Researchers used AI to sift through over 400,000 Reddit posts, uncovering potential side effects of popular weight-loss drugs like Ozempic and Mounjaro that might not be fully captured in clinical trials.
A University of Pennsylvania team analyzed more than five years of discussions from nearly 70,000 Reddit users. The study, published in Nature Health, highlighted two areas needing more investigation: reproductive symptoms, such as menstrual cycle changes, and body temperature issues like chills and hot flashes. It's important to note that these findings don't prove the medications caused these symptoms, but rather signal potential areas for closer examination.
"The underreported symptoms are leads that came from patients themselves, unprompted, and clinicians could potentially pay attention to them," said Sharath Chandra Guntuku, a senior author on the study.
While clinical trials identify major side effects, they might miss symptoms that patients find more concerning in everyday use. Social media, despite not being perfectly representative, can offer a large dataset of patient experiences. The study's first author, Neil Sehgal, mentioned that while they can't confirm causality, nearly 4% of Reddit users in their sample reported menstrual irregularities, suggesting a signal worth investigating.
This approach of using social media for drug safety isn't new, but advances in AI, particularly large language models, have made it possible to analyze vast amounts of informal text more efficiently and accurately. The AI can translate everyday language into standardized medical terms, speeding up the process significantly.
The analysis showed that gastrointestinal issues were the most commonly reported side effect, aligning with known effects. However, reproductive symptoms and body temperature changes were also frequently mentioned, along with fatigue, which appeared more often in the Reddit data than in some clinical trial reports. Researchers suggest that because these drugs interact with the hypothalamus, a brain region regulating hormones and body temperature, these reported symptoms warrant systematic study.
The team hopes these findings will encourage further research and prompt healthcare professionals to consider symptoms that patients frequently discuss online. They also plan to expand their analysis to non-English speaking communities and other platforms to see if similar patterns emerge globally, potentially creating an early warning system for emerging health concerns.
