WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
Using machine learning on activity monitoring data, we have been able to measure new information on how UK Biobank participants move. For example, we find that men spend more time in both low- and high- intensity behaviours, while women spend more time in mixed behaviours. Walking time is highest in spring and sleep time lowest during the summer. This work opens the possibility of future public health guidelines informed by the health consequences associated with specific, objectively measured, physical activity and sleep behaviours.
Extracting Meaningful Health Information from Raw Accelerometer Data in UK Biobank
UK Biobank accelerometer data will allow more precise investigation into physical activity and its health associations. We propose to investigate:
1) Which participants are actually wearing the devices, and for how long
2) How much time is spent in sleep, sedentary behaviour, and physical activity
3) How much time is spent in specific behaviours of interest such as walking and cycling
4) What physical activity patterns can be characterised from accelerometer data
Participant level summary variables will be returned to UK Biobank. This will allow non-specialists worldwide to benefit from the information currently locked within Biobank's accelerometer data. UK Biobank aims to improve the prevention, diagnosis and treatment of life-threatening illnesses. This research is relevant as insufficient physical activity is associated with a range of chronic diseases and premature mortality. However these associations are based on self-reported data which is crude and prone to measurement error. Therefore, uncertainty exists on the level of physical activity people engage in, the exact amount and type of activity that should be recommended, and which interventions are most effective. This proposal is therefore critical as it will allow epidemiologists to more accurately explore physical activity and its associated health consequences. We will use published methods to convert raw accelerometer data into meaningful summary variables to explore the role of different physical activity patterns in health. Examples include time spent in: non-wear, sleep, sedentary behaviour, light and moderate-to-vigorous intensity physical activity. These methods will be computationally optimised so that each accelerometer file can be processed efficiently. Emerging automated (`machine learning`) methods will be used to extract specific behaviours such as walking from the raw data. These physical activity variables will be assessed across subgroups (such as age, gender, socio-economic status, and health status) using standard epidemiological techniques. We require access to the full cohort of participants in UK Biobank. Note that in addition to analysing accelerometer data in those who receive the device (currently ~29k but expected to be >100k), we wish to explore if their general characteristics are similar or not to the entire cohort.