AI System mHolmes Accurately Estimates Time of Death Using Microbiome with Less Than Two Days Error

Introducing mHolmes: A Breakthrough in Forensic Time of Death Estimation

According to НВ — Техно: Scientists have unveiled mHolmes, an artificial intelligence system that determines the postmortem interval by analyzing the body's microbiome. By tracking microbial changes on deceased bodies, mHolmes can estimate the time since death with an average error margin of under two days. This advancement, detailed in the journal Nature Communications, marks a significant leap forward in forensic science.

Study Design and Key Findings

The AI was trained using openly available data from 34 human cadavers, with daily microbiome samples collected for 21 days postmortem. Impressively, mHolmes can predict the time since death for one body site based on data from another, maintaining an average prediction error below two days across multiple regions. This accuracy surpasses earlier methods, which typically had about a three-day margin of error. Additionally, the system can reconstruct the likely progression of microbial communities during the first week after death.

Researchers identified seven distinct bacterial groups linked to different stages of decomposition, including:

• Gammaproteobacteria, which flourish during the early decomposition phase;
• Clostridia, dominant in the active decay stage;
• Deinococci, present during the dry stage of decomposition.

Co-author Kan Nin likened mHolmes to "a meteorologist for microbes," explaining that the system forecasts microbial shifts to accurately estimate time since death.

Despite promising results, the researchers acknowledge the study's limited dataset size. To advance mHolmes toward practical forensic use, further validation in real-world scenarios and standardized sampling protocols are needed. This innovative approach holds the potential to greatly enhance time of death estimations and support criminal investigations.

Traditional methods for determining time of death often lack precision, sometimes resulting in substantial errors. By leveraging microbial analysis, mHolmes opens new avenues for obtaining more accurate timelines crucial to forensic cases. Ongoing research and operational testing will be essential to confirm the system’s reliability and facilitate its integration into forensic practice.