Accurate estimation of the postmortem interval (PMI) is crucial in forensic investigations but remains challenging due to environmental, individual, and cause-of-death variables. Traditional methods relying on postmortem changes (e.g., livor mortis, rigor mortis) are subjective and limited, especially for late PMI. Recent advances in artificial intelligence (AI) and computational pathology, particularly whole-slide imaging (WSI), enable data-driven PMI estimation by analyzing digital pathology images with high precision and reproducibility. Building on previous successes in using AI to estimate PMI in uninfected conditions, this study extends the method to bacterially infected mouse cadavers (Staphylococcus aureus, Escherichia coli, and Pseudomonas aeruginosa) under varying temperatures (25 °C, 37 °C, and 4 °C). The results demonstrate the model's robustness in diverse scenarios, achieving micro- and macro-area under the curve values (AUCs) of at least 0.873 (patch-level) and 0.717 (WSI-level) in training and testing sets, and no less than 0.948 (patch-level) in external validation. By leveraging easily prepared pathological sections and AI algorithms, this approach offers a practical, objective, and scalable solution for PMI estimation, enhancing forensic workflows. The integration of computational pathology with forensic science establishes a new technical benchmark for PMI estimation in both infected and uninfected cases.
Keywords: Artificial intelligence; Deep learning; Machine learning; Postmortem interval; Whole-slide imaging.
Copyright © 2025 The Chartered Society of Forensic Sciences. Published by Elsevier B.V. All rights reserved.