[istultus]

Exactly, anecdotes, with a healthy portion of the Availability and Confirmation Heuristics.

We already know that women are more susceptible to auto-immune responses and that pregnancy is one of the reasons for that disparity between the sexes (carrying a foreign invader for 9 months plays with the expression of immunity genes in a manner that occasionally - rarely, at a population level - backfires.) And we know that the risk for an autoimmune "malfunction" rises with age. Thus there will obviously be random clusters of people, with more women who've given birth and and average age that skews older than, say, 30 - to which you unfortunately seem to belong - where there are more autoimmune responses than the average population.

What's funny is that in response to this post there are multiple posts each suggesting their own conspiracy. I wonder if this actually points to something, or it too is a just random noise.

[dillydogg]

When I was getting my immunology PhD, the prevailing thought about the prevalence of autoimmune disease in women was that estrogen promoted B cell survival, leading to more anti-self B cells from being removed.

[Xcelerate]

> average age that skews older than, say, 30 - to which you unfortunately seem to belong

Haha, unfortunately you would be correct.

> in response to this post there are multiple posts each suggesting their own conspiracy. I wonder if this actually points to something, or it too is a just random noise.

After I posted my original comment full of personal anecdotes, I thought “It’s always my useless posts that get the most responses.” Checking HN later in the day confirmed my suspicion.

As a data scientist, there are certain fields of study concerning the physical world that overwhelm me with the vast number of baseless and likely wrong theories within them: psychology, sociological, and medicine. Basically any field that attempts to study the human body or mind.

We likely have sufficiently advanced ML and inference techniques by this point to discover the true root causes of allergies and autoimmune disease (and all disease for that matter), but unfortunately this is a case where the problem is the data, not the algorithm.

We have no reasonable system in place to accurately collect, process, and store vast amounts of observational health data. And beside the data quality concerns, the privacy implications of attempting to store all of that sensitive data securely are mind-boggling.

Nevertheless, with observations from billions of people, it seems like even simple techniques could extract a lot of signal from the noise, which is just too difficult with the tiny number of low sample, high variance RCT studies we currently use. When you’re trying to predict causation from one or a few variables, it’s simple. When you’re considering thousands of variables at once, it’s next to impossible.

[mlyle]

> it seems like even simple techniques could extract a lot of signal from the noise, which is just too difficult with the tiny number of low sample, high variance RCT studies we currently use.

On the other hand, RCTs gather relatively well-conditioned data. And we have enough study power across them that we end up with a lot of statistically significant, real findings.... that still have effect sizes that aren't clinically meaningful.

We even know about things with large effect sizes that are controlled by a few variables, that are rare enough that it's hard to find the patients that they apply to.

And, of course, correlation ain't causation: if we used your system and found a bunch of things, we'd still need to figure out how to reduce them down enough to be something we could test before advising practitioners to do things a certain way.