Accurate Prediction of Death Time via Integrating Microbial Community Dynamics

This study addresses the challenge of accurately estimating Postmortem Interval (PMI), the time since death, employing a data-driven approach. PMI determination is crucial in forensic investigations, and traditional methods often lack precision. We focus on utilizing a data mining approach Regularized Random Forest with cross-validation to enhance PMI prediction accuracy. Unlike conventional methods, our approach incorporates external information about the deceased, recognizing the impact of contextual factors on PMI estimation. Recent advancements have seen statistical methods leveraging dynamic changes in microbial communities to predict PMI. However, accuracy has been hindered by various sources of noise. To overcome this limitation, we propose a novel data mining approach, integrating cross-validation techniques and external information to refine PMI predictions. Through an empirical demonstration, we establish that our approach surpasses existing procedures in terms of accuracy, as validated against published datasets. This research contributes to the advancement of PMI estimation methodologies, emphasizing the importance of incorporating comprehensive data mining techniques and contextual information for more precise forensic applications.