When it comes to conversations around artificial intelligence (AI), the most popular connotations of our current time are split between uncannily-rendered images of people with six fingers generated at the cost of devastating environmental consequences, or of enabling a robot takeover that had been impending for the last century. However, the outcome of regular use of AI isn’t always so grim, or as far off into the future as we may think.
The Prime Minister’s speech ‘AI Opportunities Action Plan’ in January announced the Government’s intentions to utilise AI to revolutionise public services. He highlighted its importance in healthcare in particular by using the case of 54-year-old Deb Kelly from Stoke as an example – a woman who suffered a stroke and would’ve died had it not been for the usage of AI.
‘She was found by her son, [and] rushed to hospital where the doctors used artificial intelligence to help pinpoint the exact location of the blood clot,’ said Prime Minister Keir Starmer. ‘She said if that blood clot hadn’t been removed, she wouldn’t be here right now – I wouldn’t have spoken to her this morning. That’s the power of AI in action.’
The integration of AI in medicine is already in motion, especially so in the last five years. The Government and the NHS have been actively seeking out and funding various projects expanding the usage of AI for the purpose of early diagnoses; health predictions; and to better streamline the health service, with initiatives such as the ‘AI Diagnostic Fund’.
- First artificial intelligence trial to tackle breast cancer launched
- AI receptionists to be rolled out by the NHS to reduce missed appointments
- Hospitals coming home: the rise of virtual wards in the NHS
How does AI work?
AI is technology that simulates human intelligence by making decisions, responses and seemingly performing cognitive functions that a human brain could typically make. It does this by the use of computer-generated algorithms that rapidly analyse datasets and detect patterns to make decisions, and ultimately generate original content.
AI was first used for medical purposes in the 1970s with INTERNIST-1 – a medical consulting tool that analysed patient symptoms and generated a diagnosis. Medical AI saw another major expansion in the 2000s when computer giant IBM established the question-answer system ‘Watson’. Aside from winning first place on quiz game show Jeopardy! in 2011, Watson was also later used to identify altered proteins associated with the development of the neurological disorder amyotrophic lateral sclerosis, also known as Lou Gehrig’s disease.
Over the last five years alone, developments in the usage of medical AI have been progressing at rapid rates across multiple avenues, with a focus on diagnostics.
‘AI has a lot of potential in diagnostics,’ says Dr Arun Sau, cardiologist, clinical lecturer at Imperial University and AI researcher. ‘We know from modalities, like imaging X-rays. that there’s a vast amount of information that humans either cannot appreciate or miss or just not even something that we would ever consider.’
NHS AI Diagnostic Fund
With time-sensitive disease such as cancer, an early diagnosis could be lifesaving. In 2023, the previous Government announced that it would grant £21 million across 64 NHS trusts in what it called the ‘AI Diagnostic Fund’, to fund the development of AI treatment and diagnostic technology. This was awarded to accelerate the development of AI tools to expedite and support the diagnostic process for conditions such as cancer or heart defects.
‘Artificial intelligence is already helping to save lives from faster diagnosis of a stroke allowing faster emergency treatment, to providing patients with their personalised risk of a heart attack,’ says Dr Vin Diwakar, National Director of Transformation at NHS England. ‘This investment will allow 64 NHS trusts from across the country to harness the power of AI to tools to speed up the diagnosis and treatment of lung cancer.’
AI in cancer
One area of medical technology that had seen rapid developments with AI are cancer diagnostic tools. The NHS has already outlined in the Long-Term Plan that it aims to diagnose 75% of cancers at stage 1 or 2 by 2028. NICE already recommends AI to be used when diagnosing lung cancer, to aid healthcare professionals when scanning chest X-rays. Research for other cancers had already been underway from various different avenues as in 2023, a Swedish study screened over 80,000 mammograms – half with a radiologist, and the other half with the addition of AI. The study found that mammograms in the group screened by both a radiologist and AI detected more breast cancers than in the group screened by a radiologist alone.
With the AI Diagnostic Fund, the NHS honed its focus onto imaging technology, and highlighted the need for chest X-ray analysis for diagnosing lung cancer. An early value assessment of AI-derived software by the National Institute for Health and Care Excellence (NICE) supported the development of such tools, suggesting that they could shorten the time it takes from the receipt of a chest X-ray to a CT scan referral, and improving overall diagnostic accuracy. The first trust to utilise this technology was the South Tyneside and Sunderland Trust in the North East and North Cumbria region, who are now set to use AI alongside their X-ray kits.
Ken Bremner, Chair of the North East and North Cumbria Provider Collaborative and CEO of South Tyneside and Sunderland NHS Foundation Trust says that ‘The arrival of this technology for the region’s NHS is a very welcome addition to our fight against lung cancer and the terrible impact it has on our region’s communities.’
With lung cancer being the most common cancer-related death in the UK, and over 600,000 chest X-rays being performed in England per month, deploying diagnostic AI tools hold considerable potential for helping clinicians improve patient outcomes.
The future of AI in medicine
While helpful in medical and diagnostic processes, the future of AI sees it purely as a facilitating tool rather than a replacement for primary care staff. Professor Kamila Hawthorne, Chair of the Royal College of General Practitioners (RCGP) stated that ‘technology will always need to work alongside and complement the work of doctors and other healthcare professionals,’ and that primary care workers ‘have distinct expertise and experience in providing whole person medical care. It’s difficult to see how this could be completely replaced by AI.’
However, this doesn’t make the convenience of AI redundant as it has demonstrated the potential to be applied to broader medicine.
A recent study by Imperial College London and Imperial College Healthcare NHS Trust explored how AI could possibly predict early deaths and other health risks such as heart failure and type-2 diabetes. The study used AI to analyse over one million electrocardiogram (ECG) results to predict the possibility of new disease and other outcomes for patients.
‘An AI model can take a million ECGs and the whole trajectory of a patient, learn from that entire data set, and retain all of that information with no limit,’ says Dr Sau, who led the project. ‘There’s a real potential through automation and early detection through AI in all of these domains.’
The project trained an AI model to analyse ECG data – which are routinely taken and exist in high volumes – and observe which patient’s conditions deteriorated, contracted disease, died or survived. This analysis therefore would let the AI read new patient ECGs and predict the most likely outcome, so health professionals can implement the necessary interventions as early as possible.
‘We’re taking simple heart readings, which are one of the most common tests in the world and training AI to predict the risk of someone dying early, and also the risk of them developing serious heart conditions,’ Dr Sau explains further. ‘If the AI predicts that you’re at very high risk of heart failure, you could be monitored very closely with, for example, serial heart scans or even have preventative therapies. But anything that we want to adopt needs to be studied in rigorous randomised control trials and not just adopted without proper evidence base.’
The technology is aimed to be integrated into the NHS after five years. With the conclusion of the initial research, the project is set to move forward into randomised clinical trials with a target sample of over 1,000 patients for the initial instance before potential integration into the healthcare system.
However, aside from medical technologies, the usage of AI has expanded in other areas of the healthcare system. With promises to cut maximum waiting times from ‘18 months to 18 weeks,’ by Health and Social Care Secretary Wes Streeting, the NHS is set to roll out AI software to tackle missed appointments and reduce waiting times by predicting which appointments are likely to be missed and for what reason through an algorithm. This will ultimately leave more staff free to continue with other duties.
Primary care workers were found to spend a significant amount of time with ‘high levels of unnecessary bureaucracy and administrative processes,’ according to Professor Hawthorne. She claimed that AI ‘could help to solve a longstanding problem’ when utilised to alleviate bureaucratic pressure, as primary care staff will therefore be left with more time to ‘deliver more appointments and spend more time with [our] patients.’
NHS England found that over eight million appointments had been missed in the previous year, resulting in £1.2 billion worth of costs. The technology was developed by AI provider Deep Medical and Bart’s Health NHS Trust to minimise missed appointments and help patient’s access better care, by analysing reasons why people often miss their appointments, and offering backup slots.
‘The use of AI to help reduce the number of missed appointments is another example of how new technologies are helping to improve care for patients,’ said Dr. Diwakar. ‘Not only can these technologies help to free up doctors’ time to treat more
patients and reduce waiting times for planned care, it means a significant amount of money can be invested in frontline care rather than lost to missed appointments.’ The software was piloted in Mid and South Essex NHS Foundation Trust, and is prepared to go through further trials before mainstream integration into the system.
However, as with most digital developments concerns of safety and data protection arise – such as the NHS ransomware attacks in 2022 and 2024. The RCGP commented that while the use of AI is welcomed, ‘its use will need to be closely regulated, guaranteeing the security of patient data.’ Professor Hawthorne said that GPs are typically ‘controllers’ of patient data and are responsible for its safety, reassuring that ‘we take this responsibility extremely seriously.’
With the spike in initiatives and funding for AI research, such as the NHS AI Fund and the oncoming trial stages of various software, AI is set to be integrated into nearly all core parts of the healthcare system. And with the Plan for Change for the NHS, it’s possible that AI could be utilised even further to bring the service closer to its goals of better service, more accurate diagnoses and shorter
waiting times.