UCLH invests in AI to ease burden on doctors and nurses

Machine learning could help to diagnose diseases and redistribute healthcare professionals more efficiently

University College London Hospitals (UCLH) has unveiled plans to use artificial intelligence to ease the burden on healthcare professionals by taking on tasks that could include diagnosing cancer and lowering A&E wait times.

Bryan Williams, director of research at UCLH NHS Foundation Trust, said, "It's going to be a game-changer. You can go on your phone and book an airline ticket, decide what movies you're going to watch or order a pizza … it's all about AI. On the NHS, we're nowhere near sophisticated enough. We're still sending letters out, which is extraordinary."

The three-year partnership between UCLH and the Alan Turing Institute aims to change that with machine learning. Williams thinks that it could have ‘a major impact' on patient outcomes.

Machine learning could help the NHS to operate in new ways: for example, by diagnosing diseases or directing resources. Professionals could be deployed to wards on an as-needed basis, as the algorithms learn where doctors and nurses are needed the most (although that is sure to attract criticisms about lack of familiarity with patients, and possibly privacy concerns).

The first project will focus on improving the hospital's accident and emergency department, which like many around the country is failing to meet the government's four hour maximum wait-time target. ML algorithms could, for example, assess which patients were suffering from serious problems, and fast-track them to be seen more urgently.

Another project - already underway - is being used to identify patients who are likely not to attend appointments. Consultant neurologist Parashkev Nachev has managed to predict, with an 85 per cent success rate, whether a patient will turn up for outpatient clinics and MRI scans. He is using factors such as age, address and weather conditions in his work. The next phase will see the hospital using reminder messages, and allocating appointments to maximise the chance of attendance.

The intent behind all of this automation is that healthcare professionals will have more time to spend with their patients, Chris Holmes, director for health at the Alan Turing Institute, said, "We want to take out the more mundane stuff which is purely information-driven and allow time for things the human expert is best at."

Holmes also remarked on the importance of guarding against "learned helplessness" - when people become overly reliant on technology that they cannot solve problems by themselves.

UCLH is working to allay privacy concerns before they are raised - like those that plagued the Google DeepMind/Royal Free Hospital tie-up. The hospital will train its algorithms on its own servers and private companies will not be involved, Holmes said.

Marcel Levi, UCLH chief executive, added, "Machines will never replace doctors, but the use of data, expertise and technology can radically change how we manage our services - for the better."

We hope that the organisations can also avoid the situation brought up by one Computing writer, known informally around the office as ‘Doctor Terminator':

"YOU APPEAR TO HAVE A [BRUISED ELBOW] HOLD STILL WHILE I [AMPUTATE YOUR LEGS]"