This DHS SBIR Phase I project will lay the groundwork for the development of a Machine Learning (ML) system capable of instantaneous identification of unknown chemical and biological agents.to reduce threats of bioterrorism and chemical attacks, this effort will strengthen DHS capabilities to rapidly detect a wide range of hazardous agents.Since the DHS BioWatch program was established in 2003, threat detection technologies have evolved to become more portable, yet these innovations have not been fully leveraged to support the DHS mission to enhance Chem/Bio Detection and Defense technologies.Analytical instruments that conventionally were large laboratory instruments have been miniaturized and implemented into portable and handheld devices for on-site chemical analysis. Infrared (IR) spectroscopy, Raman spectroscopy, and mass spectrometry (MS) all have portable instruments that are commercially available.The proposed ML system will be designed to facilitate integration with such commercial devices to ultimately perform molecular structure recognition/prediction based on spectral data.This effort will investigate the feasibility of using IR, Raman and MS spectral data to train ML algorithms to recognize characteristic spectral patterns for a group of known organophosphate nerve agents.These compounds were selected because they share the same phosphorus-based structural "backbone" but differ in certain atoms bound to the central phosphorous atom.The result will be an ML system trained to recognize such molecular characteristics in spectral signal patterns to identify the aspects of a molecule needed to elucidate an unknown molecular structure.
