SRU professor leads study of medical experts’ tweets about COVID-19


Covid 19 topics on Twitter

A study of medical professionals’ tweets sent during the early stages of the COVID-19 pandemic was led by Abdullah Wahbeh, a Slippery Rock University assistant professor of computer science.

July 16, 2020

SLIPPERY ROCK, Pa. — While the internet certainly has its fair share of misinformation, it is also chock full of facts and professional opinions about COVID-19. Part of the issue for those who truly want to be in the know is how best to sift through all the particulars to answer the question, "What are the medical experts saying about this?"

Abdullah Wahbeh, a Slippery Rock University assistant professor of computer science, started asking this question back in February and he did more than scroll through social media: he analyzed the data.

"Having information is very helpful but it depends on how you use it," said Wahbeh. "We have all this information on social media and can still be confused about what information we're supposed to believe or not."

So instead of reading a few tweets about COVID-19, Wahbeh relied on computing tools to analyze more than 10,000 tweets. Wahbeh and his co-investigators didn't sample just any tweets by the general public, but those written by top medical experts.

Wahbeh profile photo


"There are techniques using machine learning and data mining that can help by harnessing information much faster than what a human can do," Wahbeh said. "The idea is to see what medical professionals are talking about in terms of COVID-19 to have better insights in what they are recommending. People who are knowledgeable about the disease are more trusted than following any 'influencer' on social media."

Wahbeh was the lead author of the study, titled "Mining Physicians' Opinions on Social Media to Obtain Insights Into COVID-19: Mixed Methods Analysis," recently published in the open-source journal, the Journal of Medical Internet Research - Public Health and Surveillance. Co-authors included Tareq Nasralah of Northeastern University, Mohammad Al-Ramahi of Texas A&M-San Antonio and Omar El-Gayar of Dakota State University.

The researchers identified 119 medical professionals who were actively discussing the COVID-19 pandemic on Twitter by using two resources: a list of health care professionals compiled by the marketing software company Onalytica based on the firm's social media influencer score, as well as the Johns Hopkins Coronavirus Resource Center that provided a list of COVID-19 experts on Twitter.

They then ran search queries using resources from the analytics company Crimson Hexagon that extracted all the tweets from their list of medical experts between Dec. 1, 2019 and April 1, 2020. The researchers then began coding a subset of 250 tweets by relevant keywords and the data analytics tools picked up on the pattern and coded the remaining 9,800-plus tweets.

Through this process, the researchers identified topics the medical experts were addressing, such as actions and recommendations; warnings about misinformation; health care systems and workers; symptoms; immunity; testing; and virus infection and transmission. They used another tool, NVivo, to organize and analyze the data, before generating word clouds that created a visual representation of what was being communicated. This technique is known as "topic modeling," which is used by statisticians to identify the abstract topics that occur in a collection of documents.

As an example, Wahbeh said that early in the pandemic there was a lot of conversations in the media about increasing testing capacity but the medical experts were discussing solutions, such as drive-thru testing and relaxing protocols to allow hospitals to use their own testing kits instead of having them provided by government agencies or other third parties.

"These posts can help identify topics that are important to the community and can serve as a gauge for measuring concerns about potential threats," the researchers wrote in the paper. "This study provides a unique perspective of medical professionals during the early stages of the pandemic."

As the pandemic evolves, more accurate information will become available, including further tweets from medical experts and additional journal articles - beyond the more than 50,000 that have already been published - by researchers about COVID-19 topics since onset of the pandemic.

"The findings are only respective to the timeframe and if we did this one more time we might end up with different insights," Wahbeh said. "Every day something new is coming out."

Still, the process to analyze huge amounts of data in an evolving public health emergency can prove to be insightful, especially for policymakers and public health officials, according to Wahbeh.

In fact, what led Wahbeh to this study was another public health concern: e-cigarette use, aka "vaping."

Prior to conducting the COVID-19 study, Wahbeh used similar techniques for a separate study, titled "Health Risks of e-cigarettes: Analysis of Twitter Data Using Topic Mining," that was accepted for publication in 2020 Americas Conference on Information Systems, which will be conducted virtually in August. For that study, Wahbeh analyzed tweets to gauge the themes and perceptions of health issues associated with e-cigarettes. The main difference between the two studies was the COVID-19 study relied on medical experts and the vaping study gauged perceptions from the general public.

"The problem these days is you have the concept of 'fake news' or misinformation and any one of us can go online and post anything about any topic," Wahbeh said. "(You have all these) opinions and conflict between what people believe in and what they don't. (In this COVID-19 study) we are looking for trusted resources to avoid misinformation, so that's why we want to develop a way to see what the collective medical experts are talking about."

Read the entire study on the JMIR website. To learn more about computer science at SRU, visit the department's webpage.

MEDIA CONTACT: Justin Zackal | 724.738.4854 |