The identification of biological stains at crime scene: a promising role for proteomics and machine learning

Identification of stains in criminalistics remains a critical challenge in forensic sciences. In particular, determining the biological origin of a trace can be crucial in an investigation in order to understand the circumstances in which it was deposited. This study aims to develop three distinct tests based on proteomics data for identifying human bodily fluids. The first test focuses on constructing specific peptide biomarkers for each fluid. The second test quantitatively assesses ratios between peptide abundances. The final test employs artificial intelligence, utilizing multioutput Random Forest, to distinguish between pure fluids and mixtures. Data from 20 individuals per fluid and 78 mixtures are analyzed. This research underscores the potential of proteomics in forensic bodily fluid identification. The evaluation of the different tests highlights their complementarity, as they focus on different features obtained from a single experiment.