Estimating Postmortem Interval of Buried Pig Carcasses by Integrating Microbial Succession Patterns with Machine Learning Algorithms.

Microbial succession serves as a promising tool for estimating the postmortem interval (PMI). However, the patterns of microbial succession in burial scenarios require further exploration. This study established a pig carcass model, including buried and surface (control) groups, to investigate this. Using 16S ribosomal RNA (16S rRNA) gene sequencing, we analyzed microbial community changes and their differences across various decomposition stages. Results indicated that the decomposition rate of buried carcasses was slower than that of surface carcasses. Following the early decomposition stages, the alpha diversity of skin and underlying soil samples from buried carcasses decreased, a trend similar to that observed in the surface group. A significant shift in bacterial communities occurred in the buried group during abdominal rupture, mirroring the pattern in the surface group. At the phylum level, the relative abundance of Proteobacteria in the skin and soil of the buried group increased during later stages, consistent with the surface group. Furthermore, the buried and surface groups each possessed unique microbial taxa that responded to PMI changes. Using genus-level data, we identified feature taxa and constructed Random Forest models for PMI estimation. In the buried group, the mean absolute error (MAE) was 5.47 days for skin and 4.91 days for soil, while in the surface group, it was 5.59 days for skin and 5.30 days for soil. Although the model's generalizability is currently limited by the sample size, the results demonstrate the predictability of microbial succession across different environmental contexts, underscoring its potential as a tool for PMI estimation in buried remains.