How AI is Helping the NHS Tackle Missed Appointments

Missed hospital appointments – known in NHS-speak as ‘did not attends’ or DNAs – are more than just an inconvenience.

In 2024, there were 11.8 million of them, costing the NHS nearly £1.9 billion. That’s money lost, but the real impact goes deeper. When appointments are missed, it’s often the people who need care the most who lose out – those with long-term conditions, mobility problems or little support at home.

People living in deprived neighbourhoods are twice as likely to miss their appointments as those in wealthier areas. The reasons? Everything from unreliable transport and rigid work schedules to caring duties and language barriers.

Why predicting isn’t enough

Of course, reducing missed appointments is vital. But it’s complicated. On the surface, predicting who might not turn up looks like a data problem. In reality, it’s about much more than that. Behind every patient is a unique set of circumstances and the reasons for not attending often don’t show up in routine hospital records.

The challenge isn’t just predicting who’s at risk, it’s also uncovering the why – and making those barriers visible so staff can step in with real solutions.

Building an AI model with the NHS

To take on this challenge, we teamed up with Royal Berkshire NHS Foundation Trust (RBFT), one of the UK’s largest trusts, caring for around a million people. Together, we developed an explainable artificial intelligence (XAI) model.

Using data from over 500,000 past appointments across more than 80 specialties, the model can predict a patient’s risk of not attending with around 92% accuracy. What makes this project unique is that it doesn’t just flag high-risk patients, it also shows the specific factors behind that risk – things like past DNAs, long travel times, mental health issues or deprivation levels.

Turning insight into action

The model was built into a decision-support system that NHS staff can actually use day to day. With clear guidance materials, staff can have better conversations with patients flagged as high risk.

Instead of a generic reminder call, they can talk through the issues and offer tailored support. That might mean covering travel costs, arranging an interpreter or finding a flexible appointment slot for someone with caring responsibilities. It’s about using AI insights to guide compassionate, personalised care but not replacing the human touch.

The impact so far

Since 2023, the system has been in use across all RBFT sites and departments, supporting nearly 700,000 outpatient appointments in 2023/24 alone. The results are clear: non-attendance among high-risk patients has been cut by 40%. Patients have also told us that the personalised approach made it easier to attend appointments they’d otherwise have missed.

Looking ahead

The success of our work, partly funded by the UK Research and Innovation Economic and Social Research Council (UKRI ESRC), has paved the way for more AI projects supported by the UKRI and the National Institute for Health and Care Research (NIHR). These include tools for earlier disease detection, forecasting disease activity and helping clinicians decide who should be cared for in virtual wards.

AI won’t solve every problem in healthcare. But used in the right way, it can help make care more equal, more effective and more human.

Authors