SMART INFORMATION FLOW TECHNOLOGIES LLC — Department of Defense STTR Phase I: N16A-T002

Description

We propose CADET (Cognitive Assessment During Evaluation and Testing), a flexible and minimally intru-sive machine learning enabled software sensor that classifies environmental, human behavior, and physiolog-ical features to measure and predict near real time workload (WL) and performance. We will combine SIFT's prior WL research in zero-intrusion techniques in linguistics, keyboard dynamics and video-derived heart rate (which achieved WL classification accuracy levels between 66-100%), with Dr. Bradley Chase's research in eye movement parameters for WL identification. CADET will tailor and further extend these measures to in-clude interaction dynamics, posture, and prosody to increase operational coverage for a Test and Evaluation (T&E) environment. CADET will include a multi-dimensional array of metrics to provide WL estimates, em-ploying statistical models to consider in-task cognitive load, and allow for changes in an operators WL rela-tive to learning. Further, CADET will customize result visualizations pertinent to different stakeholders, providing a truly flexible and actionable measure of WL useful for a wide variety of T&E situations. Addi-tionally, such a WL assessment assay would be compatible with both team and individually performed tasks. Phase I will provide the necessary requirement and design specifications to validate these measures in Phase II.