CommunitySide Effects

AI scans 400,000 Reddit posts to flag overlooked GLP-1 side effects

20 replies19 peopleApr 10, 2026☆ Follow
Summary

What do Reddit users think about using AI to mine forums for unreported GLP-1 side effects like menstrual changes?

Participants see value in the approach for spotting signals that controlled trials might miss, but stress that anecdotes are uncontrolled, self-selected, and cannot prove causation. Many note that rapid weight loss, PCOS, perimenopause, or other conditions could explain cycle shifts, and that only people with problems tend to post. Several share personal experiences of irregular cycles on these drugs while others report no change or improvement.

What this discussion establishesWhere people disagree

Whether the 4% menstrual irregularity rate is likely inflated by selection bias or still worth investigating as a real signal.

Still open

Whether any menstrual changes are caused by the medications, by weight loss, or by unrelated factors such as PCOS or menopause.

Nothing here is advice.

20 replies · 19 people
PlainPebble75archiveopening postApr 10

I appreciate them pursuing projects of this sort and noting the constraints promptly. It would be good if they extended this approach to places like these discussions and additional compounds of that type. The authors warn that the outcomes don't establish direct links. One can't conclude these medications produce the reported effects, according to the lead researcher who is a grad student. Still, close to four percent of the platform participants mentioned changes in their periods, and that rate could rise when focusing solely on women. It looks like an indication that deserves more attention.

SteadyCinder62archiveApr 10

It catches my eye, yet this lacks the structure of a controlled trial or item. The aim is obvious, with straightforward metrics that could be included to observe during subsequent phased testing.

RustCompass30archiveApr 10
↳ replying to @SteadyCinder62

You've got a point that those random personal accounts lack the rigor of proper experiments. Yet combining them all lets experts examine the details for clues they might have overlooked otherwise.

GreenMeadow54archiveApr 10

This seems like a solid application of the tech. There's an actual pattern worth checking into, though plenty of irrelevant stuff gets mixed in. The trouble here is that folks experiencing problems are naturally the ones who speak up. Nobody bothers to mention online when their routine stayed exactly the same after beginning these treatments. That means the reported rate is probably way overstated.

LevelPebble88archiveApr 10
↳ replying to @GreenMeadow54

Which set of people does the four percent refer to exactly? It can't be everyone who uses the service. The text there makes it clear otherwise.

LevelFenwick49archiveApr 10

This scenario catches my attention. Yet there's plenty of clutter that needs filtering. In particular around the shifts in monthly cycles. It raises questions about whether some of the people involved might have had undiagnosed issues such as polycystic ovarian syndrome along with other fertility-related disorders. Then comes the task of sorting out whether the drug or those prior problems caused the effects.

GreenMeadow27archiveApr 10

Just seeing two things happen together never proves one actually leads to the other. Plenty of these females already dealt with conditions such as PCOS before anything else. When those problems finally get addressed through dropping pounds or leveling out glucose and similar steps their cycles often settle back into a regular pattern including cases where menstruation had been missing for years. Sorting the whole picture will require a careful long term investigation.

RustCompass30archiveApr 10
↳ replying to @GreenMeadow27

Even so this adds value to the investigation since controlled experiments can clarify those elements first and then separate out the remaining influences to check whether the medication itself caused the improvement or whether things like reduced body mass or glucose levels were responsible instead just as you noted.

GreenMeadow27archiveApr 10
↳ replying to @RustCompass30

I never claimed this had no value at all. The point here is simply to note that two things happening together does not prove one leads to the other. In short, avoid overreacting before further research provides clarity.

SteadyTimber49archiveApr 10

I find this quite fascinating from my own experience. I've noticed irregular periods and issues with keeping my body temperature steady after starting that medication, so I'd like to see more actual research on it.

AmberSignal57archiveApr 10

I figure the fast drop in body mass overall is what triggers that effect, or at least plays some part in it. Bone strength loss works the same way.

WarmWillow50archiveApr 11

I suppose this offers a method to spot warning signs via the large number of users exceeding trial sizes by quite a bit. Even if based on reports these signs get noticed without legal responsibility before official approval happens. It works as a budget friendly approach to observe ahead of wide distribution and the lawsuits that will follow.

NorthKettle44archiveApr 11

For a fast look I copied over the summary of those stats from the piece.

SlowSparrow89archiveApr 11

Wow the distortion could turn wild once AI gulps every post from this spot.

ClearCompass37archiveMay 18

For a while I've been experiencing significant bleeding outside of my regular periods along with unpredictable menstrual timing. At first I assumed this stemmed from my birth control pills, yet I connected the dots that everything began right after initiating that medication. My initial injection happened to occur while I was already menstruating. No one has ever told me I have polycystic ovary syndrome, uterine growths, or similar conditions, and my periods were always normal before this point. Quick shedding of pounds isn't the culprit here either. I began with a minimal amount and increased it gradually over time.

AmberKettle85archiveMay 18

This seems really fascinating to me, particularly around periods. From stories I've read these meds looked like they eased minor hormone problems for some yet others could notice the reverse instead. It makes me wonder what kind of effect that might have on later studies if any.

SteadyQuill55archiveMay 18

My menstrual periods started coming at odd intervals beginning last summer. I figured this could be the onset of menopause and it might still turn out that way, though it happened right when I began that medication. This doesn't worry me at all. It seems odd how folks expect a lady to get anxious if her bleeding doesn't follow a strict timetable, yet I failed to observe the shift in my own pattern. A coworker pointed it out to me, I acknowledged it seemed strange, and promptly forgot about the matter again.

ClearMeadow73archiveMay 18

I've taken these meds sporadically across the last three years. My monthly cycle stayed steady the entire time even though I stopped the pill about four weeks back. Should be curious to watch whether any differences show up.

WarmMeadow92archiveMay 18

At my workplace the library only carries recent volumes of that publication. Still the researchers uploaded their full manuscript to a public site. It covers how online forums capture user reactions to these medications outside formal studies. They examined hundreds of thousands of discussion threads involving the two compounds and noted that thousands of people mentioned taking them. A button appears in the upper section to retrieve the document. The summary lines up with the earlier thread post so this must be the matching study.

WarmHarbour28archiveMay 19

What the fuck is Reddit?

SteadyLedger61archiveMay 22

Probably the quick loss affects things because fat around the middle is important for creating and balancing hormones, messing up the ones related to reproduction. I wonder what portion of people noticing shifts in their periods were actually approaching or in menopause, since that brings its own set of hormone problems.

Add your experience

If you have tried it, this topic is still open. Sign in to reply — it takes a minute.

AI scans 400,000 Reddit posts to flag overlooked GLP-1 side effects · ZyraTrack Community