Words: Isaac Quirke (he/him)
‘What the NHS truly needs is an injection of humanity.’
With each passing day, it becomes harder to diagnose the role of Artificial Intelligence in our society. With its questionable accuracy, poisonous environmental impact, voracious appetite for stolen datasets, and potential threat to a stable workforce, it’s a water-cooler topic everyone’s sick to death of hearing about. Apocalypse seems guaranteed, but could AI bring about a utopia? Recent NHS studies have begun introducing AI to support cancer detection in X-rays, with potential to expedite treatment for lung cancer patients. Seemingly slouching towards its deathbed, is AI the spark of life that will finally allow the NHS to hit its cancer diagnostics and treatment targets for the first time since 2015? Or is it just another sugar pill the government touts as a cure-all, meanwhile ignoring the true remedy to all its problems?
It’s hard to ignore stories like Diane McCallum’s, whose lung cancer was detected by AI software on a chest X-ray performed at New Stobhill Hospital. Through its rapid detection, McCallum, who presented with just a persistent cough, was informed of her possible cancer within hours of the investigation and had a follow-up CT scan the morning after. 7 weeks after its initial detection, Diane underwent surgery to remove the infected part of her lung, squarely within the 62 day referral-to-treatment cancer guidelines.
The AI that detected Diane’s cancer, Qure.ai, is part of the RADICAL (Radiograph Accelerated Detection and Identification of Cancer of the Lung) trial, which the NHSGGC commenced in 2024. Qure.ai almost instantaneously detects abnormalities in X-ray scans, such as lung masses and nodules. These scans are prioritised by reporting radiologists, with the intention of speeding up cancer diagnosis, consequently leading to quicker treatment turnaround. Qure.ai is also being used for all outpatient X-rays, even where there are no suspicions of cancer, improving treatment outcomes for non-symptomatic and early cancer patients.
However, frontliners aren’t convinced that stories like Diane’s can be replicated on a larger scale. In a paper published in medical journal The Lancet, oncologists warned that the effectiveness of “magic bullet” solutions are oversold. While innovations like Qure.ai could potentially support overly-strained radiology departments, they are usually costly diversions from often ignored aspects of the cancer pathway. Co-author Professor Pat Price stated in the paper, ‘If you boost diagnosis but don’t treat people on time, the newly diagnosed patients just end up stuck on long waiting lists’. By pumping money into novel, unproven technology, the government risks diverting funds from tried-and-true improvement measures, such as increased staffing, and expanded diagnostic and treatment capacity. Early results from trials similar to RADICAL also show only a slight boost in cancer detection when compared to diagnosis solely by radiologists. It’s a tough sell when you read between the lines, especially considering that the government’s 2022 10-year cancer plan, a strategy which included cancer as one of 6 major conditions to be addressed, was abandoned after less than a year.
Data safety is another source of anxiety. The NHS is a constant target for ransomware attacks, where confidential data is stolen in exchange for large sums of money. With Keir Starmer including health information in plans to build a National Data Library for training AI models and make the UK the ‘leader’ in AI, how does this put our confidential data further at risk? Since use of this dataset by private companies and for commercial purposes has not been ruled out, where can we draw the line on how our data is utilised? Only 6% of the UK’s population opt out of the use of data for research and planning purposes. Does this include this AI dataset, and what if people want to contribute to research that doesn’t include AI? With AI yet to prove its effectiveness, why is the government so willing to put our data in the hands of less-robust organisations, carving new routes for digital gangs to procure it? If it emerges that AI is a fruitless government showboat, all that could remain is a skeletal NHS, and a dark web scattered with our sensitive information.
What the NHS truly needs is an injection of humanity. If AI can provide that, if it can lighten the load for overworked healthcare staff and slash wait times for worried cancer patients, then I’m happy for the NHS to be the place where it finally feels worthwhile. However, unsurprisingly, I’m more inclined to listen to those directly tackling the issues that the government is attempting to bandage up with AI. The results seem too inconclusive and too risky to make a case for its inclusion. Government spending needs to be distributed evenly to what really makes a difference in improving health outcomes: more healthcare workers and more capacity to speak to patients, take blood, interpret images, and provide the lifesaving treatment the NHS is there to supply.
