Microbiological analysis in forensic identification and machine learning: a review

Abstract

Microorganisms that share human body space and live in the surrounding environment carry specific microbial signatures in each individual and location. They interact continuously with the surrounding environment and this interaction may provide useful information. In the body of a deceased, microorganisms are called as thanatomicrobiome. This fact presents a promising alternative tool for human identification from the traditional identification methods (fingerprint and DNA analysis) in cases where the samples for those methods are unattainable. Microbiological analysis introduces the possibility of using microbiomes as identifiers in forensic investigation and their values can be directed into four aspects; PMI inference, individual identification, tissue or fluid identification and geolocation inference. Rapid technological advancement in artificial intelligence has been widely applied to a lot of disciplines including forensic science. The use of artificial intelligence enhances the accuracy of microbial identification. Numerous machine learning frameworks have been created such as RF, SVM, and ANN alongside the expansion of metagenomic databases. Along with this, concerns regarding limitation and ethical implication of the AI application in forensic microbiology arise. As more research are conducted, people may become aware of the importance of microbiome data and their privacy. There are also concerns regarding AI-generated analysis reports being used in judicial processes because of how inexplicable the way AI works. This paper summarizes the application of microbiological analysis in forensic identification and development of forensic microbiology with the aid of artificial intelligence, specifically the machine learning technique with few examples leaning towards forensic odontology (samples from oral cavity). 

Keywords: machine learning, forensic identification, microbiological, artificial intelligence