A UK healthcare AI company, Behold.ai, says winning the ‘Nobel Prize’ in biopharmaceutical research for its X-ray analysis algorithm is important proof of the scientific validity of using Machine Learning in diagnosis.
More importantly, says the company’s CEO and co-founder, the lives of thousands of Brits facing the peril of undiagnosed lung cancer could now be saved by the approach, as an estimated 90% of chest X-rays can lead to an inaccurate result.
The award in question is the Prix Galien—an international awards program dedicated to medical progress through innovation.
The scheme is administered by the Galien Foundation, which is all about the open exchange of ideas to drive scientific innovation in medicine. In May, the body awarded the London-based company in question, Behold.ai, the 2022 prize for “Best UK Digital Health Solution.”
The award was for the company’s ‘red dot’ clinical analysis software, which aids human experts to quickly filter out normal (so-called ‘High Confidence Normal (HCN)’) lung X-rays to save radiologist time.
‘Early diagnose 55,000 additional lung cancer cases by 2028’
That matters, as 40% of images out of 100 typically indicate areas of concern, and by identifying only those with a suspected nodule, or mass, to a radiologist, care can be rapidly sped up: triaging X-rays like this could also, claims the start-up, potentially save 255,000 hours of radiology consultant time per year, as well as helping to promote more standardised diagnosis across the UK public health service.
“Lung cancer is the single largest cause of cancer-related deaths–not just in the UK, but also in Holland and in Germany, and every year it kills 2 million people globally,” Simon Rasalingham, the firm’s leader, told RockingRobots.com.
“The NHS is looking to diagnose 55,000 additional lung cancer cases by 2028. Almost half of those cases could be picked up by this one algorithm,” he added.
Rasalingham—a serial medical device entrepreneur who previously set up then sold another healthcare IT company, Medica–added that his results so far indicate that using red dot to immediately call out HCN scans could reduce backlogs in NHS lung cancer diagnosis by at least 15%, or 1.2 million X-rays, per year.
Current NHS elective care backlog, as in all treatments, not just lung cancer, is predicted to rise from today’s 6.1 million to 12 million by March 2025. At the award evening, Rasalingham stated that winning the prize would give his company “the forward momentum and recognition to implement our technology to tackle [the] NHS backlog”.
‘Harnessing AI to tackling the issues facing radiology internationally’
Rasalingham says the software is currently being trialled at NHS Trusts including the Dartford and Gravesham NHS Trust, Somerset NHS Foundation Trust, and others as part of an independent validation, which results due to be published soon in a peer-reviewed academic journal–but that this is a very important validation, too:
“There’s been a lot of question marks around the science of AI, but we have peer reviewed and published evidence which we supplied to win this award that was judged by senior independent clinicians and experts shows that this is the best scientific tool of its type in the UK in this field.”
Next steps for Behold.ai is to build on the success of the first wave of installations and show the potential for harnessing AI to lead the way in tackling the issues facing radiology internationally, starting in the UK and USA–but with the EU also very much on its radar screen.