Title: Scalar-on-function regression and functional isotemporal substitution analysis in the context of time-use data
Authors: Paulina Jaskova - Palacky University Olomouc (Czech Republic) [presenting]
Karel Hron - Palacky University (Czech Republic)
Javier Palarea-Albaladejo - University of Girona (Spain)
Ales Gaba - Palacky University Olomouc (Czech Republic)
Zeljko Pedisic - Victoria University (Australia)
Dorothea Dumuid - University of South Australia (Australia)
Abstract: How people allocate their daily time to physical activities (PA), sedentary behaviors (SB) and sleep have important health implications. PA (categorized by intensity into light, moderate and vigorous PA), SB and sleep should not be analyzed separately, because they are parts of a time-use composition with a natural constraint of 24 hours per day. To find out how relative reallocations of time between PA of various intensities are associated with health, we describe compositional scalar-on-function regression (CSFR) and a newly developed functional isotemporal substitution analysis (FISA). Instead of working with the ordinary PA categorization, the original data consisting of PA intensity distributions can be characterized as empirical probability density functions (PDFs), which better reflect the continuous character of their measurement using accelerometers. These PDFs have specific properties, such as scale invariance and relative scale, and they are geometrically represented using Bayes spaces. The CSFR of explanatory PDFs and FISA were applied to a dataset from a cross-sectional study on 24-h movement behaviors and adiposity conducted among school-aged children in the Czech Republic. Theoretical reallocations of time to PA of higher intensities were found to be associated with larger decreases in adiposity. We gained detailed insight into the dose-response relationship between PA intensity and adiposity, which would not be feasible using an ordinary approach.