AI analysis of Reddit posts uncovered potential side effects of GLP-1 drugs that may be underreported, demonstrating the value of social media in early drug safety insights.
Researchers at the University of Pennsylvania used artificial intelligence to examine more than 400,000 Reddit posts and identify symptoms reported by patients taking GLP-1 drugs, including semaglutide and tirzepatide. These medications are widely used for weight loss and diabetes, but some of the effects described online may not be fully documented in clinical trials or official reports.
The study, published in Nature Health, analyzed over five years of discussions from nearly 70,000 Reddit users. It points to two groups of symptoms that deserve closer attention: reproductive issues such as irregular menstrual cycles and temperature-related effects like chills and hot flashes.
“Some of the side effects we found, like nausea, are well known, and that shows that the method is picking up a real signal,” says Sharath Chandra Guntuku, Research Associate Professor in Computer and Information Science (CIS) at Penn Engineering and the study’s senior author. “The underreported symptoms are leads that came from patients themselves, unprompted, and clinicians could potentially pay attention to them.”
“Clinical trials generally identify the most dangerous side effects of drugs,” adds Lyle Ungar, Professor in CIS and a co-author on the study. “But they can fail to find what symptoms patients are most concerned about; even though social media is not necessarily representative, a large collection of posts may reflect additional concerns.”
Limits of Clinical Trials and Value of Patient Voices
The researchers emphasize that their findings do not prove cause and effect. “We can’t say that GLP-1s are actually causing these symptoms,” notes Neil Sehgal, the study’s first author and a doctoral student in CIS advised by Guntuku and Ungar. “But nearly 4% of the Reddit users in our sample reported menstrual irregularities, which would be even higher in a female-only sample. We think that’s a signal worth investigating.”
Ungar has studied online health discussions for years, including early efforts in 2011 to extract information about drug side effects from user-generated content. He compares these platforms to informal networks where people share experiences in real time.
“Online patient communities work a lot like a neighborhood grapevine,” says Ungar. “People who are living with these medications are swapping notes with each other in real time, sharing experiences that rarely make it into a doctor’s office visit or an official report.”
As social media use has expanded, it has become a more valuable source of health-related data, even as access to that data has grown more restricted. Researchers note that traditional clinical trials remain essential, but they can take years to complete. (Guntuku has also conducted research on methods for adjusting to shifts in platform access.)
Computational Social Listening and Scaling Challenges
“Clinical trials are the gold standard, but by design, they are slow,” says Guntuku. “This is not a replacement for trials, but it can move much faster, and that speed matters when a drug goes from niche to mainstream almost overnight.”
A major challenge in this process, which Guntuku calls “computational social listening,” has been scale. People describe symptoms in many ways, making it difficult to match their language to standardized medical terms used in systems like the Medical Dictionary for Regulatory Activities (MedDRA).
Recent advances in large language models such as GPT and Gemini have made it possible to process and organize these posts more efficiently. “Large language models have made it possible to do this kind of analysis much faster with a level of standardization that could be difficult to achieve before,” says Sehgal.
Key Findings: Reproductive and Temperature Symptoms
Although Reddit users are not representative of the general population, the reported symptoms largely align with known side effects of these medications. About 44% of users mentioned at least one issue, most often gastrointestinal problems.
However, some patterns stood out. Nearly 4% of users who described side effects reported reproductive issues, including irregular cycles, heavy bleeding, and bleeding between periods. Others mentioned temperature-related symptoms such as chills, feeling unusually cold, hot flashes, and fever-like sensations.
Fatigue was also frequently reported, ranking as the second most common complaint, even though it appears less often in clinical trial data.
Biological Explanations and Clinical Implications
“These drugs are thought to work by engaging part of the brain called the hypothalamus, which helps regulate a wide variety of hormones,” says Jena Shaw Tronieri, Senior Research Investigator at Penn’s Center for Weight and Eating Disorders and a co-author of the study. “That doesn’t mean the medications are necessarily causing these symptoms, but it could suggest that reports of menstrual changes and body temperature fluctuations are worth studying more systematically.”
In the short term, the team hopes their findings will prompt healthcare providers and researchers to pay closer attention to patient discussions online. “They’re clearly on patients’ minds, and that’s worth paying attention to,” says Sehgal.
Future research will expand beyond Reddit and include non-English communities to determine whether similar trends appear in other populations. “We don’t really know yet whether what we’re seeing on Reddit reflects the experience of GLP-1 users globally, or whether it’s particular to the kind of person who posts on Reddit in the United States,” Ungar says.
Future of AI in Drug Safety Monitoring
The researchers believe this type of AI-driven analysis could become an important tool for identifying early signals related to new medications and health trends. This approach may be especially useful for substances that gain rapid popularity online, including products sold in less regulated markets such as injectable peptides.
“The whole point of this kind of approach is that it can move quickly, and that’s exactly when it’s most valuable,” says Guntuku.
Reference: “Self-reported side effects of semaglutide and tirzepatide in online communities” by Neil K. R. Sehgal, Jena Shaw Tronieri, Lyle Ungar and Sharath Chandra Guntuku, 10 April 2026, Nature Health.
DOI: 10.1038/s44360-026-00108-y
The authors report no outside funding. Tronieri reports receiving an investigator-initiated grant, on behalf of the University of Pennsylvania, from Novo Nordisk and receiving consulting fees from Currax Pharmaceuticals, LLC. The other authors report no conflicts of interest.
Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.
8 Comments
Whay ? “Guntuku and Ungar. “But nearly 4% of the Reddit users in our sample reported menstrual irregularities, which would be even higher in a female-only sample. We think that’s a signal worth investigating.”
2 more idiots that don’t know the difference between a man and a woman.Are you kidding me? What a pair of dumb asses
You’re literally the only dumbass who doesn’t see how removing males from the total responses would impact the percentage. If a group of 100 is split between men and women, and 4% of the total group mentions this side effect, then removing the men would raise the percentage mentioning it to 8%.
Er, um, sorry, Robin your mistake was to ASSUME that Moose did not know that the percentage would increase when removing males from the calculate. This apparently enabled you to feel superior to Moose by being able to multiply by 2. It should not have…
I am serious scientist who has analyzed with like this for 3 decades, but this is such a simple little logical thing, my experience is actually not necessary…
Let’s give Moose a little credit (being able to multiply) and consider that they may be referring to the ILLOGIC of the very CALCULATION itself: including males in a calculation of menstruation data.
Because on quite obvious first principles this calculation MUST be wrong, the result is by definition a bogus number (4%),
Since it must be wrong, so obvious why only a dumbass, one would make the calculation.
UNLESS of course, which is probably the case, the authors of the study – because it is REDDIT data and they cannot contact the people who post – ACTUALLY DO NOT KNOW exactly who is male and who is female, in which case they CANNOT calculate male and female specific numbers from this data source, and thus are forced leave the results sex non-specific.
Moose: your drive to feel superior blinded you to the possibility that the study authors dont have accurate identity/counts for male and female, and therefore are NOT dumbass. You assumed that the authors are blazingly incompetent, and you have no evidence to support that and from what I see, substantial evidence against it.
Robin your drive to feel superior blinded you to the possibility that Moose DOES know that removing males will increase the percentage (of anything that is female specific). Granted, Moose might not, but I bet that, since it is SO SIMPLE, Moose does know this. But either way, assumption has no evidence to support it.
Please note that I do not feel superior to either of you. Unlike Moose and Robin) my purpose is to try and get people to be a little more charitable by BEFORE attacking, THINKING a bit more about the information they see, about what it says and does not necessarily say, and that there are often DIFFERENT explanations for what people say. Cheers.
I suffered a catastrofic injury to my abdomen. they wanted me to lose 40lbs so there would be viable tissue not fat to sew my abdominal muscles back together months of taking ozempic did the trick. However, it also caused stomach paralysis. Now I have to take medication in order to eat and now my body is rejecting all food despite the medication . Now I have another side effect of the Ozempic, low blood sugar. my inability to eat is making it hard for me to keep my blood sugar above 70. and now I’m borderline anorexic. All I’m saying is think long and hard before you inject that poison, because Those side effects are PERMANENT!
If it’s to good to be true……….it’s nit
Yes, Wegovy and Ozempic can cause nausea, but stimulating the Vegas nerve alleviates it. Its an unfair practice that Navitus and other pharma insurance agencies will only approve semiglutides. Zepbound works way better but we are kept from it.
I find it interesting, nocne mentioned hairloss! I know I am not the only struggling with this side effect. I had to dig deep to find it.
It’s the opposite for me, my hair grew back on Ozempic, after falling out from taking Jardiance. Usually losing hair is from malnutrition, low iron and low zinc, and too fast weight loss. I had none of these, therefore it was probably the jardiance. Stopped it, took nothing for 6 months. Hair stopped falling. Went on Ozempic 6 months ago, great book sugar control, modest weight loss, and a full head of nice hair again. Gotta eat lots of protein and fats for good hair.