Use of Bayesian Networks for the investigation of the nature of biological material in casework
Lydie Samie, Christophe Champod, Séverine Delemont, Patrick Basset, Tacha Hicks, Vincent Castella
Abstract
Chemical and staining methods, immunochromatography, spectroscopy, RNA expression or methylation patterns, do not allow to determine the nature of the biological material with certainty. However, to our knowledge, there are few forensic scientists that assess the value of such test results using a probabilistic approach. This is surprising as it would allow account for false positives and false negatives and avoid misleading conclusions. In this paper, we developed three Bayesian Networks (BNs) to assess the presence of blood, saliva and sperm in the recovered material and combine potentially contradictory observations. The approach was successfully tested using 188 traces from proficiency tests. We have implemented an online user-friendly application (https://forensic-genetic.shinyapps.io/BodyFluidsApp/) that allows forensic scientists to assess the value of their results without having to build Bayesian Networks themselves. They can also input their own data, use the application to identify a potential lack of knowledge and report their conclusions regarding the presence of sperm, blood or/and saliva considering uncertainty.