Artificial intelligence (AI) will diagnose broken bones under new NHS plans.
Patients arriving at A&E with suspected fractures will have their X-rays reviewed by a machine, under proposals in draft guidance published on Tuesday.
The NHS spending watchdog has provisionally approved four different AI tools to be used in emergency departments across England.
A&E staff will be able to use AI to speed up diagnoses and reduce the number of missed fractures, which are the most common diagnostic errors made in emergency departments.
About 3 to 10 per cent of fractures were missed in urgent care, the guidance said, because of the pressure on doctors who do not specialise in radiography.
The National Institute for Health and Care Excellence (Nice) said there was enough evidence to show AI was better - and never worse - than doctors in A&E at diagnosing broken or fractured bones.
It has recommended that four products - TechCare Alert, BoneView, RBfracture and Rayvolve - are used in urgent care settings in England, while also generating further evidence to assess the technology's benefit in a real-world setting.
AI could help fill staff gaps
Mark Chapman, the director of health tech at Nice, said the use of AI could help fill the void caused by the high number of vacancies in radiography.
"Every day across the NHS thousands of images are interpreted by expert radiologists and radiographers, but there is a high vacancy rate within these departments across the country and more support is needed to manage their workload," he said.
"These AI technologies are safe to use and could spot fractures which humans might miss given the pressure and demands these professional groups work under."
He added: "Using AI technology to help highly skilled professionals in urgent care centres to identify which of their patients has a fracture could potentially speed up diagnosis and reduce follow-up appointments needed because of a fracture missed during an initial assessment."
In studies, the AI tools improved diagnosis rates consistently when used in emergency settings.
In one example, the BoneView tool increased the accuracy from 61 per cent to 74 per cent with the help of AI, while for RBfracture it went from 74 per cent to 83 per cent, and with Rayvolve from 79 per cent to 94 per cent.
A separate study with TechCare Alert involved radiologists, so saw a smaller increase, from 92 per cent to 95 per cent.
According to Nice, the "true cost of implementing and using AI technologies for fracture detection is uncertain" as data was from retrospective studies.
During modelling, the cost per scan was estimated at £1, with centres using the technology advised to ensure the cost per scan is similar to the estimate in the draft guidance.
The draft guidance said "further evidence" was needed on the cost of implementation, adding: "These costs are important for understanding the financial investment that is needed and also the feasibility and sustainability of integrating AI technologies into routine healthcare."
A consultation on the draft recommendations will run until Nov 5.