Fatigued or Ready to Train? - Modelling Availability to Train at Team GB Hockey via a Binary Time Series Mixed Model
Hannah Frick (UCL)
Friday 10th February, 2017 15:00-16:00 Maths 203
A topic of practical importance for athletes and coaches is monitoring fatigue in order to avoid non-functional overreach. Definitions of fatigue vary across sports science literature and no single definite fatigue marker has been identified. Teams and Clubs typically resort to monitoring training load and collecting data on multiple physiological and psychological indicators for their athletes. One outcome of non-functional overreach is athletes being unavailable for training, a crucial measures for coaches to assess if the prescribed training load was too high or not.
In this talk, I present joint work with Ioannis Kosmidis on modelling availability to train for athletes of Team GB Hockey. To account for heterogeneity between athletes and the time series structure of the availability measurements, we employ a binary time series mixed model. In its originally proposed form, the model allows to include auto-correlation in the availability to train and random intercepts for the athletes. Motivated by additional explanatory variables on psychological well-being, we extend this to a more general structure for the random effects to be able to include random slopes as well as team dynamics.