The WEALTH project aims to develop machine learning models to accurately estimate behaviours from research grade and commercially available wearable accelerometer devices so that easy to apply wearables can measure the daily time spent in active behaviours. A second aim of the project is to test the feasibility of linking estimated behaviours from accelerometry with context data obtained using Ecological Momentary Assessment (EMA) in which participants respond to questions through their smartphone. This will produce an integrated data collection system that will simultaneously capture activities and eating episodes.

To develop and validate machine learning models, 600 participants will be recruited to obtain lab-based and particularly data under real-life conditions by wearing a range of state-of-the-art accelerometer devices including wearables for 9 days to provide time series data of 24 hour behaviours focusing on raw accelerometry data. We further collect context data on physical behaviours and eating episodes using Ecological Momentary Assessment (EMA) to label time stamps during the free-living assessment period.

The EMA applied in WEALTH will provide a method of passively assessing the real-time impact of physical behaviours such as prolonged sedentary time and periods of physical activity on eating behaviour. This is an important development, because existing tools generally measure the two behaviours separately. The ability to target specific activities (for example, evening-time sedentary behaviour) for study, and to deliver EMA based questions during the behaviour will be important in future studies where the interaction of behaviours is critical. The combination of raw accelerometry data from wearable devices and EMA context data will deliver a measurement solution linking 24h behaviours and their context to eating episodes. Tools provided by WEALTH will include the assessment via wearables and fitness trackers providing low-cost wrist-worn movement-sensors.