The National Institute for Health and Care Research (NIHR) has identified 10 potential AI interventions that could address five key challenges for the NHS.
These healthcare challenges include detecting heart disease, diagnosing lung cancer, predicting disease progression, personalising cancer and surgical treatment, and alleviating pressures on emergency departments.
The innovations outlined include:
- Heart disease:
- A smart stethoscope that research showed could identify those with heart failure correctly nine out of 10 times. This could help GPs prioritise referrals and save the NHS time and money.
- An AI application used routine clinical data in conjunction with a blood test that measures heart damage and was able to correctly predict whether patients in emergency rooms had experienced a heart attack or not. This could speed up patient flow, boost early treatment and stop unnecessary admissions.
- Lung cancer:
- Researchers investigated whether AI could decide whether abnormal growths on a CT scan were cancerous or not. Two studies explored separate AI interventions for large nodules and small nodules, comparing efficacy against the British Thoracic Society-recommended Brock score. Both innovations could predict cancer more accurately that the Brock score.
- Disease progression:
- Researchers explored whether AI could predict the progression of wet age-related macular degeneration (wet AMD) – this is the biggest cause of sight loss in the UK; one in four people with wet AMD in one eye are expected to get in their second eye as well. AI was better than five out of six experts at determining whether the disease would develop.
- An AI could see when those with ulcerative colitis (bowel ulcers and inflammation) would get flare-ups. This could streamline and standardise assessment in this area.
- Personalised treatment:
- AI could help clinicians determine which drug combinations are most likely to benefit lung cancer patients in just 12-48 hours.
- An AI tool was found to be able to predict the risk of death for patients with Covid-19 a month after their surgery.
- A&E pressures:
- Researchers took 100,000 ambulance and emergency department records from across Yorkshire and found that, using AI, they could correctly predict which people didn’t need to visit A&E, with an 80% accuracy.
- An AI tool was able to predict how many beds a busy London hospital would need in four and eight hours’ time. This was proven to be better than the usual planning method, which is based on the previous six weeks’ demand.
“Over just the last few months, our awareness of the potential of artificial intelligence to change our lives – in both good and bad ways – has risen to new heights,” said the NIHR’s scientific director for innovation, Professor Mike Lewis.
The secretary of state for health and social care, Steve Barclay, added: “It is encouraging to see the results of these studies, which show that with the help of AI, doctors could soon be able to detect heart disease more quickly, predict the progression of diseases, and personalise cancer treatments.”
The NIHR has indicated that more research is required to fully understand how these tools could work in practice, the long-terms effects of implementation, and their overall value for money.
Image credit: iStock
