Grant: Canada Postdoctoral Research Award (CPRA) program

Alzheimer's disease (AD) affects more than 700,000 Canadians and is projected to impact 1.7 million by 2050. Despite advances in research, early diagnosis remains limited because current biomarker tests are invasive, costly, or not widely available. This project aims to develop non-invasive, affordable, and scalable tools for early detection and monitoring of AD by combining smartwatch sensor data with blood-based biomarkers using artificial intelligence. We hypothesize that physiological and behavioural signals collected from smartwatches, such as heart rate variability, sleep, activity, and skin conductance, may help predict levels of blood markers related to AD, including amyloid, tau, and other proteins involved in brain health. In this multi-site study, we will collect data from older adults (aged 50-90) in Kingston (Ontario, Canada) and Brazil. Participants will include both cognitively healthy individuals and those with probable AD. Each participant will be assessed every three months over one year (five visits total). At each timepoint, we will collect blood samples and measure brain-related proteins using ultrasensitive methods, such as SIMOA and Olink proteomics. Participants will also wear Fitbit Sense 2 smartwatches continuously for passive monitoring, and complete digital tasks to assess memory and attention. Our goal is to test whether smartwatch data can be used to estimate biomarker levels and cognitive performance using AI-based models. If successful, this would offer a proof of concept for software that passively tracks AD-related changes, potentially enabling earlier, more accessible, and more equitable detection of dementia risk across diverse populations.