Multimodal-multisensor Analytics for Detecting Anxiety Phases in Individuals Experiencing High Anxiety

This thesis aims to advance objective assessments of anxiety to address the drawbacks of current clinical assessments. It uses multiple methods, including semi-structured interviews, lab-based data collection, signal analysis techniques, and multimodal-multisensor analytics. In total, 147 subjects participated in qualitative and quantitative data collection studies. Its results detected high-anxious vs. low-anxious individuals, conceptualized four anxiety phases, and detected all those phases in 65% of high-anxious individuals by fusing three physiological and behavioral features; a 30% improvement compared to the best unimodal feature. Overall, this thesis is a fundamental contribution toward the long-term aims of minimizing the burden of anxiety disorders.