Sports analytics and modeling of wearable-sensor data
Wearable sensors and portable devices such as global-positioning system watches, accelerometers, and bicycle-mounted power meters are used to collect unprecedented observational data from exercising individuals. An obstacle to translating these data into actionable insights is a lack of algorithms for extracting key information.
We are working to devise improved metrics and algorithms for exercise and physical activity monitoring devices. Ongoing projects include improving existing models such as Dr. Eric Banister's impulse-response model (developed here at SFU ~50 years ago), extending the critical power model (Puchowicz et al. 2020), and constructing hybrid physiological-statistical models that reflect both the physiology of the individual and additional sources of variance known to impact the signal. A critical aspect of the work is to ensure that the model outputs are interpretable by end-users and therefore enhance their ability to make decisions based on the data.
Our work in sports analytics is described in a feature story on the SFU Key Big Data Initiative website.
As a result of the expertise we have developed, Dr. Clarke has been tasked to provide guidance on the implementation of data science within the Canadian high-performance sport system.
Dr. Clarke and his group are core members of the SFU Sports Analytics Group, a collection of researchers, students, and coaches that are interested in applying data and modeling to improve sports performance. Currently, Dr. Clarke organizers the SFU Sports Analytics Group virtual seminar series.
Collaborators: SFU Sport Analytics Group, Canadian Sport Institute Pacific, Michael Puchowicz (Arizona State University), Nathan Townsend (Hamad Bin Khalifa University, Qatar), Philip Skiba (Advocate Lutheran General Hospital, Chicago USA).
Funding: SFU Key Big Data Initiative, Mitacs, Own The Podium, Canadian Sport Institute Pacific