News, Secondary Care

How university-led research is driving the use of AI for diagnosis and screening in conditions including breast cancer and dementia

The University of Aberdeen is working with NHS Grampian on AI breast screening technology; University of Sheffield is working on an AI tool to speed up dementia diagnosis; and Newcastle University’s COLO-DETECT study aims to use AI to detect early signs of colon cancer.

The University of Aberdeen recently announced that it was working with NHS Grampian and Kheiron Medical Technologies on an AI breast screening technology which is said to detect abnormalities which traditional screening methods would have missed. To test the solution, 220,000 mammograms from more than 55,000 patients were analysed by the software, named “Mia”, which was found to be successful in identifying interval cancers which would have gone undetected until the women developed symptoms, the researchers said.

According to the research team’s findings, Mia would have suggested recalling 34.1% of the women who went on to develop cancer in between screenings. In the process of testing Mia, the research team also determined the importance of fine-tuning AI tools to suit local populations and conditions, after the solution was initially deemed too sensitive in recalling patients unnecessarily. This led Professor Anderson, Chair in Health Data Science at the University of Aberdeen, to note that “ongoing quality assurance monitoring is essential”.

At the University of Sheffield, researchers have been working on an AI tool which could speed up dementia diagnosis, spotting the early signs of the disease which might warrant referral to a specialist. CognoSpeak uses AI to analyse language and speech patterns in conversations between patients and a virtual agent, and has been found to be “as accurate at predicting Alzheimer’s as the current pen-and-paper-based assessments”.

Accessed through a web browser, the assessment can be carried out from a patient’s home. Whilst early trials have shown that the technology boasts an accuracy of 90% “for distinguishing people with Alzheimer’s from people that are cognitively healthy”, the tool will now be trialled more extensively as researchers look to recruit a further 700 patients from memory clinics across the UK, thanks to an injection of £1.4million in funding from the NIHR.

Elsewhere, Newcastle University has announced the closure of its COLO-DETECT study, “the largest trial of its kind in the world”, which recruited more than 2,000 patients and was found to be capable of analysing data “much quicker than people can”, detecting extra polyps and abnormalities. The University of Warwick has also received £2.3million in funding to develop a screening tool for colon cancer, which will involve data from 10 NHS hospitals across the UK and more than 10,000 biopsy samples, potentially revealing the possibility for AI to transform the diagnosis process for the disease.

Guiding the future development of AI for diagnosis and screening

Last month we wrote about the CDEI’s timely publication of its AI assurance techniques, that stresses the importance of ensuring trustworthiness in AI solutions, providing advice on evaluation and quality checking, as well as good practice guidance.

NICE’s “one-stop-shop for for AI and digital regulations for health and social care” has also been developed in order to provide developers with insight into regulations on AI solutions within the health and care sector, focusing on the safe implementation and use of AI-driven interventions.

In June, the DHSC announced that £21million would be invested in AI technologies. This funding will allow NHS trusts to bid for funding to roll-out available AI solutions quickly, with aim to mitigate some of the pressures on diagnostic services.

NHS England has also been working to help ensure that NHS staff are prepared for the introduction of AI-driven technologies, and to understand what challenges lie ahead for their implementation. In its report titled “Developing healthcare workers’ confidence in artificial intelligence”, it explores how confidence in AI among healthcare workers can be developed through education and training.