Being under development since 2022 by researchers at the University of Waterloo, the UbiLab Misinformation Analysis System (U-MAS) has quietly emerged as one of the most robust AI tools in the fight against health misinformation online. Now in active use, the system is helping researchers and public health officials monitor the ever-growing tide of false health claims circulating on platforms like X and Instagram.

“If [these false claims are] shared and become an infodemic,” warns PhD student and lead developer Irfhana Zakir Hussain, quoted by as quoted by CTVnews, “that becomes a very bad public health outcome.”

Built as a big-data ecosystem, U-MAS has been designed to detect and analyze misleading content on social media in real time.

According to its developers, U-MAS offers exactly the kind of early warning system public health experts need. “Ideally, you’d be using the system for a variety of use cases and monitor them over time to see what needs to be spoken about and what does not,” explains Hussain.

U-MAS has already tracked false claims about the war between Russia and Ukraine, and its current focus is on vaccine hesitancy and misinformation related to fluoride, heatwaves and diet.

In one study, the system scanned 500 Instagram posts containing the term “fluoride-free.” While many reflected lifestyle choices, others veered into conspiracy territory.

Supported by co-investigator Dr. Zahid Butt and principal investigator Prof. Plinio Morita, U-MAS is built using a combination of Python, the Twitter API, and the Elastic Stack. Its five major components include a topic model (LDA), sentiment analyzer, and misinformation classifier, all trained on expert-validated data.

In tests, the system performed impressively. “The LDA topic models achieved relatively high coherence values,” researchers noted. The sentiment analyzer performed at a correlation coefficient of 0.72 but could be improved in further iterations. The misinformation classifier attained a satisfactory correlation coefficient of 0.82 against expert-validated data.”

The tool’s developers hope to make U-MAS available to a broader set of users in the near future. Currently, it pulls data from X (formerly Twitter) and Instagram, but there’s an ambition to include YouTube, Facebook, and other platforms soon.

A 2022 WHO study found that over 60% of pandemic-related posts on social media contained some form of misinformation. In Canada alone, COVID-19 falsehoods were linked to over 2,800 preventable deaths in 2021, according to the Council of Canadian Academies.

Also read: Turmeric supplements cause liver damage: US woman’s case exposes dangers of unverified health advice 

Do you have a health-related claim that you would like us to fact-check? Send it to us, and we will fact-check it for you! You can send it on WhatsApp at +91-9311223141, mail us at hello@firstcheck.in, or click here to submit it online