An NHS Trust is bringing in artificial intelligence (AI) software to predict potentially missed appointments and offer back-up bookings. But whether the software is accurate enough to help the health service cut waiting times remains to be proven, an expert told Tech Monitor.
The software, which has been developed by Deep Medical, will use algorithms and anonymised data to break down potential reasons why a patient might not attend an appointment. It will use a range of external insights such as the predicted weather, traffic and patients’ jobs. Based on the data, the appointment will be rearranged for the most convenient time for patients.
The NHS says that Deep Medical’s software will back-up bookings to ensure no clinical time is lost.
The system also implements intelligent back-up bookings to ensure no clinical time is lost, maximising efficiency.
It is being trialled by Mid and South Essex NHS Foundation Trust, which supports 1.2 million people. Its rate of ‘did not attend’ (DNA) appointments averages 8%. The national average is 6.7%.
New NHS AI push could save it billions
In January 2023, NHS England launched its drive to reduce an estimated 7.8 million missed hospital appointments per year, which equals a £1.2bn cost to the health service. Of the 122 million appointments booked in 2021/22, around 6.4% were missed – around 650,000 a month.
Missed appointments contribute to longer waiting lists. In January 2023, NHS England said that around four in five people are on an NHS waiting list for an outpatient appointment. Over three in five outpatient appointments are follow-ups.
According to NHS England, when at full scale the AI software could free up an additional 80,000-100,000 patients to be seen each year at Mid and South Essex. An additional five NHS Trusts are set to be tested this year. Tech Monitor contacted NHS England to confirm which trusts would be testing the software.
Amanda Pritchard, NHS chief executive, said that the pilot shows that the f NHS is at the “forefront of innovation,” using the latest technologies to ensure patients have the best possible experience.
“The system will help ensure patients receive ‘smart’ appointments, that are convenient and fit into people’s increasingly busy lives,” she said. “It is a win-win for patients and the NHS alike.” Pritchard added that smart appointments will help free up doctors to treat more patients, save taxpayers’ money as well as help to reduce waiting times.”
Bringing in AI isn’t about people ‘forgetting appointments’
Wider use of the technology could save the health service “hundreds of millions of pounds every year” according to Steve Barclay, health and social care secretary. Barclay said it would also “boost our efforts to cut waiting lists so patients can get the care they need more quickly.”
However, according to Deep Medical’s co-founder Dr Benyamin Deldar, the appointment of the AI software isn’t about people forgetting an appointment – it’s about challenging health inequalities.
“Whilst working on the frontline over Covid, I felt the impact of the pandemic on our services,” he explained. “We’ve created a system that can better identify and offer support for patients, whilst importantly tackling the NHS waiting list.”
Deldar is a member of the NHS Clinical Entrepreneur Programme (NHS CEP) – a development programme aimed at equipping NHS staff to develop commercial skills, knowledge and experience to help them digitally transform healthcare without leaving the national health service.
“Being a part of the NHS CEP has given me the framework and skills to scale innovations with purpose,” he said.
AI might predict missed appointments, but it can also get it wrong
This isn’t the first time technology has been used to solve the ‘missed appointments’ challenge. In 2019, University College Hospital, London, created an algorithm using records from 22,000 MRI scan appointments as part of a broader programme around machine learning in the NHS.
The algorithm identified 90% of patients who turned out to be a no-show. “On average we estimate this could save £2-3 per appointment,” Parashkev Nachev, UCL Queen Square Institute of Neurology and the National Hospital for Neurology and Neurosurgery at UCLH, told the Guardian.
However, it also incorrectly flagged around 50% of patients who did attend their appointments as a risk of not doing so.
Sam Smith, who leads policy at independent organisation medConfidential, told Tech Monitor that while data could help with DNAs, hospitals will still require sufficient infrastructure to be able to communicate with their patients.
“We look forward to reproducibility to show if this AI is any more useful than astrology,” he said.
Read more: NHS brings in extra scrutiny for data platform due to Palantir concerns
