Open wearables initiative highlights standards in open-source algorithms
The Open Wearables Initiative (OWEAR), which promotes and facilitates the use of open-source algorithms for digital medicine, has announced a new program designed to identify and support open-source algorithms that are being adopted rapidly and may become de facto industry standards.
This OWEAR program uses the number and quality of references to “crowdsource” the identification of potential de facto standards. Once a potential standard is identified, OWEAR will create and maintain a searchable catalogue that displays key information about the studies in which the algorithm has been employed, including the therapeutic area, population(s) studied, and the number of participants.
Geoffrey Gill, OWEAR co-founder and president of Shimmer Americas comments: ‘Establishing industry standards is critical in gaining widespread acceptance of digital medicine technologies. Regulatory agencies and healthcare professionals cannot be expected to evaluate myriad algorithms that measure essentially the same phenomenon. Meeting OWEAR’s reference-count criterion indicates that the authors trusted the algorithm enough to use it and that the software is in good enough shape to be used by others.’
OWEAR plans to create a new catalogue for each open-source algorithm used in a wearable sensor or other connected health technology that generates more than 25 published papers and meets other quality metrics. OWEAR searchable catalogues will provide a snapshot of, and links to, the published wearables research using open-source algorithms in multiple therapeutic areas. The reference-count criterion indicates that the algorithm has garnered significant industry adoption. Shimmer Research has agreed to donate the staff time to create and maintain these OWEAR catalogues.
GGIR, a popular package for transforming wrist-worn acceleration data into activity and sleep metrics, is the first open-source package recognized under this program. The OWEAR catalogue displaying all the published papers for GGIR is now available at www.owear.org. It features more than 300 references, including 114 that were published in the past year alone.
‘OWEAR is always looking for new ways to facilitate the use of open-source algorithms. Identifying and evaluating whether an open-source package is appropriate for a specific application can be extremely time-consuming. This program streamlines that process by identifying packages that have been used extensively and allows the user to display relevant references in seconds,’ states Mr Gill.