In 2017, a company developed an AI algorithm capable of detecting breast cancer between 13 and 34 percent more accurately than two expert radiologists working together. It is still not routinely implemented in the UK. Not because the evidence was weak. Not because the technology failed. But because the regulatory and adoption infrastructure was never built to receive it.
This is the paradox at the heart of healthcare innovation, and it is precisely the challenge that regulatory sandboxes were designed to solve. The question we must now be honest enough to ask is: not just whether they are solving it, but how do we optimise their impact so that the innovations that enter them actually reach the patients who need them.
During WHX Dubai 2026 (formerly Arab Health), I had the privilege of convening a discussion that brought together healthcare leaders, clinicians, legal experts, and system innovators with direct experience of building and operating sandboxes across the Middle East and the UK. What emerged was not a celebration of sandboxes as a concept. It was a clear eyed assessment of where they work, where they stall, and what needs to change. I am grateful to Prof Sam Shah for contributing their perspective to this piece. Between us, we have worked on sandbox design and innovation adoption from Riyadh to London, and the pattern we keep finding is the same.
The sandbox is not the problem. The exit is.
Sandboxes, at their best, are structured programmes with four components: clear objectives, defined eligibility criteria, robust guardrails, and supervised environments in which innovators, regulators, and providers co-create together. This is not a new idea. The NHS experimented with test beds and digital development pilots’ years before the term sandbox gained currency. The frameworks were recognisable. The outcomes were not.
The failure was almost never at entry. It was at exit. The sandbox got the innovation in. Nobody built the way out.
Time and again, across geographies and across decades, the pattern repeats: a promising innovation enters a structured test environment, generates real-world evidence, and then nothing. The data sits behind closed doors. The funder moves on. The pathway to scale was never designed. At every discussion we had at WHX Dubai, the same question cut through: who had answered the “and then what?”
This is not a criticism of ambition, it is a structural design problem, and it is one that health systems can no longer afford to treat as secondary.
Professor Sam Shah | Director of Health Data Management, Neom Health
“When we talk about sandboxes in healthcare, we really mean structured programmes with four key components: clear objectives, defined eligibility criteria, robust guardrails, and a graduation pathway. The problem is that we have become good at designing the first three and consistently failing to design the fourth. I have been involved in building sandbox environments from London to Riyadh, and the pattern is the same everywhere: the sandbox fills up with promising innovation, generates evidence, and then the pathway to scale was never built. We have seen these things play together. We have not seen them scale.”
What good sandbox design looks like
The most instructive examples of sandbox environments that have successfully connected testing to adoption share a structural feature that most health systems, including the NHS, have not yet replicated at scale: they build the adoption commitment into the design of the environment itself, before the first company enters.
In the most sophisticated models, hospital groups and health networks have begun investing directly in companies operating within structured innovation environments, not as philanthropists, but as strategic investors with a pre-committed intent to adopt if the product graduates successfully. That changes the economics fundamentally. The company has a runway and a route to scale. The provider has skin in the game. The sandbox has an accountability mechanism that most pilot programmes lack entirely.
For the UK, this points toward a model in which groups of NHS trusts, integrated care systems, or networks of practices commit in advance, as a condition of a company entering a structured environment, to adoption if defined outcome thresholds are met. It would not be a guarantee of commercial success. But it would change the capital risk profile for health technology investment fundamentally, and it would create the accountability that turns a sandbox into a genuine graduation pathway rather than a well-resourced cul-de-sac.
Leadership sponsorship is not optional. A sandbox without a senior champion who can issue directions, unblock decisions, and stay the course will drift. This is one of the most consistent lessons from sandbox environments I have observed, and it applies equally in London as it does in Riyadh.
The three barriers between the sandbox and the market
There are three structural barriers that consistently determine whether a promising innovation reaches scale or stalls at pilot. Sandboxes address none of them directly. Understanding them is essential for anyone designing a sandbox environment, commissioning innovation, or investing in health technology companies.
- The first is procurement: NHS procurement frameworks were not designed for value based, outcomes-linked products, and most health technology companies arrive without understanding how to navigate that.
- The second is IP and data: NHS contract provisions make it genuinely difficult to train AI algorithms on NHS data in the way the product requires, and while the new IP policy released in late 2025 is a step forward, the gap between policy intent and contract reality remains substantial.
- The third is product regulation: we are completely outpacing where regulation currently sits for software and AI as medical devices. All three are solvable. None of them are being solved fast enough.
These three barriers; procurement architecture, IP and data provisions, and product regulation maturity, are not new discoveries. They have been named, reported, and recommended against for years. What is different about the current moment is that there is, for the first time in some years, genuine political will and institutional momentum in London to do something about them. The London sandbox being developed with the MHRA represents a real opportunity, if it is designed with the graduation pathway already built in, and if the three structural barriers are addressed as part of the design rather than deferred to later.
The NHS is perfectly wired to maintain the status quo, unless we change the wiring
There is an uncomfortable truth about NHS innovation adoption that I want to name directly, because it was said plainly in our WHX Dubai discussion and it is rarely acknowledged in official accounts of why innovation stalls.
The NHS already has proven innovations, technologies that have cleared NICE evaluation, that have been assessed and approved, that it systematically fails to adopt. The problem is not regulatory. It is organisational. It is nobody’s job to drive adoption. It is everybody’s job to keep running their healthcare services in precisely the way they are currently running them. Passing innovations from one queue to another, from the sandbox to NICE, from NICE to procurement, from procurement to implementation will not fix the adoption problem if the underlying system is designed to absorb and neutralise change rather than to scale it.
For a London sandbox to produce different results than its predecessors, the accountability for adoption has to be built in from day one, not added when the testing phase ends. This is precisely why the design of the graduation pathway matters more than the design of the sandbox itself. A sandbox without a committed adopter at the end of it is not a pathway. It is an extended pilot with better branding.
Three principles for sandboxes that actually work
Drawing from direct experience across sandbox environments in the UK and internationally, three principles stand out consistently as the difference between a sandbox that scales and one that stalls.
The London–Middle East connection
One of the most important observations from our WHX Dubai discussion, and one that is rarely made explicitly, is that the learning between health systems on sandbox design and innovation adoption is no longer running in one direction. For decades, the assumption has been that the NHS is the exporter of healthcare expertise and other systems are the importers. That model is changing.
Professor Sam Shah | Director of Healthcare Data Management, Neom Health
“One thing I have learned from working on sandbox environments in multiple jurisdictions is that the temptation to copy paste models is almost always a mistake. The problems look similar on the surface, regulatory barriers, data governance, procurement friction, but the underlying architecture is different in every place. What the most mature sandbox environments have in common is not a specific design, but a specific discipline: they treat regulation as something to be tested and evolved, not bypassed. The most useful thing the UK and the Middle East can do for each other right now is not to exchange blueprints, but to exchange the honest account of what has not worked and why.”
The momentum around the London MHRA sandbox and the innovation infrastructure work being led through Imperial College Health Partners represents a genuine opportunity to take the lessons from more mature sandbox environments, including the design discipline around graduation pathways and adoption commitment, and apply them from the outset, rather than discovering their absence at the end of a costly pilot phase.
The moment we are in
Healthcare systems globally are under pressure. Outcomes are stagnating. Workforce models are under strain. The technologies to change this exist. The regulatory appetite to test them is growing. What remains missing is the architecture to take what works in a sandbox and reliably deliver it at scale.
The innovation already exists. The question, the one that we opened this discussion with at WHX Dubai, and that Sam and others in that room have been working on for years, is whether the system is ready to receive it. Not just to test it. To receive it, adopt it, and make it reliably available to the patients who need it.
At CF, we work with health system leaders, innovators, and investors across the entire healthcare sector to build the end-to-end infrastructure to make that happen, from strategy through implementation, accelerated by data, digital, and AI. If you are working on this challenge in London, in Riyadh, or anywhere in between, we would be glad to continue the conversation.
About CF
CF is a leading consultancy dedicated to making an enduring impact on health and healthcare. We work with leaders and frontline teams to improve health, transform healthcare, embed life science innovation and boost growth through investment. With unmatched access to UK healthcare data and award-winning data science expertise, our team are a driving force for delivering positive and meaningful change.
About the authors
Dr Zahra Safarfashandi, Partner & Middle East Lead, CF
Zahra is a clinician and healthcare management consultant specialising in clinical and digital transformation at scale. Her NHS work includes leading elective recovery and digital technology rollout programmes with NHS England. She leads CF’s commercial presence across the Middle East and convened the WHX Dubai 2026 panel from which this piece draws.
Professor Sam Shah, Director of Health Data Management, Neom Health
Sam is a former Director of Digital Development at NHS England and has led sandbox design and implementation across the UK and the Middle East. He was instrumental in building the first Dubai health sandbox and has worked on sandbox design in Saudi Arabia as part of Vision 2030’s health transformation agenda.
