Title
Physiology-Based User State Assessment in the Presence of Physical Activity
Meeting link: https://teams.microsoft.com/meet/475115634587746?p=gp1cIlVwmWTOVdS34g
Abstract
Wearable physiological sensing has enabled the assessment of cognitive and human factors during human–system interaction. However, much of the existing research has been conducted in controlled environments, where users remain stationary while performing relatively simple tasks. Real-world deployment introduces substantially greater complexity due to physical activity, motion artifacts, physiological confounding, and demographic variability. This talk presents approaches for developing robust physiology-based user state assessment systems under these real-world conditions. The work addresses challenges associated with disentangling cognitive state from the effects of physical activity and developing fair and generalizable models. It combines domain-knowledge-driven physiological feature extraction, signal processing, and artificial intelligence, including multi-objective learning strategies for deep neural networks. Finally, current challenges in real-time user state assessment will be discussed in the context of increasingly capable AI systems, spanning both software agents and embodied AI.
Bio:
Abhishek Tiwari is a Research Associate at the University of Quebec in Gatineau, where he works with the Cybersecurity and Digital Trust Unit. His current research spans cybersickness assessment in Virtual Reality, demographic bias in human factors models, and trust repair in Human–AI teaming. He also works as an Independent Research Consultant with the National Research Council of Canada, where he develops physiological signal-processing pipelines and integrates foundation models into research workflows.
Physiology-Based User State Assessment in the Presence of Physical Activity
Meeting link: https://teams.microsoft.com/meet/475115634587746?p=gp1cIlVwmWTOVdS34g
Abstract
Wearable physiological sensing has enabled the assessment of cognitive and human factors during human–system interaction. However, much of the existing research has been conducted in controlled environments, where users remain stationary while performing relatively simple tasks. Real-world deployment introduces substantially greater complexity due to physical activity, motion artifacts, physiological confounding, and demographic variability. This talk presents approaches for developing robust physiology-based user state assessment systems under these real-world conditions. The work addresses challenges associated with disentangling cognitive state from the effects of physical activity and developing fair and generalizable models. It combines domain-knowledge-driven physiological feature extraction, signal processing, and artificial intelligence, including multi-objective learning strategies for deep neural networks. Finally, current challenges in real-time user state assessment will be discussed in the context of increasingly capable AI systems, spanning both software agents and embodied AI.
Bio:
Abhishek Tiwari is a Research Associate at the University of Quebec in Gatineau, where he works with the Cybersecurity and Digital Trust Unit. His current research spans cybersickness assessment in Virtual Reality, demographic bias in human factors models, and trust repair in Human–AI teaming. He also works as an Independent Research Consultant with the National Research Council of Canada, where he develops physiological signal-processing pipelines and integrates foundation models into research workflows.
Previously, Abhishek was a Lead Data Scientist at Myant Inc. in Toronto, where he led a Canadian Space Agency-funded project on wellness monitoring and management for astronauts, Prior to this role, he was a MITACS Postdoctoral Researcher at Myant, working on psychophysiological modelling using smart textiles. He received his B.Tech. and M.Tech. degrees from the Indian Institute of Technology Kharagpur in Engineering Product Design and Manufacturing, with a specialization in Industrial Electronics, in 2016. He subsequently completed his PhD in Biomedical Signal Processing at the Institut national de la recherche scientifique (INRS), University of Quebec, under the supervision of Prof. Tiago H. Falk at the MuSAE Lab.
Host: Vishnu Narayan