AI analysis of Reddit posts reveals potential unreported side effects of weight-loss drugs
Using artificial intelligence, researchers examined over 400,000 Reddit posts from nearly 70,000 users of GLP-1 drugs like Ozempic and Mounjaro. They identified reports of menstrual cycle changes, chills, hot flashes, and fatigue that are not commonly listed in clinical trial data. The findings are not proof of causation but suggest areas for further investigation.
The University of Pennsylvania team analyzed over five years of Reddit data, drawing from roughly 70,000 users who discussed semaglutide and tirzepatide medications. Their AI approach successfully detected known side effects like nausea, which validated the methodology, while also surfacing less documented reports involving menstrual irregularities and temperature-related discomfort. Nearly 4% of sampled users mentioned menstrual changes.
The research builds on earlier efforts dating to 2011, when Ungar participated in one of the first attempts to mine online user content for drug safety signals. The authors emphasize that social media captures spontaneous patient conversations that may never reach formal reporting channels, though they caution that such data cannot establish causation.
This research could influence how clinicians discuss GLP-1 medications with patients, potentially prompting more targeted questions about menstrual changes or temperature discomfort during follow-up visits. It may also encourage regulators and pharmaceutical companies to consider social media mining as a supplementary tool for post-market surveillance. However, the findings could generate unnecessary concern if patients interpret associations as proven causation, underscoring the need for careful communication about what such data can and cannot demonstrate.