Priorities in behavioural epidemiology are shifting from examining the health impacts of individual activities (e.g., sitting, sleep or physical activity) divorced from the rest of the day, to examining the integrated impact of all daily activities.
Novel statistical methods based on compositional analysis now allow the inclusion of all daily activities in the same analytical model and enable us to determine relationships between 24-hour time use and health outcomes.
However, we are not yet able to describe optimal durations that should be spent in activities across a day for a single health outcome, let alone multiple health outcomes. This is critical as there is not much sense in encouraging people to sit less, sleep longer or exercise more if we don’t know what the optimal durations are.
It is also not necessarily helpful to optimise time use solely for one health outcome, because this may have negative effects on other valued health outcomes.
We are exploring new analytical methods by combining machine learning, optimisation theory and compositional data analysis.
We will apply these new methods to pooled data from large, population-based studies, allowing us to confidently optimise 24-hour time use: from individual health outcomes, and ultimately for overall health and wellbeing.
Our research will provide evidence to underpin 24-hour activity recommendations for both general and clinical populations, and to guide the design and implementation of behavioural interventions.
It will enable future evidence-based surveillance of time-use behaviours and robust economic evaluation of potential lifestyle interventions.