The use of chemical forensic impurity profiling for estimating the strength of evidence for source attribution are reported. The strength of evidence was measured as a likelihood ratio. The relative integrated areas for a suite of 14 impurity profiles were employed to calculate similarity scores between DC samples taken from the same and different production batches. Models for discriminating between batches were developed and cross validated against the DC data. The models were then challenged with DC samples coming from different bottles of a single DC batch used in the model development and from Sarin (GB) samples prepared from two different DC batches. The performance of each models was characterized by equal error rates, empirical cross-entropy plots and extent of misleading evidence based on Tippett plots, and receiver operating characteristic plots.

Score based likelihood ratio approaches to chemical profiling of methylphosphonic dichloride (DC) and derived products

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.