Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.
Fitness monitor 2

One of today’s great health challenges is physical activity: the lack of it is associated with around five million deaths worldwide each year. The Nuffield Department of Population Health (NDPH) cohort studies follow tens of thousands of people, tracking lifestyle and health over years. The huge data resource created is then analysed to find associations between health issues and a range of factors – including physical activity. One issue is that asking people to record the physical activity they do can give inaccurate results. Technology may have an answer.

Dr Aiden Doherty is part of the team at NDPH, bringing his expertise as a computer scientist to bear on the large-scale health research programmes run by the department. Dr Doherty has a range of devices used to get more objective measures of activity. These include wrist-worn accelerometers – ‘Think of it as a research-grade Fitbit’ – and wearable cameras. Already, over 100,000 participants in the Oxford-led UK Biobank population study have worn the devices, and 200 additional volunteers around Oxford have worn the cameras. The images offer better understanding of the readings from the accelerometers, showing whether people are cycling or walking, for example.

Aiden explains: ‘At the moment, health advice is based on gross estimations. The Chief Medical Officer recommends 150 minutes of moderate to vigorous intensity exercise each week, but there are no guidelines on light activity, and little on the consequences of too much sitting. We don’t know whether it should be regular daily activity or if you get the same effect from being a weekend warrior. The new measurement technology will help us in future, informing health guidelines and helping other researchers – for example, if interested in the genetics of obesity, it is important to know about physical activity.’

The technology makes a significant difference – studies found that in the US 50% of adults thought they met the minimum 150 minutes of moderate exercise, while in England 38% did. In both countries, the technology suggested just 5% were hitting the target.

The challenge is the amount of data gathered. In just one week, wrist-worn devices return tens of millions of movement readings, while cameras produce tens of thousands of images. That is where Aiden’s expertise is vital. He creates the processing techniques that can turn these millions of data points into information useful for health researchers. For the UK Biobank, an analytical task that could have taken almost 3½ years was reduced to a matter of days.

Following a degree in Computer Science from Dublin City University, Aiden stayed on to complete a PhD on an Irish government research fellowship. He chose to look at the then-emerging field of wearable technologies and ‘life-logging’. Supervised by Professor Alan Smeaton and supported by leading tech companies, he looked at how best to process the data from a lifelog of images collected by a wearable camera. Then, in 2010, he received a prestigious Marie Skłodowska-Curie actions Research Fellowship from the European Union, which enabled him to come to Oxford and apply his knowledge to health issues.

‘It was an opportunity to have a more applied focus,’ he says. ‘Population health works on the most important problems. The overall aim is to better measure lifestyle health behaviours. We’re only scratching the surface of what we can do. As the technology improves, there will be the opportunity to include it in more studies.’

Now supported by a fellowship from the British Heart Foundation Centre for Research Excellence at Oxford, Aiden adds: ‘The key thing for me is to continue learning. The Big Data Centre opens in January and I’ll be working with people in genomics, engineering, population health and across medical sciences. Five years ago I was not hopeful about automated analysis of sensor and image data in health datasets. Now, at least semi-automated analysis is possible, and I am very excited about the advances my colleagues and I are making.’