The Pathway to Regulatory Approval of LLM-based Apps: It’s Likely a Trek, Complex & Costly

Recently, TrialSite reviewed an interesting white paper authored by Dr Hugh Harvey and Mike Pogose, both employed at Hardian Health, a clinical digital consultancy based in Haywards, Heath, West Sussex, United Kingdom. Founded in 2018, this consultancy advises researchers, businesses, and investors on how to develop digital healthcare solutions. Their recent paper investigates the prospects of what could be classified essentially as validating ChatGPT and other large language models (LLMs) in the context of Food and Drug Administration (FDA), UK Medicines & Healthcare products Regulatory Agency (MHRA) as well as the European Medicines Agency (EMA) and other life science focused regulatory bodies. With seemingly wide-open clinical applicability, TrialSite now reports on ongoing studies demonstrating the prospects of using LLMs such as ChatGPT in both the clinic and in research operations. But similar to the concept of the Gartner Hype Cycle, the UK-based authors issue a cautionary outlook. An apex of over enthusiastic hyperbole is followed by a trough of disillusionment before the inevitable, hardened embrace, compliant and at scale. If too much hyperbole ensues, critics will have the day, then slowing down the true potential of this revolutionary technology.

What’s the pathway toward true medical and research/GxP—hardened use cases for this ground-breaking technology? Well, that’s the focus of this important white paper.