Bayesian network tool for analyzing the cost-effectiveness of bulk forensic trace DNA profiling

Introduction of DNA profiling into the forensics toolkit has revolutionized police work and enabled solving numerous high-profile investigations, sometimes decades after the crimes were committed. However, as the demand for DNA profiling has increased with the availability of scalable technology, continuous bulk analysis of samples still implies challenges for forensics laboratories. To aid the forensic process management, we developed a Bayesian network as a novel tool for quantitative decision-making at forensic DNA laboratories. Using this network, we analyzed the cost-effectiveness of bulk trace DNA profiling on a dataset of over 21,000 analysis results of DNA samples from the National Bureau of Investigation Forensic Laboratory (NBI-FL) in Finland collected over the span of 1 year. Our results show that, in the context of Finland, filtering out samples based on their type and the quantity of extracted DNA enables a substantial reduction in processed sample volumes with relatively low risk of missing usable DNA profiles or matches in a database search. We show that DNA from touch samples or samples with low quantities of DNA has a low probability of producing a usable profile and that, under mild assumptions, processing thereof is unlikely to be cost-effective in routine workflow. This is important as such samples comprise the majority of the analyzed material at the NBI-FL. This suggests forensic sample processing protocols could be redesigned for a higher efficiency of DNA profiling for investigative purposes. Furthermore, our model provides a template for other forensic laboratories to apply similar analyses to their workflow.