From sub-source to source: Interpreting results of biological trace investigations using probabilistic models

Abstract

The current method of reporting a putative cell type is based on a non-probabilistic assessment of test results by the forensic practitioner. Additionally, the association between donor and cell type in mixed DNA profiles can be exceedingly complex. We present a probabilistic model for interpretation of cell typing results. This Bayesian network can be used by forensic practitioners to form a substantiated opinion about the cellular material(s) that a sample is composed of and about the attribution of DNA donor to cell type.