatty oils is critical to ensuring pharmaceutical quality and safety. In this study, we present a rapid method for classifying and detecting adulteration that combines supervised machine learning (ML) with supercritical fluid chromatography-mass spectrometry (SFC-MS). SFC MS was employed to analyze and establish data for 14 free fatty acids (FFAs) in five common fatty oils. Three different ML models were successfully utilized to identify these five fatty oils. Furthermore, the FFA composition of 45 mixed fatty oils with varying adulteration ratios was rapidly identified using the same models, among which the random forest (RF) algorithm demonstrated exceptionally high accuracy (100 %) by correctly classifying all samples. By integrating ML with FFA analysis, this study provides valuable insights for the rapid identification and detection of adulteration in fatty oils.