A study using machine learning, optimization algorithms, and data from the San Francisco Police Department Criminalistics Laboratory to examine the cost-effectiveness of processing DNA samples in a sexual assault kit finds that using the optimization framework to choose samples deemed most likely to yield useful evidence can increase the yield of DNA profile matches in criminal databases by 45.4% without increasing cost; processing all samples in a kit, however, is only slightly less cost-effective than a selective approach but can more than double the yield of positive DNA profile matches, according to the authors.
###
Article #20-01103: "A cost-effectiveness analysis of the number of samples to collect and test from a sexual assault," by Zhengli Wang, Kevin MacMillan, Mark Powell, and Lawrence M. Wein.
MEDIA CONTACT: Lawrence M. Wein, Stanford University, CA; e-mail: lwein@stanford.edu
Proceedings of the National Academy of Sciences