We are creating a unified UKRI website that brings together the existing research council, Innovate UK and Research England websites.
If you would like to be involved in its development let us know.

Site search
Back to blog

Smartphones for healthcare research: data at our fingertips

by Guest Author on 10 Dec 2019

Anna Beukenhorst, a PhD student at the University of Manchester, was commended in this year’s Max Perutz Science Writing Award. She sets out the challenges of studying the link between the weather and arthritis symptoms – and how collecting smartphone data from patients could help them better manage their pain.

with MRC Executive Chair Professor Fiona Watt

Anna with MRC Executive Chair Professor Fiona Watt

When motorcycle legend Barry Sheene moved to Australia, it was not for another Grand Prix victory. It was for the weather. Sheene had retired from racing because of his arthritis. Like many arthritis patients he believed that the damp British weather worsened his arthritic pains.

Barry Sheene’s hypothesis has never been proven, even though many researchers tried to. Ideally, we would like to place arthritis sufferers in a controlled experiment, with rain on one day and sunshine on the next. We would measure their pain, while holding everything else constant. Statistics help us determine what weather conditions (such as temperature, or humidity) aggravate pain. Alas, such an experiment is not possible, though. We have to rely on observing people in their daily lives. Patients have to record their pain and the weather in diaries to pinpoint the association. These observational studies come with challenges. We can’t hold everything else constant, so how do you know it is really the weather changing someone’s pain? The culprit can also be a third factor, like medication, mood, or physical activity. We can only unpick the link between pain and weather by observing many participants for a long time, and record the potential third factors alongside pain. Previous studies had too few participants, too short time frames or little information on third factors to conclusively underpin Barry Sheene’s hypothesis.

We wanted to have another go and started Cloudy with a Chance of Pain. We hoped we could get better daily life data from…people’s pockets. Most people carry a smartphone wherever they go. So we made an app: Cloudy with a Chance of Pain. Reporting pain only takes one swipe. In ten seconds participants also record their mood and physical activity. Smartphones know the location of their owners, which allows us to link pain reports to accurate weather data, even for commuters or frequent travellers.

Our plan worked. Our study, with 13,000 participants, is ten times bigger than the biggest previous study. Our participants reported 5 million symptoms during 15 months, linked to hourly weather reports from the closest Met Office station. My role was analysing this humongous dataset. I investigated where our participants were from (every postcode from Land’s End to the Orkneys), on how many days they reported their pain (our most faithful participant for 457 days), how their pain changes during the week (Sundays are least painful).

Sometimes, smartphone data caused new problems. I discovered that iPhone users had many gaps in their location data. To link their pain reports to the right weather station, I had to find a way to fill those gaps. Luckily, most people have strong habits: depending on the day and time, they are at home, in the office or at the supermarket. If participants had a location gap, I let the computer analyse where they usually were at that day and time, and used the weather from that location.

To link the weather and pain, I use the so-called ‘case-crossover design’. It is the statistification of common sense. You may have inadvertently used it in your own life. Have you ever had a headache and wondered what triggered it? Maybe you asked yourself: “What did I do differently, compared to days that I feel fine?” I do the same for each of our participants. I look at days of high pain and quantify what was different in the weather compared to days that they felt fine.

The first results look promising. It would be fantastic to unravel the age-old question of the weather and pain. Our participants tell me it would help them manage their pain better. On days with high pain, they often can’t drive their car, go to work or even walk. Even if patients cannot move to Australia, they can anticipate ‘being under the weather’. They can take medication, avoid activities that are heavy on their joints or plan in extra physiotherapy. Some participants suggested that the NHS could cover holidays to places with very un-British weather based on our results, but with 18 million arthritis sufferers that may not be feasible.…

Cloudy with a Chance of Pain has a second result. We have shown the potential of smartphones for research: better data from the daily lives of thousands of patients. Even though mobile health has been in newspapers for years, our study is the first that succeeded in getting symptom and location data from so many participants on so many days. Hopefully, this will inspire other researchers, doctors and patients to build their own apps. The methods I showcased will help them make the most of their big datasets. Smartphones will help us answer many more healthcare questions, now we finally have daily life data at our fingertips.

Read the paper How the weather affects the pain of citizen scientists using a smartphone app

Check out Cloudy with a Chance of Pain on BBC Breakfast



No comments have been posted

Leave a reply

You may use basic HTML in your comments. Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.


From category

Share this: