University of Southampton researchers to use machine learning and big data to tackle NHS winter pressures

Researchers from the University of Southampton are involved in a programme of rapid research projects aiming to ease winter pressures faced by the NHS.

This year, the NHS is under considerable strain compounded by COVID-19 and a record flu season, and the cost-of-living crisis.

The Southampton studies have been launched by Health Data Research UK (HDR UK) with funding from the National Institute for Health and Care Research (NIHR).

Dr Nazrul Islam, Associate Professor of Epidemiology and Medical Statistics, and Dr Hajira Dambha-Miller, an NIHR Clinical Lecturer in General Practice and a practicing GP, will lead a project to understand which combinations of Multiple Long-Term Conditions (MLTC) are associated with the highest risk of hospitalisation and death over the winter season.

Around 14 million people in England live with MLTC – two or more chronic conditions such as diabetes chronic obstructive pulmonary disease and arthritis. People living with MLTC have a higher risk of being admitted to hospital and of dying during the winter season. However, it is not known which combinations of MLTC are associated with the highest risk of these poor health outcomes, and if vaccinations (against COVID-19 or seasonal flu) can lower this risk.

Dr Islam and Dr Dambha-Miller’s project will use machine learning approaches to analyse patient data to answer these questions to help develop interventions and strategies to reduce the risk for this vulnerable group of patients.

Dr Dambha-Miller said: “This work will complement our ongoing research on multiple long-term conditions and social care needs and advance our understanding of the risk and vulnerabilities of people living with MLTC.”

Dr Islam added: “This is one-of-a-kind initiative using such a large volume of data to produce robust evidence on the combinations of long-term conditions that are associated with a higher risk of winter hospitalisation and death. Our findings will potentially save lives through a combination of identification of high-risk groups for earlier or prioritised implementation of preventive measures, and proportionate allocation of resources.”

Dr Islam is also working on two other projects within the programme with his collaborators at the Office for National Statistics and University College London.

Professor Michael Boniface, Director of the IT Innovation Centre at the University of Southampton, will lead a project to tackle complex hospital discharge processes. The work brings together NIHR ARC Wessex, the University of Southampton, University Hospital Southampton and Hampshire Hospitals.

Professor Boniface explains: “Currently around 20 per cent of beds in Southampton General Hospital are used by patients who are well enough to leave hospital but can’t because of delays arranging onward care. Using computer algorithms, we aim to help the NHS understand what causes patients to be delayed and use the insight improve discharge planning.”

This work is the continuation of a project – called PROCED – already underway at NIHR ARC Wessex.

Dr Dan Burns, of Digital Health and Biomedical Engineering (DHBE), will work alongside Professor Matt Inada-Kim from Hampshire Hospitals NHS Foundation Trust on a project called Predicting Hospital Length of Stay in Acute Respiratory Infections Patients.

Dr Burns explains: “We aim to find patients with acute respiratory infections – or ARIs – who are at higher risk of a long hospital stay and find ways of predicting whether a patient will have a long stay when they next get an acute respiratory infection. A very high proportion of admissions throughout the winter are ARI-related and we need to improve this pathway to help ease the impact of winter.

“We want to understand how individual patients recover from ARIs and this will involve looking at a wide variety of factors: age, gender, ethnicity, previously diagnosed diseases, staffing levels in their wards, and how they recovered from previous infections. By understanding how length of stay is affected by these factors, the hope is to develop more effective ways of monitoring and treating ARI patients, thus reducing the amount of time patients are in hospital for and reducing the impact on the NHS.”

Professor Matt Inada-Kim has previously worked with Professor Boniface and Dr Burns on using machine learning and health statistics to support people recovering from Covid infections in the community.

Overall, 16 projects have been launched across the UK. Each project is designed to generate findings in just a few months so that they can be implemented for future winters. After being selected in December 2022, the studies will start in January, produce results by the end of March, and publish their findings later this year.

Professor Cathie Sudlow, Chief Scientist at HDR UK – the UK’s institute for health data science, which is delivering the projects – said: “As a doctor who has previously treated patients in the emergency department, I am all too aware of the enormous challenges faced by the healthcare system this winter. It’s critical that we use data rapidly, securely and responsibly to support the NHS, its workers, and the patients who rely on it for their care.

“By using existing data, research teams, and infrastructure these projects are able to respond rapidly to evolving pressures on the NHS. Within three months, they will have honed in on key pain points in the health service, and developed evidence-led recommendations on how best to manage resources and prevent unnecessary illness through the winter.”