Integrating temporal dynamics of both microbial taxonomic and functional characteristics as a tool for forensic time since deposition estimation.

Predicting the time since deposition (TsD) of body fluid stain at a crime scene is highly valuable in forensic investigation, as it can connect a DNA-identified stain donor to a crime or estimate the post-mortem interval in forensic case involving cadaver. Previous study has demonstrated the applicability of microbial taxa for TsD prediction. In this proof-of-concept study, we explored the combined use of taxonomic and functional characteristics in human saliva microbiome for TsD prediction. We investigated the potential correlations between the temporal dynamics in microbial communities and the TsD of saliva stains, utilizing the relative abundance of microbial genera, amplicon sequence variants (AsVs), and the metabolic function pathway profiles. The results revealed that changes in the metabolic function pathway profiles of microbial communities could contribute to TsD estimation of the aged saliva stains. We developed ensemble models with multiple heterogeneous algorithms, also combined microbial species and metabolic function pathway profiles for TsD estimation. The TsD prediction model, developed using the relative abundance of the top 50 microbial genera and metabolic functional pathway profiles with the xgboost algorithm, achieved high accuracy (R = 0.76, mean absolute error = 6.28 days) in a 60-day deposition. Besides, we further interpreted the model and identified potential biomarkers for TsD estimation of saliva stains. The current study provided the first evidence to explore the potential of using multiple data modalities for predicting the TsD of aged saliva stains, offering valuable insight for forensic investigation.