‘Break out of silos’ to share research technology and data

Tying state-of-the-art digital research infrastructure to particular research communities should give way to more equitable landscape, conference hears

April 16, 2024
A falling silo, illustrating moving beyond siloed education tracks
Source: iStock

Working with and accessing sensitive data has always come with a bureaucratic burden, but both the barriers to and opportunities of data-fuelled research are rapidly shifting, the Digital Universities UK conference has heard.

Speaking at the Times Higher Education event, Richard Gunn, programme director for digital research infrastructure at UK Research and Innovation, described the efforts being taken by the funding body to rethink the national digital research infrastructure for a new age.

This presented an opportunity to address problems with the current landscape, such as the inequality that exists between different disciplines.

“In the past you’ve had siloed provision, with state-of-the-art digital research infrastructure tied to particular research communities,” he told the event, held in partnership with the University of Exeter.

“That inequality led to real barriers, for example to interdisciplinary research, and creates all sorts of practical difficulties too. For us, it’s about breaking out of silos of admittedly excellent activity and enabling researchers to access the right technology and skills to solve a particular problem.”

In order to improve on the status quo, he said, it was important to acknowledge why the current uneven approach had arisen – not least that digital infrastructure is “a complex adaptive system, driven by bottom-up factors”, which meant that a “command and control approach” to replacing it would not work.

What was required, he argued, was “a change programme in three dimensions: technology, policy and culture, and as we know it’s often the culture the requires the most time.”

But the prize for improving the current landscape, he said, was considerable: “We want to move away from this idea that your access to digital infrastructure is depending on your funding council or research topic…[That] way of doing things is unsustainable, we need to move on and we know if we can get this right it can be transformational.”

Speaking alongside him was Emma Gordon, director of the Administrative Data Research UK strategic hub at the Economic and Social Research Council, who also acknowledged problems with the current approach to data access.

She described a frustrating process in which researchers can get stuck in a loop of applying for access to owners of often highly sensitive datasets, getting feedback on the request, resubmitting and so on – often resulting in months of delay.

This could be compounded by a lack of clarity in advance about the data that researchers will ultimately get access to – a problem, she suggested, that could be resolved if they were able to access “low-fidelity synthetic data”.

This would be in a form that “describes the real data, but doesn’t maintain the statistical relationship between them, and so it couldn’t ever be misused, but it would allow you to understand what variables are there, how they are structured, and to develop and test code before you get access to the data. It would also allow you to develop a much better research application to access the data, because you would properly understand it.”

Such an approach would, she argued, save considerable time and expense on all sides by reducing the demand for costly computational time to analyse data within a live environment, allowing researchers to develop code outside of the system, and improving outputs.

A third speaker at the session, which was titled “Optimising the UK’s digital research infrastructure in support of a low-bureaucracy, impactful research culture”, was Michael Ball, head of data science at the Medical Research Council.

He highlighted the transformation of medical sciences in recent years, with ever greater use of data to the point now that some disciplines are almost entirely computational.

As a result, he said, data used by researchers now was not just gathered by the teams in question, but increasingly gathered externally, and “access to someone else’s data requires additional management and bureaucracy”.

The NHS, he said, was pioneering a new approach to access via its “secure data environment network”, for which it was “explicitly saying the model is shifting from download of data to accessing it where it sits”.

“That requires new thinking – you are now taking your tools and computation to the data, rather than taking it away and using it in your own environment,” he explained.

“But if that sounds more difficult, there is a flip side to this, and that is that the NHS is trying to make its data more available for research, so there’s a great prize to be had here that can open up whole new worlds of research.”


Register to continue

Why register?

  • Registration is free and only takes a moment
  • Once registered, you can read 3 articles a month
  • Sign up for our newsletter
Please Login or Register to read this article.

Related articles

Reader's comments (1)

What about the owners of the data, who gave consent presumably that the data would be used for a specific purpose, once you open the silo, you risk creating risks for loss of privacy. Simple anonymisation is not enough. In the longer term you risk communities withdrawing cooperation if the scope & integrity of the study cannot be defined and guaranteed.