B1638
Title: Doubly online changepoint detection for monitoring health status during sport activities
Authors: Mattia Stival - University of Padova-Dipartimento di Scienze Statistiche (Italy) [presenting]
Mauro Bernardi - University of Padova (Italy)
Petros Dellaportas - Athens University (Greece)
Abstract: An online framework is provided to analyze data recorded from smartwatches during running activities. In particular, we focus on identifying variations in the behavior of one or more measurements caused by physical condition changes such as physical discomfort, periods of prolonged de-training, or even malfunction of measuring devices. The framework considers data as a sequence of running activities, where one activity is a time series collecting over time physical and biometric data. We combine classical changepoint detection models with an unknown number of components to Gaussian state-space models to detect distributional changes between a sequence of activities (multivariate time series). The model considers multiple sources of dependence due to the sequential nature of subsequent activities, the autocorrelation structure within each activity, and contemporaneous dependence between different variables. We provide an online Expectation-Maximization (EM) algorithm involving a sequential Monte Carlo approximation of changepoint predicted probabilities. As a byproduct of our model assumptions, the proposed approach processes sequences of multivariate time series in a doubly online framework. While classical changepoint models detect changes between subsequent activities, the state space framework coupled with the online EM algorithm provides the additional benefit of estimating real-time probabilities that a single activity is a changepoint.
