Advancing Statistical Methods for the Analysis of Complex Biologging Data Collected from Humans and Animals

> français

Collaborative Research Team Project 22: 2022–2025

Technological advancements have revolutionized how we study and monitor the movement, behaviour, and health of humans and animals in their free-living conditions. Biologging instruments now allow us to monitor an individual’s movement (e.g. position and/or acceleration), physiology (e.g. heart rate, glucose level), and environment (e.g. ambient temperature/light) in places that were rarely available to researchers and clinicians before. Commercial wearable devices (e.g. fitbits, oura rings, apple watches) are used in health, sport, and behavioural sciences to understand a broad range of human behaviours and physiological processes (e.g. sleep, heart rate), often with the goal of improving clinical outcomes (e.g. improving prevention and intervention) by developing algorithms that automatically detect and predict health events. In ecology and veterinary sciences, specialized versions of these biologging instruments (e.g. GPS collars, depth sensors) are used to understand how the behavioural states of animals change through space and time, detect the environmental variables that are important to animals, or identify changes in condition that could be indicative of sickness.

Although siloed in their respective disciplines, medical and ecological researchers both collect similar biologging data, and their studies often share common analytical goals and challenges. First, wearable devices used by patients and animals often collect multidimensional, high-frequency and highly correlated time-series data that are difficult to process and analyse. Second, the goals of both medical and ecological researchers often involve understanding the underlying behavioural or physiological processes and using these time-series to identify when behaviours or physiological states are exhibited. Lastly, both fields recognize the need to capture behavioural and physiological processes occurring at multiple scales for identification of a wider range of behaviours and medical conditions. The increasing ubiquity of these biologging instruments, and the shared goals and statistical challenges faced by those that use them, highlight a need to join forces. Our CRT will advance the development and application of statistical methods to biologging data by bringing together researchers across the fields of statistics, ecology, and medicine, including several government scientists and practitioners. In particular, we will use the hidden Markov model framework to bridge these distinct human and animal research fields.

Team leaders:

Vianey Leos Barajas – University of Toronto
Marie Auger-Méthé – University of British Columbia

Collaborators:

Joanna Mills Flemming – Dalhousie University
Nancy Heckman – University of British Columbia
Alán Aspuru-Guzik – University of Toronto
Catalina Gomez Salazar – Fisheries and Oceans Canada
Nigel Hussey – University of Windsor
Shelley Lang – Fisheries and Oceans Canada
Marianne Marcoux – Fisheries and Oceans Canada
Juan Morales – Universidad Nacional del Comahue
Abigail Ortiz – University of Toronto
Yannis Papastamatiou – Florida International University
Andrew Trites – The University of British Columbia

Comments are closed.