TL;DR
The medicines and devices regulator has tied up with Manchester University NHS Foundation Trust to trial health technology on real patients inside ordinary care. The first venture under that partnership is a Manchester Sandbox, and AI-enabled devices are expected among its earliest occupants. Findings will be circulated across the health service rather than kept local.
What the sandbox does
The stated target is a specific and familiar failure. Plenty of tools look convincing during development and then behave differently in a working hospital, and nothing in the usual approval route reliably catches the difference. Putting products in front of actual patients and clinicians, under oversight, is meant to generate evidence on safety, effectiveness and impact that a controlled trial does not produce.
Early candidates are expected to include devices that flag which patients with a long-term condition are most likely to deteriorate, so treatment can start sooner. The launch at MFT’s Oxford Road Campus showcased wearables and ambient voice technology — the documentation tools already spreading through NHS consulting rooms.
There is a second audience here beyond patients. Developers routinely complain they cannot tell what evidence the NHS actually wants before committing to generate it. The programme is explicitly meant to make that clearer.
Lawrence Tallon, who runs the MHRA, said the challenge is “understanding how they work in practice and where they can make the biggest difference”. MFT chief executive Mark Cubbon put the constraint more bluntly: patients “should never have to choose between access and safety”.
James Frith, the health innovation minister, said products would be trialled under “strict oversight and guardrails”.
Looking forward
A roundtable with providers, developers, researchers and patient representatives is planned before the year is out, and the regulator will ask device makers to register interest in the next phase.
Read this alongside the rest of the morning. The same regulator’s independent commission proposed staged authorisation and lifelong monitoring for healthcare AI, and eight trusts began a year-long surgical AI trial.
Three separate announcements, one shared premise: the evidence that matters is generated after deployment, not before it. Vendors selling into the NHS should expect that assumption to start appearing in procurement questions.