Wearables group seeks collaboration on decentralized tech
Founded in September 2019, the Open Wearables Initiative (OWEAR) has worked to bring clinical research professionals together to hammer out solutions involving wearable sensors and other decentralized trial technology. Now, at the beginning of its second year, the group is considering expansion of its mission, in an effort to seek broader industry input and greater consensus.
Outsourcing-Pharma (OSP) discussed OWEAR’s work with Geoffrey Gill (GG), cofounder of the group and president of Shimmer Americas. He touched upon the initiative’s work to date, as well as its plans for the road ahead.
OSP: Could you tell me a little bit about OWEAR, especially about the progress made since its founding?
GG: When we first started OWEAR, we had a specific vision – to create a single location where people could access open source code for algorithms that analyzed data from wearable sensors. We have achieved that vision and an index of wearable algorithms has been up on the www.OWEAR.org website since May 2020, but we have done a lot more.
One of our first steps was to establish a Working Group to help guide the initiative. This group has representatives from five of the largest pharma companies, a top clinical research organization, and several leading industry organizations.
Based on feedback, we soon expanded our vision to include indexing open data sets of wearable data and benchmarking algorithms. The data set index also launched in May 2020 and we are in discussions with a number of pharma companies to sponsor our benchmarking programs which are based on the Dream Challenge process pioneered by one of OWEAR’s founders, Sage Bionetworks.
OSP: Why is the group important?
GG: Few people dispute that digital technologies have the potential to revolutionize healthcare and perhaps even bend the cost curve. To achieve this potential, we need to establish generally accepted digital indicators of health – sometimes called clinical endpoints.
If we cannot agree on whether someone is getting better or worse and how to measure it, it is difficult to see how we can make decisions on how to treat a patient. OWEAR is working to address this challenge through collaboration and sharing of software and data.
The reason that this is such a challenge is that we are measuring things like activity and gait that have never been measured continuously before. Today, almost all wearable devices rely on their own proprietary algorithms to arrive at these measures; the problem is that these proprietary algorithms arrive at different results.
Take a familiar example, such as step counts. Everyone who has worn a step counting device knows that they can be inaccurate. But perhaps more importantly, different products can come up with wildly different results – varying by up to 40% in one study.
Even different versions of the same products give different results. This is not because the sensors are different – they all use accelerometers that are roughly equivalent – it is because the algorithms used to process the sensor data are different, but for a measure to be useful in healthcare, it needs to be carefully validated.
With the proprietary models, each algorithm/wearable needs to be validated independently. But with thousands of potential clinical endpoints and hundreds of wearable devices available, that approach would require literally hundreds of thousands of expensive validation studies.
OWEAR is working to establish open source algorithms that anyone can use; that way the algorithm only needs to be validated once and the wearable only needs to prove that it is producing appropriate data. This transforms an insuperable problem into something manageable. Plus, it promotes collaboration because everyone can share in the benefits of open source.
OSP: How has collaboration between companies that sometimes could be considered competitors worked?
GG: All of the different parties involved with OWEAR are incredibly collaborative in our work together. Everyone recognizes the importance of the work and the immensity of the challenge. It is clear that no one company can solve this challenge on their own.
Furthermore, the work we are doing is truly pre-competitive. No one, aside from some wearables companies, has any thought of building a business because they have better algorithms than other companies. Everyone will benefit from our success.
OSP: What are some of your goals for the short- and long-term future?
The long-term goals are easy – to accelerate the adoption of digital medicine through collaboration and sharing. Having rapidly achieved our initial targets, we need to figure out where our next effort can have the greatest impact. That is our current focus.