AI Identifies Unexpected Symptoms from Unapproved Weight-Loss Drug Retatrutide

A University of Pennsylvania study, published in JAMA Network Open on October 14, 2026, utilized artificial intelligence (AI) to analyze nearly 150,000 Reddit posts concerning the unapproved weight-loss drug retatrutide, uncovering unexpected user-reported symptoms including increased appetite, sleep disturbances, and elevated heart rates[1][2][8]. Increased appetite was the most frequently reported symptom among users with classifiable symptoms, observed in 17.1% of such users[2][8][10].

This research offers a new method for identifying potential drug safety signals from real-world social media data, particularly for substances acquired outside regulated channels[1][9]. The ability to detect symptoms not initially flagged in controlled clinical settings is critical for public health, as unapproved products lack verifiable safety or dosing[1][5].

What the records show

The University of Pennsylvania study, published in JAMA Network Open, used a large language model to analyze 148,640 Reddit posts and comments related to retatrutide[2][3][9][10]. Researchers screened approximately 40,000 Reddit users and identified about 14,000 who stated they had taken unapproved retatrutide[9]. From this group, 13,589 users reported current use, and 7,823 had at least one mapped symptom after excluding weight-loss and appetite-suppression terms[3][10].

Among users whose posts contained classifiable potential symptoms, increased appetite was the most frequently reported, affecting 17.1%[2][8][9][10]. Other unexpected symptoms included sleep disturbances, such as insomnia and abnormal dreams, and elevated heart rates[1][2][8]. These findings diverge from the Phase 2 clinical trials for retatrutide, where gastrointestinal complaints were the most common side effects[8][10].

The timeline

Reddit posts analyzed by the study spanned from May 2021 to December 2025[2][10]. The University of Pennsylvania study was published on October 14, 2026[1][2][3]. Retatrutide, an investigational drug developed by Eli Lilly, is not yet approved by the U.S. Food and Drug Administration (FDA)[2][4][5][6]. Eli Lilly has stated plans to submit a Biologics License Application (BLA) for retatrutide to the FDA in the first quarter of 2027[6].

What changed and for whom

This research highlights a novel application of AI in pharmacovigilance, the monitoring of drug safety, by identifying potential drug safety signals from social media data[1][9]. Users obtaining investigational weight-loss drugs online are outside conventional drug-safety monitoring, creating a surveillance gap that AI can help address[9][11]. For individuals considering or using unapproved versions of retatrutide, these findings suggest potential side effects beyond those identified in official trials[1][8].

What the subject has said publicly

Eli Lilly is developing retatrutide, a triple agonist for GLP-1, GIP, and glucagon, as an investigational weekly injectable peptide[2][5]. As of October 2026, retatrutide has not received FDA approval or approval from any other regulatory body globally[2][5][6][7]. Products promoted as retatrutide online often have unverified ingredients, purity, and quality, posing health risks[5][9]. Jena Shaw Tronieri, Senior Research Investigator at Penn's Center for Weight and Eating Disorders and a co-author of the study, noted that because retatrutide works differently from existing GLP-1 drugs, it is possible its effects on the body are also different[8]. Researchers, including lead author Neil Sehgal, emphasize that the findings are hypothesis-generating signals and not conclusive proof of causation, given the challenges in verifying the purity or composition of unapproved products[1][3][9][10]. The analysis cannot assess incidence, causality, or comparative safety[10].

What we could not confirm

The study could not verify the identity, purity, dose, route, or duration of use for the retatrutide products discussed on Reddit[10]. It also could not definitively establish that retatrutide caused the reported symptoms or that the experiences of Reddit users reflect those of clinical trial participants[8][9]. The analysis cannot determine the incidence of these symptoms or assess comparative safety against approved drugs or clinical trial data[10].