COVID-19: the open data pandemic?

As the world grapples with how to forecast and forestall the impact of COVID-19, researchers have stepped up, producing mathematical and computational models that predict the pandemic’s spread and its consequences at breakneck speed. 

In many cases, COVID-19 research is posted online on pre-print servers designed for papers that are yet to be formally vetted in a peer-review process. Since January, scientists have posted almost 2500 articles to two such servers alone – bioRxiv and medRxiv.

“There’s been an unprecedented use of preprint servers and it has served a really important purpose,” says Raina MacIntyre, head of the Biosecurity Program at the Kirby Institute at Australia’s University of New South Wales. 

“That has meant that people have had access to the data straight away, to give them information that could be important in terms of a whole range of issues, whether it’s public health control or clinical treatment or interventions for flattening the curve.”

This rapid, open sharing of data is a first for global pandemics. 

Last time, during the H1N1 “swine flu” pandemic of 2009, bioRxiv and medRxiv were still years away, and formal publication in journals lagged weeks or months behind the spread of the virus. 

“Research was often only available through publication in journals,” says James McCaw, an epidemiologist at the University of Melbourne, Australia, who has been modelling the COVID-19 pandemic and advising government officials.

The change reflects a broader movement in science towards a culture of open sharing of results, and digitally archiving datasets so that researchers can view, use and appraise each other’s work. 

Now, an international group of scientists is calling for researchers to be even more open. 

“At this time of crisis, it is more important than ever for scientists around the world to openly share their knowledge, expertise, tools, and technology,” they write in a letter published in the journal Science. 

Scientists frequently publish the results of their modelling efforts, but they often don’t share the lines of code underlying their models. 

“We have to open up the hood,”, says Columbia University climate impact modeller Cynthia Rosenzweig, a co-author of the Science letter. 

This would mean that all of the assumptions, algorithms and sources of data built into a model can be scrutinised by others. 

George Milne, a complex systems modeller at the University of Western Australia who is advising the Western Australian government, says that sharing of data is particularly important when policy decisions hinge on that information. 

“Not all modelling may be useful and not all may provide ‘defendable’ and clear guidance to health authorities,” he says. “This is the whole point of making models and their results available for scrutiny.” 

There are other upsides to openness. For one, it can accelerate science, says Rosenzweig. 

Rosenzweig co-leads the Agricultural Model Intercomparison and Improvement Project (AgMIP) a global community of more 1000 climate, crop, livestock, environment and economic modellers who share models to predict how global agriculture and food security could be affected by climate change. 

Sharing of models via data repositories ultimately leads to the development of better models, faster, according to Rosenzweig. 

The same could happen for COVID-19. “What COVID is showing us is we’ve got to do the science faster and better,” she says. 

Dyani Lewis will take an in-depth look at COVID-19 modelling in the forthcoming issue of Cosmos magazine. You can subscribe to the magazine here.