Driving is one of the most cognitively and perceptually demanding activities of daily living, and changes in everyday driving behaviour can carry signal about underlying neurological and cognitive change. Our work in this area combines GPS, inertial, and in-vehicle sensing with machine learning to characterize naturalistic driving, develop digital markers of cognitive decline in aging populations, support evidence-based driving decision-making in dementia, and examine how concussion affects driving in adolescents and across the lifespan. Projects in this pillar: 1) Driving Decision-Making in Dementia 2) CCNA Driving and Dementia Team 3) Concussion & Driving Across the Lifespan 4) Naturalistic Driving Methods External Collaborations: 1) The DRIVES Project (Washington University in St. Louis — Babulal Lab) 2) ROAD-Skills (McMaster University — Vrkljan Lab)

The environments older adults inhabit — neighbourhoods, green spaces, road networks, and the built environment more broadly — shape daily behaviour and health trajectories in ways that are only beginning to be quantified. Our work develops geospatial and multimodal methods for measuring environmental exposure as it unfolds in daily life-space mobility, linking these dynamic measures to cognitive, physical, and mental health outcomes in aging populations. We contribute scalable computational tools — including satellite-derived greenspace quantification, route-based exposure modelling, and accessibility analysis — that bring geographic information science into close dialogue with health and aging research. Projects in this pillar: 1) Greenspace Exposure 2) Built Environment & Driving Behaviour 3) Computational Urban Accessibility

Cognitive aging is heterogeneous: individuals with the same clinical diagnosis can follow very different trajectories of daily function, and population-level statistics often obscure what matters most at the individual level. Our work develops individualized, longitudinal digital phenotypes — constructed from passively collected mobility, gait, sleep, and activity data — to detect early, person-specific change along the cognitive risk continuum, from cognitively healthy aging through mild cognitive impairment to Alzheimer's disease and related dementias. We are particularly interested in multimodal sensor fusion, person-specific modelling, and methodological approaches for working with small, deeply-phenotyped cohorts. Projects in this pillar: 1) Digital Phenotyping of Cognitive Aging 2) Life-Space Mobility & Cognitive Aging 3) FIT from GAIT — Sensor-Driven Frailty Monitoring External collaborations: 1) Digital Phenotypes of Healthy and Pathological Cognitive Aging (University of Zurich — Langer Lab, SNSF-funded) 2) Smart Insole / Aging in Place (Université de Montréal & UofT — Tannou and Wang Labs)

AI for aging must move beyond predictive accuracy alone. It requires models that are equitable, interpretable, robust across heterogeneous populations, and responsive to the real-world contexts in which technologies are encountered, understood, and used by older adults, including those experiencing cognitive change. Our work develops bias-aware machine learning methods for health data, evaluates generative AI approaches for personalized dementia care, and advances methodological frameworks for designing, validating, and translating AI systems for older adult populations. This is particularly critical because older adults, especially those with cognitive impairment, multimorbidity, disability, and diverse social or cultural backgrounds, remain systematically underrepresented in the datasets used to train many contemporary AI models. Projects in this pillar: 1) Digital Stories with Generative AI for Dementia 2) Trustworthy & Fair Machine Learning for Health 3) Foundation Models for Aging