COVID-19 data analytics research supports frontline

17 Jun 2020


Back in March 2020 Professor Michael Boniface and his team of mathematicians, physicists, statisticians, data scientists and engineers set about an unlikely task. Could they use data and other information sources with mathematical models of infectious disease to help Southampton's emergency department forecast the number of people likely to be admitted?

What's unusual is that the team are not doctors, nurses or medical professionals, but a group of scientists and engineers turning their skills to support the health service at a time of a pandemic.

Michael says "The sheer notion of rapidly bringing a team together to provide real-time answers and forecasts for the NHS in order to help save lives shows the value of diverse teams, especially data analysts working with clinicals to solve complex health system challenges"

"We had to trade off some elements," he continues. "What we needed was timely results, rather than perfect ones, and an approach that improved over time as global understanding of the disease and its impact increased"

Michael is the Director of the IT Innovation Centre, Electronics and Computer Science at the University of Southampton, and for over 20 year’s he’s been working with industry and public sector organisations, including healthcare providers, to innovate ways of modelling and connecting data and information for public and provider benefit.

Wessex NIHR ARC Director Alison Richardson, had seen Michael's work and offered the support of the ARC making him part of the Workforce and Health Systems theme.

What the team has discovered is that they were able to map and forecast the numbers of patient admissions to the University Hospital Southampton, and for the ICU and general Covid-19 ward.

It is what Professor Michael Boniface describes as 'hyper-local' modelling, looking at just one hospital and the factors behind demand for beds and resources.

The modelling and reports offered NHS managers a range of scenarios too, factoring in uncertainties like social distancing polices. The reports were one source of intelligence helping the hospital manage limited healthcare resources at a time of crisis.

The work has been in shared with the wider ARC modelling and Operational Research teams, and across the Wessex region, with Dorset exploring its use.

A further bid for funding is now to bring the research together with lessons learnt from other major hospitals in Spain and Italy, helping European health systems recover and reimagine emergency care delivery.

Team

Acknowledgements

We thank Professor Ben Macarthur and Dr Francis Chmiel for help and support in the early development and validation of the SIHAUDR epidemic model, and Professor Dave Woods and his colleagues who have recently joined the team to explore ways to improve handling of uncertainties.