Body Fluid Identification by Nanoparticle Enhanced LDI MS

Body fluids contain valuable information about human body and are the most significant evidence in forensic investigation at crime scenes. Their characterizations can provide investigative leads and contributes to case proof. Identifying different kinds of body fluids is critical but challenging. Traditional forensic methods rely on enzymatic assays as presumptive tests, which are destructive and invasive to samples, as well as time-consuming and complex in procedures, resulting in their inability to meet the demand of rapid detection. Here, a body fluids (including serum, plasma, tear, and urine) integrated identification platform through machine learning of metabolic fingerprinting (MF) was developed, using nanoparticle enhanced laser desorption/ionization mass spectrometry (NPELDI MS). The direct MFs of multiplex body fluids were acquired with fast speed (10 s per sample) and high reproducibility (coefficient of variation, CV < 15%) at sample volume of 5 nL, coexisting with minimum sample invasiveness. Further, two strategies (four classifications and dichotomy) were conducted to identify diverse body fluids through machine learning with the area under the cures (AUCs) of 0.969 for four classifications and 0.973 to 1.000 for dichotomy, respectively. Finally, a biomarker panel was established, and it improved the performance of body fluids identification with AUC of 1.000. Our work provides a frontier reference for differentiating body fluids through metabolomics and innovative perspectives for forensic application.