MIT seeks to speed up pandemic research with AI-aided journal

Hoping to speed up coronavirus-related research, the Massachusetts Institute of Technology is creating a new journal using artificial intelligence tools to rapidly compile scientific findings for open review and sharing.

The online journal, called Rapid Reviews: Covid-19, will cull articles from preprint formats, add peer reviews and openly share the results regardless of whether the process ends in a traditional decision to publish.

RR:C19 will be managed by MIT Press and led editorially by Stefano Bertozzi, a professor of health policy and management at the University of California, Berkeley.

Its creation was driven by several objectives for fighting the pandemic and perhaps beyond, said Amy Brand, director of the MIT Press, including increased speed and scientific accountability at a moment that is demanding rapid advances and greater interdisciplinary integration.

“We want to align with what the research community is doing and what it wants,” Dr Brand said. “But we also want to build in more quality control and more accountability, and that can exist certainly beyond Covid-19.”

The structure is similar to the system in which the Gates Foundation has been openly publishing peer reviews on research work conducted by its grant recipients, Professor Bertozzi said.

One key goal of such systems, he said, was to avoid situations in which a poor piece of scientific work was sent to multiple journals then published by the only journal where reviewers overlooked the flaws.

“Basically, you get ‘no’ from Dad. So then you go to Mom, and Mom doesn’t know that Dad said ‘no’ and why. And so you ask Mom, and maybe she says ‘yes’,” Professor Bertozzi said of current practices in academic publishing.

RR:C19, established with a $350,000 (£280,000) grant from the Patrick J. McGovern Foundation, is the latest in a series of efforts to help coronavirus-focused scientists triage a flood of inputs of varying reliability from their colleagues.

Another initiative, outlined last month by several major open-access publishers including Plos and eLife, created a common pool of experts in virus-related fields who stand ready to quickly review journal submissions on a shared basis.

That coalition now has more than 1,600 reviewers. Plos also is eager to proactively review submissions on preprint servers, and to incorporate comments left on preprints into its formal peer review processes, and it has been talking with MIT Press and others about expanding such methods, Plos representatives said.

For now, Dr Brand said, the McGovern Foundation support means that RR:C19 will not charge author fees – the typical method that open-access journals use to support themselves without subscription revenue.

Aided both by AI technologies and graduate students with expertise in a variety of fields – including economics, psychology and mental health – RR:C19 hopes to review dozens of articles a month, adding the peer reviews and any resulting exchanges with the authors.

In some cases, the end result will appear as an RR:C19 article; in others, the product will remain on the preprint server where it was found. That second option, Dr Brand said, hopefully will give researchers the freedom to let RR:C19 quickly assess and highlight their findings without abandoning the possibility of later publication in a more prestigious journal.

Even if scientists quickly find effective Covid-19 treatments or cures, Dr Brand hopes that RR:C19 will show lasting value. Its life beyond the pandemic, therefore, could be a measure of how effectively it helps to transform scientific norms and processes. “It will evolve accordingly,” she said.

paul.basken@timeshighereducation.com