A Machine Learning Approach for Using the Postmortem Skin Microbiome to Estimate the Postmortem Interval

Area of Science:

• Forensic Science
• Microbiology
• Bioinformatics

Background:

• The human microbiome plays a critical role in health and disease.
• Estimating the postmortem interval (PMI) is crucial in forensic investigations.
• Current PMI estimation methods have limitations, especially in uncontrolled environments.

Purpose of the Study:

• To investigate the potential of the skin microbiome as a tool for estimating PMI.
• To develop a predictive model for PMI using microbial data from decomposing cadavers.
• To establish a machine learning approach for forensic application of necrobiome data.

Main Methods:

• Sampling of skin microbiomes from nasal and ear canals of decomposing human cadavers.
• Development of statistical regression models to predict PMI.
• Utilizing a k-nearest-neighbor regressor trained on comprehensive microbial data.

Main Results:

• The complete dataset, not just indicator species, improved model training.
• Microbial genus and family levels were more informative than species level for PMI prediction.
• The developed k-nearest-neighbor model predicted PMI with an average error of ±55 accumulated degree days (ADD).

Conclusions:

• Skin microbiota is a promising indicator for estimating PMI in forensic death investigations.
• A machine learning approach using necrobiome data provides a viable method for PMI prediction.
• This study presents a successful proof-of-concept for microbial analysis in forensic science.