Donor-Independent Metabolomics Enables Bloodstain Age Determination at Crime Scenes.

Determining when bloodstains were deposited remains an unsolved challenge in forensic science, limiting investigators' ability to reconstruct events and verify suspect timelines. Here, we develop a metabolomics-based approach combining liquid chromatography-mass spectrometry (LC-MS) with machine learning to estimate bloodstain age independent of donor-specific variation. Through untargeted analysis and degradation studies, we identified 51 time-dependent biomarkers and transformed their intensities into stable ratios that normalize for individual differences and blood volume. Using samples collected under controlled environmental conditions, we achieve high accuracy for forensically relevant timeframes with prediction errors of ∼7 h for fresh bloodstains and near-perfect classification of samples as recent (<60 h) or aged (>60 h). Validation on two independent data sets confirms strong performance under typical indoor conditions, while highlighting sensitivity to extreme environmental fluctuations. By addressing key biological and technical sources of variability that have hindered translation to practice, this study establishes a robust analytical framework for bloodstain age estimation. The approach offers a practical foundation for future operational implementation and has the potential to substantially improve forensic timeline reconstruction.