Nottingham University NHS Hospital Trust
Using predictive insights to drive down waiting lists – an NHS first
Following the fallout from the COVID-19 pandemic (2020 – 2021), National Health Service (NHS) waiting lists in the UK are at an all-time high. Getting the numbers down is critical and every bed matters – but understanding how many will be available at any given time, particularly in the face of ongoing COVID-19 disruption, is difficult. To reduce waiting lists, NHS trusts must use the resources that are available as efficiently as possible.
Working with Nottingham University NHS Hospital Foundation Trust (NUH), we combined expertise in data and analytics with in-depth knowledge of the healthcare sector to develop and launch a digital-twin-enabled bed scheduling tool. This provides a live picture of predicted bed usage across the Trust for the month ahead, reducing the number of cancelled life-changing operations and helping ensure precious beds don’t lie empty.
The tool, as the first of its kind in an NHS hospital, leverages patterns in patient data from the past four years to make its predictions. Thanks to artificial intelligence, the tool’s predictions are increasingly accurate – and achieved over 95 percent accuracy against actual bed occupancy within three months of roll-out. That enabled the Trust to schedule extra operations. Over the next year, NUH anticipates saving £1.6 million from a reduction in the number of cancelled operations – although challenges with staffing and bed pressures due to delayed transfers of care may impact on realising the full opportunity. The new scheduling tool will also help drive waiting lists down meaning thousands of patients won’t have so long to wait for life-changing operations.
- Developed the first-ever digital-twin-enabled elective scheduling tool for an NHS hospital, which is reducing wait times across the Trust
- Achieved over 95 percent accuracy in predictions of bed use
- Enabled additional life-changing operations to be scheduled and performed within three months of roll-out
Supporting the road to recovery
The impact of the pandemic on NHS waiting lists means millions of people are waiting longer for non-emergency life-changing operations. These include procedures like cataract procedures, hip and knee replacements, hernias, and prostate surgery. At Nottingham University NHS Hospital Foundation Trust (NUH), for example, 60,000 people are waiting for treatment. The average wait time is 11 weeks, up from just under eight before the COVID-19 pandemic.
Making the best use of available resources, including operating theatres and beds, is key to getting waiting times down. But for the teams that manage bookings, making full use of all beds across the Trust’s sites is a complex business. COVID-19 infections among clinical staff, for example, can result in whole lists of operations being cancelled at short notice, resulting in empty beds on wards a few days later. For patients this can lead to longer, more painful, and sometimes life threatening wait times. With better insight earlier, these beds could be booked for other patients next in line for other procedures. Equally, scheduling teams might cancel operations because they’re not confident there will be a bed available for patients after surgery. The opportunity to fix this could not be more urgent.
“Using our finite resources for maximum impact is key to recovering from the pandemic and reducing waiting times for patients,” says David Campion, Divisional Manager at NUH. “It’s a mission-critical challenge and we were confident PA could help us find a way to overcome it.”
Combining digital and healthcare expertise
Building on our long-lasting and fruitful relationship together, NUH is now one the most advanced trusts in the country in terms of using data to inform strategic and operational planning, and optimise services for patients.
Our leading data science and analytics capabilities, with experience of developing sophisticated predictive solutions within the NHS and other sectors, were perfectly suited to the challenge. They drew on a deep well of previous experience from various projects at the junction of health and technology and applied knowledge to forecast likely length of stay in different parts of the hospital.
A user-first approach
The frontline staff supporting the COVID-19 pandemic response and their needs as primary users were uppermost in our minds from the start. What was the challenge for the booking teams in offering patients a date for their treatment 28 days ahead of their operation? What information did they need from the tool to book patients a date with confidence? The insights gathered at this stage, through interviews with stakeholders, formed the foundation of our work.
“We wanted to make sure the tool would really meet the needs of the scheduling teams. We also wanted to make sure it could eventually be adapted to support similar resource management challenges across different hospital departments,” says David Dalby, ENT Specialty Manager at NUH.
Years of patient clinical task data held by NUH within NerveCentre – five million rows in total – became the key to unlocking vital insights into future bed occupancy. The data captures each patient’s prognosis on admission, logs their route through the hospital, and records how long they stayed. To uncover the insights the data held, we wrote the algorithms that are the beating heart of the ‘digital twin’– a digital copy of the hospital which uses resources like beds just as the actual hospital would in the future.
“With the digital twin, we’re extracting the full value from our extensive data sets for the first time,” says Mark Simmonds, Divisional Director for Medicine, NUH. “We’re using patient data to address live problems across the Trust, make informed decisions, and improve services for the thousands of patients we serve.”
Using the data provided by NUH, our experts built in artificial intelligence and machine learning algorithms into the solution. This allowed the tool to recognise new patterns in the data – like extended time in hospital due to multiple unplanned ward moves or the impact of a patients previous medical history – and refine predictions accordingly. As a result, these become increasingly accurate over time.
In collaboration with NUH, we overlaid the data and analytics with an easy-to-use interface that gives a visual representation of resource availability and makes insights meaningful for the scheduling teams. With further iterations, the tool better reflected user feedback, successfully raising the score gauging ‘customer satisfaction’ to 9.5/10. With the tool ready for wider use, our teams ran training for staff from different parts of the hospital.
Enhancing patient outcomes – one bed at a time
Working in tandem with the Trust, the result was the first ever ‘digital twin’ for use in an NHS hospital. The ‘digital twin’ allowed us to develop a bed scheduling tool providing an accurate overview of bed use for the month ahead. It updates three times a day, combining 120 simulations of the hospital’s future state into a prediction of bed use for different hospital wards, specialisms, and bed types.
Three months from roll-out, accuracy of predicted bed availability hit over 95 percent, meaning virtually no gap between predicted and actual bed occupancy. New insights from the digital twin quickly started to drive up the number of non-elective procedures the Trust carries out every month. Over the course of a year, NUH estimate reductions in cancellations and a subsequent rise in procedures completed will be worth about £1.6 million to the Trust.
Looking ahead, NUH will be able to use the solution to predict and prepare for future waves of COVID-19 and more familiar winter pressures. And, though our collaboration in developing additional functionality for our tools, we will soon be able to predict a medically safe for discharge date for every patient at the point of admission – transforming the way NUH manages patient flow. We’re also expanding our relationship with the Trust to take full advantage of the digital twin by creating outputs for other hospital departments including operating theatres and diagnostic testing capacity.
Every extra operation that can take place thanks to the digital twin represents a life made more comfortable. Every cancellation that can be avoided means one less patient going through the anxiety and upset of having their surgery delayed. Technology as deployed in the digital twin really is helping to create a more positive human future.
“We really value the partnership approach that PA brought to the development of data and analytics solutions,” says Colin Monckton, Director of Performance and Information at NUH. “And their focus on real-life use cases makes all the difference. Because of our collaboration, we’re able to get life-enhancing surgery to as many people as possible, as quickly as possible. It will literally change lives.”