Inside the NHS plan to revolutionise breast cancer diagnosis with help of AI
The NHS is launching the world's biggest trial of AI to detect breast cancer, screening two-thirds of 700,000 women.
Five different types of AI will read about 462,000 mammograms at 30 breast cancer screenings later. The other 238,000 will be read in the usual way by two radiologists.
It's part of Health Secretary Wes Streeting's national cancer plan to improve survival and treatment for the disease.
If successful, hospitals may no longer need to use a 'second reader' system, where two radiologists study every mammogram for signs of breast cancer to make sure nothing is missed. Instead, AI can be the 'second reader', leaving doctors free to do other work and reducing wait times for diagnostic test outcomes.
Research has suggested that AI is a promising oncology tool. A previous study of 116 495 women undergoing at least three consecutive rounds of mammography screening every two years found that AI could be used to identify women at high risk.
What are the risks of using AI in cancer diagnoses?
A study published in Cancer discussed the limitations of using AI in oncology, or cancer treatment.
AI algorithms are only as good as the data and assumptions they are given, so biased representations of patient groups and medical scenarios can lead to AI models too closely resembling specific data sets and inaccurate generalisations. In general, not using real-world distributions of data can decrease the reliability of AI output.
Human biases can also affect how AI algorithms are used. Factors like physicians' knowledge, technological illiteracy and age impact whether AI can be used in practice.
AI also cannot perfectly replicate human decision-making, and it's unable to take into account factors like patient composure and mental condition that aren't captured in data but are extremely important for assessing patient risk.
Professor Katharine Halliday, the president of the Royal College of Radiologists, also voiced concerns, noting that AI carries 'major risks' because it would need to be designed so that it could produce reliable results for different groups of women regardless of ethnic background. She also noted that the NHS consultant shortage is expected to reach 40% by 2028, before the trial's results will be ready.
The NHS screening programme invites women between 50-70 for screening every three years via their GP surgery, meaning about 2.1 million women have the checks each year.
'This landmark trial could lead to a significant step forward in the early detection of breast cancer, offering women faster, more accurate diagnoses when it matters most', said Professor Lucy Chappell, chief executive of the National Institute for Health and Care Research, which is funding the £11m study.
Samantha Harrison, the head of strategic evidence at Cancer Research UK, added: 'More cancer cases are diagnosed every year, and innovations like this could help reduce pressure on NHS staff and cut waiting lists. We need more research to understand how AI could help with cancer screening, so it’s fantastic to see the UK government funding this trial.'
Cut through the noise and get practical, expert advice, home workouts, easy nutrition and more direct to your inbox. Sign up to the WOMEN'S HEALTH NEWSLETTER
You Might Also Like
