Time-Since-Death Estimation: A Review Of Ai Applications In Forensic Postmortem Interval Analysis

Abstract

Estimation of the postmortem interval (PMI) is an essential but difficult part of forensic analysis that has hitherto been based on subjective, frequently inaccurate markers. The recent developments in Artificial Intelligence (AI), especially Machine Learning (ML) and Deep Learning (DL), have accelerated a revolution in PMI estimation through the possibility of combining and processing various high-dimensional datasets. This paper discusses the fundamental concepts in AI that are being used for PMI estimation, including describing supervised and unsupervised learning models, neural networks, and convolutional and recurrent structures with specific orientations to process complex biological, chemical, and imaging data. It also discusses the contribution of multimodal data such as biochemical markers, transcriptomics, microbiomes, imaging, and environmental inputs in the improvement of accuracy and robustness of AI-based PMI models. Main research findings are examined, including the better performance of AI models like Artificial Neural Networks and Random Forests compared to conventional forensic practices. Notwithstanding these developments, issues remain, most notably around data availability, standardization, validation, and ethical implications related to the interpretability and implementation of AI in judicial proceedings. The article finishes with an appeal for more collaborative, standardized, and ethically informed research to move AI-based PMI estimation from experimental potential to operational forensic actuality.