Follow us
P247 – Transcriptome-based Predictive Framework for Accurate Estimation of Postmortem Interval in Forensic Practice
Accurate estimation of the postmortem interval (PMI) remains a major challenge in forensic science. In this study, we developed a transcriptome-based predictive framework for PMI estimation, utilizing publicly available RNA sequencing (RNA-seq) data derived from 363 dorsolateral prefrontal cortex (DLPFC) gray matter samples. Gene expression profiles (logCPM) were first filtered via Spearman correlation analysis with PMI (FDR < 0.05, |ρ| > 0.2). Subsequently, bootstrap-based stability selection combined with elastic net regression (α = 0.5) was applied to identify and rank candidate feature genes, from which an optimal panel of 100 genes was selected based on cross-validated performance. Among all tested predictive models, the support vector machine (SVM) model achieved the best performance, with a correlation coefficient (r) of 0.843 and a mean absolute error (MAE) of 6.98 h. The elastic net model exhibited slightly inferior but comparable performance (r = 0.839, MAE = 7.23 h), whereas the random forest and XGBoost models showed notably reduced predictive accuracy. To further validate the biological relevance and practical applicability of the identified markers, qPCR verification was performed on an independent brain tissue sample cohort with controlled postmortem intervals, which confirmed consistent temporal expression trends of the selected genes. Collectively, these findings demonstrate the great potential of integrating transcriptomic biomarkers into routine forensic workflows, and the established predictive framework provides an objective and quantitative approach to enhance the accuracy and reliability of PMI estimation in actual forensic casework.
Authors
- Yiping Hou (Institute of Forensic Medicine, West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University, China)
- Qingxin Yang (Institute of Forensic Medicine, West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University, China)
- Hong Zhu (Institute of Forensic Medicine, West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University, China)
- Zheng Wang (Institute of Forensic Medicine, West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University, China)
On the same topic
