Intended for healthcare professionals

×

Error message

Access denied. You must log in to view this page.
News Fact Check

Is the NHS rolling out AI technology to prevent falls?

BMJ 2025; 389 doi: https://doi.org/10.1136/bmj.r716 (Published 08 April 2025) Cite this as: BMJ 2025;389:r716
  1. Ella Hubbard,
  2. Elisabeth Mahase
  1. The BMJ

Last month NHS England issued a press release announcing a “nationwide rollout” of an “artificial intelligence tool that predicts falls and viruses.”1 It said that the tool, developed by the care provider Cera, was being “rolled out across the NHS” and “can predict a patient’s risk of falling with 97% accuracy, preventing as many as 2000 falls and hospital admissions each day.”

The press release quoted senior government officials backing the tool as a “perfect example of how the NHS can use the latest tech to keep more patients safe at home and out of hospital.” The announcement was covered across several media outlets, including ITV News, the Independent, and the London Standard.23 The NHS Confederation issued a response, welcoming the rollout of the technology, but warning of the need for robust evaluation of AI in the health service.4

How will the NHS roll out the AI?

The BMJ approached Cera and NHS England to ask what a national rollout would look like or who it involved. This led to clarification that the tool will be used only by Cera staff. The “nationwide rollout” seems to refer to Cera’s intention to expand use of the AI tool across its services and to extend its reach by winning more social care contracts. But, despite the confusion caused by the press release and the potentially misleading media reports, NHS England declined to comment further.

Can the tool predict a patient’s risk of falling with 97% accuracy?

When asked for the evidence to support the claims for the AI tool, NHS England directed The BMJ to Cera.

NHS England’s press release said that the tool could predict a patient’s risk of falling with 97% accuracy. Cera defended this claim, saying that it derived from “technical evaluations” that compared the tool’s categorisation of patients into low, medium, and …

View Full Text

Log in

Log in through your institution

Subscribe

* For online subscription