Weight loss and diabetes medications such as semaglutide and tirzepatide have changed the lives of many people struggling with obesity and blood sugar problems.
These drugs can help patients lose significant weight, reduce appetite, and improve diabetes control. Their popularity has grown rapidly in recent years, with millions of people now using them worldwide.
The medications belong to a group known as GLP-1 receptor agonists. They mimic natural hormones in the body that help regulate hunger, digestion, and insulin release. Because of their effectiveness, demand for these drugs has increased dramatically.
But as more people begin taking these medications, scientists are trying to better understand all of their possible side effects.
Researchers at the University of Pennsylvania recently explored a new way to study this problem using artificial intelligence and social media.
Their findings, published in Nature Health, came from analyzing more than 400,000 Reddit posts written by nearly 70,000 users over a period of more than five years.
The researchers used AI systems to search through discussions where people talked about their experiences using semaglutide and tirzepatide. They wanted to identify symptoms users commonly mentioned while taking the medications.
Some of the reported side effects matched what doctors already know.
Nausea, stomach discomfort, and digestive problems were frequently discussed online. These symptoms are already listed in official medical information for GLP-1 medications and are considered common side effects.
However, researchers also discovered several complaints that may not yet be fully reflected in current clinical trial data or drug labeling.
One of the most surprising findings involved reproductive symptoms.
Nearly 4% of users who reported side effects also mentioned menstrual changes. These included irregular periods, bleeding between cycles, and unusually heavy bleeding.
Researchers also noticed repeated discussions about body temperature changes. Some users described chills, feeling unusually cold, hot flashes, or fever-like sensations.
Fatigue was another major concern discussed by many users online. According to the researchers, tiredness ranked as the second most common symptom mentioned in Reddit discussions, despite receiving less attention in some clinical studies.
The researchers emphasized that the study does not prove these medications directly caused the symptoms. Instead, the findings identify patterns that scientists believe deserve closer investigation.
Sharath Chandra Guntuku, one of the senior researchers involved in the study, explained that the method appears capable of identifying real signals because it successfully detected already-known side effects such as nausea.
The scientists say the newer symptoms should now be studied more carefully in future clinical research.
The study also demonstrates how artificial intelligence is changing medical science.
In the past, analyzing large numbers of online discussions was extremely difficult because people describe symptoms using many different words and expressions.
One person may say they feel “freezing all the time,” while another may describe “cold flashes” or “body chills.” Older computer systems struggled to organize this type of information consistently.
Modern large language models, including systems similar to GPT and Gemini, can now process huge amounts of text and connect everyday language to standardized medical terms much more efficiently.
Researchers say this makes it possible to analyze public health conversations on a much larger scale than before.
Social media may provide valuable information because patients often discuss experiences online that they never report during medical appointments.
Professor Lyle Ungar, another author of the study, compared online patient communities to neighborhood conversations where people exchange personal experiences in real time.
The researchers believe these online discussions may help scientists identify possible concerns much earlier than traditional reporting systems.
The study found that about 44% of Reddit users mentioned at least one side effect from GLP-1 medications.
Researchers believe some of the unexpected symptoms could potentially relate to how the drugs affect the hypothalamus, a region of the brain involved in controlling hormones, body temperature, appetite, and metabolism.
Still, the researchers repeatedly stressed the importance of caution.
Reddit users do not represent the general population. The platform tends to attract younger users, more male users, and many people from the United States.
People posting online may also be more likely to discuss unusual or negative experiences.
Because of this, the findings cannot replace carefully controlled clinical trials.
The researchers say their study should instead be viewed as a way to generate new research questions and identify possible warning signs more quickly.
The team now hopes to expand its work beyond Reddit by studying other social media platforms and non-English-speaking communities.
They believe AI-assisted social media analysis may become especially useful for tracking fast-growing health trends, new medications, and products sold in less-regulated markets.
As more health conversations move online, researchers believe social media may provide some of the earliest clues about emerging medical issues.
Overall, the study highlights both the promise and the limitations of using artificial intelligence to monitor medication safety. The findings suggest that social media discussions may reveal symptoms patients care deeply about, even if those symptoms are not yet fully documented in clinical research.
The reports of menstrual changes, temperature-related symptoms, and fatigue may deserve further scientific investigation. However, the research cannot prove direct cause and effect, and online discussions can sometimes be incomplete or biased.
Larger clinical studies will still be necessary to confirm whether these symptoms are truly connected to GLP-1 drugs. Even so, the study demonstrates how AI may help scientists monitor real-world patient experiences much faster than traditional methods alone.
