A patient-developed open-source app linked to glucose monitoring devices delivers better control for type 1 diabetics than traditional FDA-approved tools, highlighting the need for regulatory reform to keep up with rapid technological innovation.
For people with type 1 diabetes, the appeal of software built outside the traditional medical device system is easy to understand. One health economist who has managed the condition for more than three decades says an open-source app linked to his glucose monitor and insulin pump has delivered steadier control than the regulated tools he had relied on for years. The software, created by patients and developers dissatisfied with the existing commercial app, is free, runs on an iPhone and, in his account, has achieved results that standard FDA-approved systems had not.
That experience captures a wider tension in digital health: the gap between the pace of software development and the slower machinery of regulation. The FDA treats much medical software as a device, a framework designed to check that products work as intended and do not contain dangerous flaws. Yet the same approach can make approval costly and cumbersome, especially when code is updated far more frequently than traditional hardware. The result, the author argues, is that the agency can become a brake on innovation rather than a safeguard for it.
The FDA has tried to adapt. In its Digital Health Software Precertification Pilot Programme, launched in 2017, the agency tested a model that would assess a developer’s quality systems and organisational practices rather than re-reviewing each product in full. The programme was intended to speed access to software-based medical devices, and the FDA said at the time that nine companies, including Apple, Fitbit, Johnson & Johnson, Roche, Samsung, Tidepool and Verily, would take part. When the pilot concluded in 2022, the agency said the exercise showed that a new regulatory paradigm was needed to keep up with fast-changing technology.
The article argues that Congress will have to finish the job. One possible route would be to revive and expand pre-certification, allowing the FDA to certify developers that meet defined standards and then give them more room to iterate within that framework. The author points out that the next five-year medical device user-fee cycle, to be negotiated by the FDA and industry groups and sent to Congress next January, could provide an opening for statutory changes that would modernise oversight for software and generative AI.
Without that, the piece suggests, patients will keep turning to open-source systems and accepting risks that are not fully understood. It also says manufacturers are caught in the middle, facing rules and timelines that make it hard to innovate at the speed of software. As the FDA has shown in other areas, including recent expansion of automated insulin-delivery indications, it is already working to widen access to advanced diabetes technology. But the article’s central warning is that incremental fixes will not be enough: in a world of machine learning and rapid updates, regulators need new tools, not just more staff or faster reviews.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
