Hacker News Reader

Comment by danieltillett | Hacker News Reader

I have to agree with you here - the separation is too good. You just never get anything in the real world that works this well with complex diseases.

View on HN

This is true, but everything is complex only up to the point when you begin to understand it. Scurvy was a complex disease until the discovery that it was cured and prevented by citrus fruit(1). Now we have the technology to skip right to the discovery of vitamin C, a miraculous 100% accurate test for scurvy.

I know I'm oversimplifying, but I think the point is valid. This is why discovery of a biomarker is so crucial -- it begins to "decomplexify" the disease so we can properly define it and start to find treatments for it.

(1) Ok, it's never really that simple: http://mentalfloss.com/article/24149/how-scurvy-was-cured-th...

Maybe link to the original blog article [1], your link is just a summary.

[1] http://idlewords.com/2010/03/scott_and_scurvy.htm

Thanks -- I just copied the first link I found in a search engine.

It is suspicious but look at the scatterplots: https://www.pnas.org/content/pnas/early/2019/04/24/190127411... It really is pretty much perfect separation on their metric. When the effect size is that large...

One thing that would be interesting to investigate in the future would be to see how the test behaves for CFS patients who are not as severely ill, or for non-CFS patients that are currently ill from a different disease.

Exacly, just sorting it would show it better, it feels like it should just output percentage with (arbitrary/based on population %age/whatever) threshold that is outside of the test.

AIDS was once considered a very complex disease, but it has a very accurate (HIV) test today. My big concern would be if, like cancer, there are really many diseases and Chronic Fatigue symptoms are common, not differentiating.

That could still be very interesting. Eg. There are some other illnesses like MS where patients can get fatigued in a similar way to CFS, but not always. If we find a similar response between the fatigued and non fatigued MS patients for example, that could still point to something that is going on causing the fatigue in both cases. Ron Davis’s line on this was to point out having one disease doesn’t mean you can’t have any others.

There are a lot of other interesting overlaps showing up. One of the more interesting to me (because I have it) is that about 40% of CFS patients are testing positive for small fiber polyneuropathy on punch biopsies that appear to be autoimmune in origin. Is that itself an underlying cause of some 40% of CFS cases or is it just a common comorbid condition that whatever triggers CFS also triggers?

I've done a lot of reading on CFS and fibromyalgia, and this has been my opinion too.

For starters, these are both diagnosis of exclusion; these labels are also used by lazy doctors who lack a deeper understanding of their field (anecdotally, both personal and otherwise), or who are reluctant to run expensive tests for rarer conditions; and many people with these conditions have been found to have small fiber neuropathy.

AIDS, like CFS, was never considered a disease in the medical sense. They both are called “syndrome”, which is medical short-hand for “interesting enough to study further, but frankly, we don’t know shit about what’s causing this”.

https://en.wikipedia.org/wiki/Syndrome: ”A syndrome is a set of medical signs and symptoms that are correlated with each other and, often, with a particular disease or disorder.”

That’s why AIDS became “HIV infection” when medical science discovered what caused AIDS.

If this research gets confirmed, and a mechanism is found, that will happen here, too.

However, as you say, if that happens, there’s a good chance there are people currently diagnosed with CFS who don’t show this biomarker.

If so, those people run the risk of getting stigmatized even more than current CFS sufferers do.

The real solution to that problem, I think, would be for the world to accept that, if you have a mental problem, a) that doesn’t imply you put it there, and b) it doesn’t imply you should be able to get it out without outside help.

Only a few centuries ago, people were being blamed of being “unfaithful to god” when they got sick, be it due to a bacterial infection or due to mental illness. I see blaming people for their mental problems as a remnant of that.

On the other hand, there are times when blaming people for their misfortune is appropriate, and serves a social purpose. The hard part is telling when it is appropriate. It does seem pretty clear though that we tend to both give ourselves too much credit for our successes, and too much blame for our failures, overall.

My impression is that people give themselves the credit for their successes, and blame somebody else for their failures.

My successes are due to my moral worth, and others' failures are due to their moral failings.

Who care is the separation is "too good." It's a lead. This is what science is. You find leads and followup on them. Lots of dead ends in the process.

You should care if it “too good” because it could be a sign of overfitting or tampering with data during analysis to reach a desired conclusion. Replication should shed some light here.

Of course skepticism is warranted however an important distinction to make is that they are measuring a whole cell phenotype, rather than some surrogate such as gene expression or a circulating molecule. I work on cancer biomarkers, and phenotypic readouts work much better than you might expect.

Reply on news.ycombinator.com