The unpredictability surrounding the spread of COVID-19 was one of the biggest challenges facing colleges at the start of the autumn term.
As colleges grappled with the return to face-to-face teaching that the government encouraged, there were crucial decisions to make. How could they strike the right balance between maximising on-campus teaching time and minimising the spread of the virus? There were many questions, but few definitive answers.
Using data modelling to be proactive
While musing over these problems during the summer, Sean Mackney, principal at Petroc College in Devon, realised that data modelling could help.
He decided to seek help, which he found on the college’ doorstep at the University of Exeter’s institute for data science and artificial intelligence. Working under Professor Gavin Shaddick, data scientists had already developed COVID-19 prediction models for Devon NHS trusts, and readily agreed to help Sean and colleagues at the Association of Colleges (AoC) create something similar for the education sector.
As a result, an app unique to further education (FE) was launched in early November, which means colleges can be proactive, rather than reactive, around COVID-19 outbreaks and rises in infection rates.
Its concept is simple: colleges enter information about their local community infection rates using figures obtained from the Office for National Statistics, and choices they could make about bubble sizes, social distancing measures, attendance rates on different days and timetabling over a set period of time.
The tool, available for free to all UK colleges, then shows likely infection and staff and learner self-isolation rates for each scenario.
'Effective use of technology'
Although still in beta stage, more than 100 colleges are now using it and it has quickly proved its worth. The app has already been shortlisted for an AoC Beacon Award in the Jisc-sponsored category for ‘effective use of technology’. More importantly, it has positively impacted colleges’ decision-making around COVID-19.
Sean cites two particular examples at Petroc:
“Firstly, we discovered quite early on in our use of the app that, if we take learners off campus for a block of time, like a long weekend for instance, that has a significant impact on lowering infection rates and, therefore, lowering the numbers in self-isolation.
“As a result, when we were planning for the start of the September term, we designed with a blend of digital and in-person teaching because the model told us that a mix of on-campus and remote learning was ‘safest’.
“Secondly, we knew that learners in practical disciplines had suffered disproportionately from having to study online. They were not making progress as rapidly as others taking academic subjects because they couldn’t develop practical competencies.
“The solution is to bring them on to campus for all their learning. So, we put them altogether in a bubble that doesn't have any contact with anybody else; they eat in that same bubble and they must stay in this one zone of the campus.
“We ran this proposal through the model to see whether it had a detrimental impact on the level of self-isolation, and it didn't. So that information enabled us to make life better for those learners and, hopefully, improve their outcomes.
“Going forward, we expect that, given the link between engagement and success, colleges’ use of the app will have a positive impact on retention and achievement.”
Tacking digital disadvantage
The app has also allowed colleges to tackle digital disadvantage and optimise access to education for all learners.
Some learners, for example those who don’t have a device or adequate broadband at home, need more in-person learning on-site for a greater proportion of their study, and the data modelling gives confidence that this can be managed in a COVID-secure way.
Sean and the FE sector is indebted to Gavin and his team at the university, who offered their time and resource for free – as they did for the NHS since the beginning of the pandemic. Funding towards the FE app’s development was later secured from City and Guilds, but money was never an issue.
“We really wanted to help in any way we could, and it was very fulfilling to be able to use our skills and expertise in a meaningful way. I think it's about the most important thing a statistician could do, to chip in and help in this time of crisis. It was the right thing to do and the team has been utterly committed to both the NHS and FE apps.
“There's a team of about five people working on these projects, including both staff and students. They’ve been absolutely amazing and incredibly dedicated, working a lot of late nights and early mornings to fit in this work in with their regular jobs and studies.
“The collaboration has been important to us, too – working in partnership with the NHS and education locally. “
But none of this would have been possible without buy-in from university leaders, as Gavin says:
“Right from the outset we’ve had support from the vice chancellor’s executive group, and Exeter IT, which provided us with computing hardware and IT solutions in incredibly short turn-around times. All of this enabled us to work at speed.
“The nature of this work and the need for solutions almost immediately meant we were working at a pace that we are not often used to in academia. We have certainly learnt a lot through these collaborations, with real-life problems and time constraints driving innovation and close collaboration. It's been very exciting - data science in the fast lane.
“The NHS and colleges needed something that worked and was accurate as soon as possible, and so we have refined the modelling and apps as we go along. It’s been great working closely with users to develop the work and to ensure that it can produce the information they require to make decisions.”
Because the app was developed in partnership with stakeholders, it’s easy to use. Sean’s executive leadership team has responsibility for plotting COVID-19 scenarios for Petroc and is doing so regularly.
“It’s an ongoing piece of work. The app’s beauty lies in its simplicity and flexibility – you don’t need a degree in data modelling to be able to use it. Once you’re used to it, it takes about 10 mins to input the data for a scenario and we use the results to respond quickly to change the college set up, if we need to.
“That could mean, for example, that if we know the number in self-isolation is likely to increase significantly then we can invest more time, money and energy into improving and expanding our digital learning offer.
“The modelling enables us to strike the right balance between keeping our learners learning and the teachers teaching on campus and keeping them all safe.”
Colleges can find out more about the app and how to access it by contacting Tammi Jahan (email@example.com) at the AoC.