AI review of 400,000 posts on Reddit identified recurring accounts of unexpected symptoms among people taking GLP‑1 receptor agonists, including Ozempic, Wegovy, Mounjaro and Zepbound. The symptoms mentioned with notable frequency included menstrual changes, chills, hot flashes and pronounced fatigue, patterns that emerged from automated text analysis rather than clinical records.
The study used natural language processing to scan large volumes of patient-reported content and surface common themes. Because the material is self-reported and drawn from public discussion forums, it cannot establish that the medicines caused the symptoms. Reporting biases, overlapping health conditions and concurrent treatments can all influence what appears online; researchers therefore treat these findings as potential signals rather than confirmed side effects.
Despite those limitations, the patterns identified carry practical relevance for pharmacovigilance. Social media mining can reveal clusters of experiences that formal trials and spontaneous reporting systems might miss, especially for effects that are subjective or intermittently reported. Detecting consistent mentions across hundreds of thousands of posts suggests topics that merit targeted epidemiological study and closer examination in controlled settings.
The discovery underscores the growing role of digital surveillance in drug safety monitoring and the need for health professionals and regulators to consider diverse data sources when assessing adverse events. Further work is required to quantify any risks and to determine whether the signals identified reflect drug-related effects, underlying conditions, or reporting artefacts. In the meantime, clinicians are advised to continue relying on established safety data and to discuss unexpected symptoms with patients using these therapies.
