Getty Images
“Opportunistic” AI screening can detect diseases doctors weren’t looking for
This is the sixth article in a six-part series examining how AI is changing medical research and treatment.
When Will Studholme, 58, found himself in an accident and emergency at an NHS hospital in Oxford in 2023 with gastrointestinal symptoms, he didn’t expect a diagnosis of osteoporosis.
The disease, strongly associated with age, weakens and weakens bones, increasing the risk of fracture.
Mr Studholme was found to have been suffering from a serious case of food poisoning, but early in the investigation into his illness he underwent an abdominal scan.
This analysis was then carried out using artificial intelligence (AI) technology which identified a collapsed vertebra in Mr Studholme’s spine, a common early indicator of osteoporosis.
More tests followed, and Mr Studholme not only received a diagnosis, but also a simple treatment: an annual infusion of an osteoporosis drug that should improve his bone density.
“I feel very lucky,” Mr Studholme says. “I don’t think this would have been possible without AI technology.”
Will Studholme
While Will Studholme was being treated for food poisoning, AI found signs of osteoporosis
It is not uncommon for a radiologist to notice something incidental in a patient’s imaging – an undetected tumor, a problem with a particular tissue or organ – apart from what he had initially searched.
But applying AI behind the scenes to systematically comb through scans and automatically identify early signs of common preventable chronic diseases that could be brewing – regardless of why the scan was initially ordered – is new .
Clinical use of AI for opportunistic screening, or opportunistic imaging, as it’s called, “is just beginning,” notes Perry Pickhardt, professor of radiology and medical physics at the University of Wisconsin-Madison , who is one of those who develop the algorithms.
It is considered opportunistic because it takes advantage of imaging already performed for another clinical purpose – whether it is suspected cancer, lung infection, appendicitis or abdominal pain.
It has the potential to detect previously undiagnosed diseases at an early stage, before symptoms appear, when they are easier to treat or prevent progression. “We can avoid a lot of the lack of prevention that we missed before,” says Professor Pickhardt.
Regular medical or blood tests often fail to detect these diseases, he adds.
Getty Images
CT scans contain a lot of information that is not examined
There’s a lot of data in CT scans related to tissues and organs in the body that we don’t really use, notes Miriam Bredella, a radiologist at NYU Langone who also develops algorithms in this area.
And while the analysis could theoretically be done without AI by radiologists taking measurements, it would take a lot of time.
The technology also has benefits in terms of reducing bias, she notes.
A disease like osteoporosis, for example, is considered to primarily affect thin, older white women. Doctors therefore do not always think to look outside this population.
Opportunistic imagery, on the other hand, does not discriminate in this way.
The case of Mr Studholme is a good example. Being relatively young for osteoporosis, being male and with no history of fractures, it is unlikely that he would have been diagnosed without the AI.
In addition to osteoporosis, AI is trained to help opportunistically identify heart disease, fatty liver, age-related muscle loss and diabetes.
Although the primary focus is on CT scans, for example of the abdomen or chest, work is underway to opportunistically glean information from other types of imaging, including chest x-rays and mammograms.
The algorithms are trained on several thousand previous labeled scans, and it is important that the training data includes scans from a wide range of ethnic groups if the technology is to be deployed across a wide range of people, experts emphasize .
And there is supposed to be some level of human review: if the AI finds anything suspicious, it will be sent to radiologists for confirmation before then being reported to doctors.
The AI technology used to examine Mr Studholme’s scanner is owned by Israeli company Nanox.AI, which is one of the few companies working on AI for opportunistic screening – with many more focusing on use of AI to help with accurate and rapid diagnoses of the specific conditions for which the analyzes are actually carried out.
Nanox.AI offers three opportunistic screening products aimed at helping identify osteoporosis, heart disease, and fatty liver disease, respectively, from routine CT scans.
Oxford NHS Hospitals began testing Nanox.AI’s osteoporosis-focused product in 2018 before officially rolling it out in 2020.
According to Kassim Javaid, professor of osteoporosis and rare bone diseases at the University of Oxford who led the introduction of the algorithm.
Further trials of the algorithm are also underway at hospitals in Cambridge, Cardiff, Nottingham and Southampton. “We want to gather the evidence needed to use it across the NHS,” says Professor Javaid.
Yet while the technology can benefit individuals, it has wider ramifications that need to be considered, says Sébastien Ourselin, professor of healthcare engineering at Kings College London, who directs the AI Center for Value Based Healthcare.
A big issue that needs to be balanced, he notes, is the number of additional patients that using technology can create. “This increases the demand on the health system, without reducing it,” he says.
First, individuals flagged by opportunistic screening as potentially having disease will likely need additional confirmatory testing, which requires resources. And if the AI is inaccurate or too sensitive, it could lead to a lot of unnecessary testing.
Then, services need to be put in place for the additional people who eventually receive a diagnosis.
The extra load is a technology challenge, Professor Javaid admits – but there are solutions.
Patients with confirmed fractures in Oxford are referred for follow-up to a fracture prevention service largely staffed by nurses so as not to overburden doctors. “AI forces you to change lanes,” he says.
And in the long term, Professor Javaid believes, being able to identify more people with osteoporosis at an early stage and receive the preventative treatment they need will save the NHS money. “Fractures are one of the main reasons people end up in the hospital,” he says.
Mr Studholme witnessed the ravages of osteoporosis firsthand: his mother fractured both hips. In the past, it was just seen as an old person’s condition with nothing they could do, he said. “I feel privileged to be able to do something before my bones turn to chalk,” he says.
