In early March 2020, on the cusp of the global acceleration of the pandemic, a popular Twitter user called ZDoggMD (in real life, a doctor named Zubin Damania) launched a pro-vaccine rallying cry: #DoctorsSpeakUp. The hashtag, intended to call on real doctors to share the positive realities of immunization with the world, was instead almost immediately hijacked by anti-vaccines. Recently published research by a team at the University of Pittsburgh used supercomputers to understand how the event went wrong – and how similar efforts might be protected from such hijacking in the future.
Using Twitter’s filtered feed interface, the researchers extracted all publicly available tweets using the hashtag #DoctorsSpeakUp on March 5, 2020. Five percent of those tweets – around a thousand – were assessed using thematic content analysis, allowing researchers to study associations between tweets. sentiment, account type (likely human or robot) and content of the tweet (e.g. personal story, statement, etc.). Researchers used a tool called Botometer to assess the likelihood that a given account is likely to be a bot.
To perform this data intensive analysis, the researchers turned to local supercomputing resources at the Pittsburgh Supercomputing Center (PSC). There, they used the Bridges System for a while before it was retired in mid-February 2021, when they switched to the Bridges-2 System (which officially began production operations this spring).
“We worked with [the] Pittsburgh Supercomputing Center since before Bridges, has been running for the duration of Bridges, and we are now on Bridges-2, ”said Jason Colditz, University of Pittsburgh researcher and one of the authors of the article, in an interview with PSC Ken Chiacchia. Colditz noted that there are “terabytes and terabytes of data that we have collected on Twitter over the course of several years,” but the data has moved quickly, requiring stability and availability. “And that’s really where working with PSC has been beneficial,” he said.
Using the analyzes powered by a supercomputer, the researchers came up with valuable information: 78.9% of all tweets studied were anti-vaccination; 79.4 of tweets from users claiming to be healthcare professionals supported vaccination; and 96.3% of tweets from users claiming to be parents (but not healthcare professionals) were anti-vaccination. While bots made up only a small portion of tweets, tweets from anti-vaccination bots were five times as numerous as tweets from pro-vaccination bots. In addition, a higher percentage of anti-vaccination tweets linked to scientific information compared to pro-vaccination tweets, although the researchers noted that anti-vaccination tweets were likely to distort the research.
The researchers concluded that the hijacking was a “highly coordinated response of dedicated anti-vaccine antagonists.” Going forward, they noted, “it would be helpful to ensure that pro-vaccine messages consider hashtag use and pre-develop messages that can be initiated and promoted by pro-vaccine advocates. “.
“This is, I think, a really good time to look at social media to get a sense of what’s going on in these communications,” said Colditz, “and how we might, as public health advocates, being able to alleviate some of that… tough road we see with people hesitant or downright opposed to indulging in vaccinations for the current pandemic. ”
To read the document, which was published in the May 2021 issue of Vaccine, Click here.
To read the report on this research by Ken Chiacchia of PSC, click here.