In most clinical trials, participants have to make regular trips to a clinic where medical staff carry out a battery of tests to gather data for the trial. These take up time, cost a lot of money, and only provide snapshots of that person’s health at that specific moment in time. But what if some of this information could be gathered using a standard, off-the-shelf fitness tracker connected to a smartphone, as people go about their daily lives? That’s the question a team from IMI project BigData@Heart set out to address.
They embedded a consumer wearable device into part of a randomised clinical trial on a heart condition called atrial fibrillation (AF). Patients with AF have an irregular, often fast heart rate, and the number of people with AF is increasing rapidly. The RATE-AF trial (RAte control therapy Evaluation in permanent Atrial Fibrillation) was designed to compare two commonly-used treatments (beta blockers and digoxin) and assess their effect on quality of life, functional status and control of heart rate.
In a sub-study of the main trial, 53 trial participants used a fitness tracker to gather information on their heart rate and physical activity. Over a period of 20 weeks, more than 140 million data points for heart rate and 23 million data points for physical activity were collected. These were made available to the researchers via the open-source RADAR-Base platform, which allows for automated and secure gathering and integration of data from different devices. RADAR-Base is an output of another IMI project, RADAR-CNS. In parallel, the trial participants continued to undergo the same in-person tests at the clinic as the rest of the RATE-AF participants (such as electrocardiograms to look at heart rate and six-minute walk tests for physical activity).
The BigData@Heart team analysed the information, creating new artificial intelligence (AI) techniques to account for missing data, avoiding the over-optimistic view of AI that is usually presented in similar studies. This was based on a previously published BigData@Heart framework for transparent AI that aims to help researchers globally to improve the application of AI in healthcare.
On the medical side, the trial revealed that digoxin and beta-blockers have a similar effect on heart rate, even when differences in physical activity are taken into account. Crucially, the results from the wearables were equivalent to the results from the standard tests the trial participants went through in the clinic.
Presenting the results in Nature Medicine, the team notes that their approach “could lead to wearable devices contributing to, or even replacing resource-intensive clinical tests and visits in the future”. What’s more, data from wearables “may better reflect the real-life, day-to-day variations in heart rate and physical activity”.
The study should also allay fears that trials involving wearables might be impractical in older age groups; the average age of the participants was 76 years, and many had never used a smartphone before.
“People across the world are increasingly using wearable devices in their daily lives to help monitor their activity and health status,” said the lead author of the study, Professor Dipak Kotecha at the University of Birmingham in the UK. “This study shows the potential to use these new technologies to assess response to common treatments, and make a positive contribution to the routine care of patients.”
BigData@Heart and RADAR-CNS are supported by the Innovative Medicines Initiative, a partnership between the European Union and the European pharmaceutical industry.