With vaccination rates continuing to decline in various countries and misinformation spreading virtually unchecked, the risk of recurring infectious diseases such as measles is once again rising significantly. In the United States and Canada, this has already led to new outbreaks of diseases that had been under control for decades. Scientists at the University of Waterloo have therefore developed a data-driven method that can help public health services intervene earlier and more accurately.
The researchers analysed how vaccine scepticism develops via social media and developed a model that considers these digital dynamics as an ecological system. Just as viruses spread among humans, misinformation can move rapidly between users on social platforms. ‘We see social dynamics as a contagious process,’ explains Chris Bauch, Professor of Applied Mathematics. ‘Misinformation can spread in the same way as a disease, which gives us the opportunity to apply mathematical models that can predict tipping points in a system.’
Tipping point
Central to the research is the concept of the tipping point. This is the critical moment when a system abruptly shifts from stability to crisis. This principle is well known in ecology and medicine, but also appears to be applicable to social processes such as the decline of herd immunity. The team trained a machine learning model to recognise patterns in social media posts that indicate such an impending tipping point.
The method was tested on tens of thousands of public posts on X (formerly Twitter) from California in the period leading up to the major measles outbreak of 2014. Traditional prediction methods, such as simply counting sceptical posts, provided only a few days' warning time. In contrast, the tipping point analysis gave much earlier signals that vaccine scepticism in the region was gaining enough momentum to make an outbreak possible. By comparing data with regions where no outbreaks occurred, the researchers were able to confirm the reliability of their model.
Evidence-based decision-making
This approach fits within the University of Waterloo's broader strategy to strengthen evidence-based decision-making and trust in science. Through the TRuST initiative, philosophers, data scientists and communication specialists, among others, are working together to better understand how and why scientific trust erodes and how it can be restored.
Although the model has been tested on a text-based platform such as X, it can be adapted for TikTok or Instagram. Analysing image and video data does require more computing power. Ultimately, the researchers want to develop a practical tool that helps health authorities monitor populations and intervene early before an infectious disease actually spreads.
Stopping a pandamic with smartwatches
Five years after COVID-19 emerged, research shows that global pandemic preparedness has improved little, leaving the world vulnerable to future outbreaks.Earlier this year Finnish-American study explored how smartwatches could help detect infections earlier and reduce presymptomatic spread. Since nearly half of COVID-19 transmissions occurred before symptoms appeared, early detection is crucial for preventing large-scale outbreaks.
Researchers from Aalto University, Stanford, and Texas A&M found that modern smartwatches can predict COVID-19 infections with about 88% accuracy, and influenza with around 90%, using signals such as heart rate, breathing patterns, and skin temperature. When people learn they might be ill, they typically reduce social contact by 66–90%, which greatly lowers transmission. Modeling based on real-world epidemiological and behavioral data shows that smartwatch alerts could cut disease spread by 40–65%, even with moderate compliance.
While technology alone cannot replace broader preparedness, wearables may offer a valuable, less disruptive tool for future pandemic response.
