AUTHOR=Iancu Lavinia , Bonicelli Andrea , Procopio Noemi TITLE=Decomposition in an extreme cold environment and associated microbiome—prediction model implications for the postmortem interval estimation JOURNAL=Frontiers in Microbiology VOLUME=15 YEAR=2024 URL=https://www.frontiersin.org/journals/microbiology/articles/10.3389/fmicb.2024.1392716 DOI=10.3389/fmicb.2024.1392716 ISSN=1664-302X ABSTRACT=Introduction

The accurate estimation of postmortem interval (PMI), the time between death and discovery of the body, is crucial in forensic science investigations as it impacts legal outcomes. PMI estimation in extremely cold environments becomes susceptible to errors and misinterpretations, especially with prolonged PMIs. This study addresses the lack of data on decomposition in extreme cold by providing the first overview of decomposition in such settings. Moreover, it proposes the first postmortem microbiome prediction model for PMI estimation in cold environments, applicable even when the visual decomposition is halted.

Methods

The experiment was conducted on animal models in the second-coldest region in the United States, Grand Forks, North Dakota, and covered 23 weeks, including the winter months with temperatures as low as −39°C. Random Forest analysis models were developed to estimate the PMI based either uniquely on 16s rRNA gene microbial data derived from nasal swabs or based on both microbial data and measurable environmental parameters such as snow depth and outdoor temperatures, on a total of 393 samples.

Results

Among the six developed models, the best performing one was the complex model based on both internal and external swabs. It achieved a Mean Absolute Error (MAE) of 1.36 weeks and an R2 value of 0.91. On the other hand, the worst performing model was the minimal one that relied solely on external swabs. It had an MAE of 2.89 weeks and an R2 of 0.73. Furthermore, among the six developed models, the commonly identified predictors across at least five out of six models included the following genera: Psychrobacter (ASV1925 and ASV1929), Carnobacterium (ASV2872) and Pseudomonas (ASV1863).

Discussion

The outcome of this research provides the first microbial model able to predict PMI with an accuracy of 9.52 days over a six-month period of extreme winter conditions.