Detection of Chemical Warfare Agents by Colorimetric Sensor Arrays.

We report the successful use of colorimetric arrays to identify chemical warfare agents. Methods were developed to in-terpret and analyze a 73-indicator array with an entirely automated workflow. Using a cross-validated first-nearest-neighbor (1-NN) algorithm for assessing detection and identification performance on 632 exposures, at 30 min post-exposure we report, on average, 78% correct chemical identification, 86% correct class-level identification, and 96% correct red light/green light (agent versus non-agent) detection. Of 174 total independent agent test exposures, 164 were correctly identified from a 30 min exposure in red light/green light context, yielding a 94% correct identification of CWAs. Of 149 independent non-agent exposures, 139 were correctly identified at 30 min in a red light/green light context, yielding a 7% false alarm rate. We find that this is a promising approach for the development of miniaturized, field-portable analytical equipment suitable for soldiers and first responders.