Penn Study Published in Nature Health Flags GLP-1 Symptoms for Further Study
The analysis turns a large, self-selected set of patient conversations into research leads—not evidence that semaglutide or tirzepatide caused the symptoms discussed.
Loading page…
The analysis turns a large, self-selected set of patient conversations into research leads—not evidence that semaglutide or tirzepatide caused the symptoms discussed.
Listen to this story
Researchers at the University of Pennsylvania used large language models and MedDRA mapping to structure 410,198 Reddit posts about semaglutide and tirzepatide treatments collected from May 2019 through June 2025. The analysis surfaced menstrual changes, temperature-related symptoms and fatigue as areas for follow-up, alongside expected gastrointestinal complaints. It demonstrates how AI can mine informal patient language for pharmacovigilance signals, but the results cannot establish incidence or causation.
The dataset included 67,008 users who identified themselves as taking one of the medicines.
Fatigue appeared in 16.7% of the symptom analysis; reproductive symptoms appeared in nearly 4% of users reporting side effects.
LLMs translated informal Reddit descriptions into MedDRA categories for cross-post symptom analysis.
A University of Pennsylvania study newly published in Nature Health used AI to analyze more than 400,000 Reddit posts about semaglutide and tirzepatide. The researchers found recurring reports of menstrual changes, temperature symptoms and fatigue that may deserve more study, but said the posts do not show that the medicines caused them.
The study examined discussions of semaglutide products, including Ozempic, Wegovy and Rybelsus, alongside tirzepatide products Mounjaro and Zepbound. Its underlying dataset contained 410,198 posts from May 2019 through June 2025; researchers identified 67,008 users who said they were taking one of the medications.
People do not generally describe symptoms in medical terminology. The team used large language models to identify self-reported medicine use and side effects, then map informal descriptions to MedDRA, a standardized vocabulary for classifying medical conditions and adverse events. That let the researchers analyze posts as comparable symptom categories rather than isolated anecdotes.
Posts mentioning semaglutide or tirzepatide covered May 2019 through June 2025.
This share applies to identified Reddit users, not all people taking the medications.
Gastrointestinal symptoms were the predominant category among users who described side effects, matching established concerns around these medications. Fatigue was the second-most frequently reported complaint, appearing in 16.7% of the study’s symptom analysis, although the authors said it reached reporting thresholds in relatively few clinical trials they reviewed.
The findings are hypothesis-generating, not a safety verdict. Voluntary posts cannot establish how often symptoms occur or whether medication use caused them; symptoms could also reflect weight loss, other health conditions or unrelated factors. The posts also did not consistently provide treatment details needed to test timing or dose relationships.
Reddit users also are not representative of all GLP-1 users; the authors note that they tend to be younger, more likely to be male and disproportionately based in the United States. The team plans to examine other platforms and languages. Whether the same patterns appear in more representative data is the unresolved step before these online signals can guide stronger conclusions.
Loading discussion...
Join the conversation
Explain who should validate the signal first.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.