Hacker News Reader

Comment by jbandela1 | Hacker News Reader

> But Joy's superpower is so unusual that researchers all over the world have started working with her and have discovered that she can identify several kinds of illnesses — tuberculosis, Alzheimer's disease, cancer and diabetes.

The story for diagnosing Parkinson’s sounded plausible until this sentence. With Parkinson’s, you could imagine that she had some sort of sensitivity in her smell to a certain biomarker.

Now with a plethora of vastly different diseases (even cancer is really a myriad of diseases grouped together), the suspicion of a confounder goes way up.

Perhaps instead of diagnosing Parkinson’s, she is actually sensing some signal that indicates inflammation or some other distress signal.

Or else people, prior to the manifestations of these diseases tend to make subtle, unconscious changes to their hygiene.

View on HN

That is not how i read that sentence.

It sounds like you are assuming that every one of those illneses smell the same to her. That is that she can tell that someone has Parkinson’s or tuberculosis, or Alzheimer's, or cancer or diabetes, but she can’t tell which one they have.

The way i read it is that she can identify which one people have based on how they smell. Totaly made up example: tuberculosis smells peppery, while Alzheimer minty, and so on and so on. (Admittedly this also assumes that the journalist was sloppy about cancer. Probably she was only tested on specific types of cancers. It is very unlikely that all cancers would smell the same. But this is something which is very easy to get jumbled up by the journalist.)

Ultimately it doesn’t matter. If it verifies through a properly designed test protocol they will publish a paper about it. If not, they won’t. So if it matters we will hear about it.

If you read the report OP linked, it says they did a blind test where she was given shirts that had been worn by a variety of people some of who had Parkinsons for differing length of time, and some who had not, and she was able to get them all, including what they though was a false positive until that person came back months later to say they had developed it (similar to how she's been able to smell it on her husband before he was symptomatic).

Why should she not be able to detect some kinds of cancers? Some dogs can.

In the end, all this should not be so difficult. Take the head space, then do MS of a bunch of people with these diseases and the head space and MS of healthy people as a comparison. The delta gives the disease.

I'm not sure any of today's mass spectrometers yet can match human smell in sensitivity, at least to some scents with larger more complex molecules.

"with larger more complex molecules."

What are "larger, more complex, molecules"? Without trolling, I think scent cuts off at a MW 330 or something.

But it's not as good with distinguishing complex organic molecules one from another, no? I.e. between various proteins of a similar size.

Mass spectrometry gives results that are a bit like the ingredient lists for food. Tells you precious little about the actual product.

You don't smell proteins.

Mass specs can detect down to the ten thousands? IIRC or so molecules.

But I don't think they can do that at the same time as having trillions of some other uninteresting molecule like water in them.

It's basically a dynamic range problem.

Mass specs would have this problem, yes. That's why they are usually paired with chromatography of some sort (usually GC/MS or LC/MS). The GC or LC more or less reduce a mixture into components, which are fed into the mass spectrometer to analyze what those components are.

Yep the Ms that does this is tethered to chromatography first (sometimes even 2 or 3 different separation schemes in tandem)

I bet most (type 1) diabetics could smell other diabetics if they were trying. High blood sugar especially changes your body odor a ton. You sweat a lot more sugar for your skin bacteria to digest.

Not to discount the woman or the article, we should totally be researching this stuff

Anybody can learn pretty quick to smell diabetics. Low blood sugar leaves one's breath bitter and metallic, high blood sugar causes fruity, sweet breath. It's extremely obvious if you're at all close to them.

Wow, til.

Is there anything they can do about the breath? (ditto people on keto diet)

On a ketogenic[diet] ... your body uses fat instead of carbohydrates for energy. This shift causes a major spike in ketones, the source of acetone breath. ... Sweet-smelling breath may be a short-term side effect of this type of diet.[0]

[0] https://www.webmd.com/diabetes/acetone-breath-diabetes

So low blood sugar causes sweet breath.

Ketones are the "metallic" smell. I wouldn't describe it as sweet.

It's literally the homeless alcoholic smell.

They used to diagnose it by tasting urine.

I read the article as saying she can tell them apart, not merely sorting them into healthy/sick.

And I wouldn't be surprised if things like "cancer" were actually a confounder (I'm thinking along the lines of detecting the body's reaction rather than the tumor itself)--but it's still useful information. They've already used her information to find a albeit imperfect test for Parkinsons. What if you had a similar test for cancer? It would tell the doctor to start looking in fashions they would not do for the general population.

If the goal is early diagnosis and the separate detections aren't getting confounded, why does it matter what parallel chain of causation leads to Parkinson in the patient and a correlating reading in the nose or other test?

If no one in a normal state has the same mix of fear, anger and confusion and this leads to a microbial change around sweat glands, then that is a valid test of greater accuracy than many existing medical tests.

In the tests, she's presumably smelling people known to have the disease or not, and maybe she can pick up on that knowledge through subtle social cues.

Impressive in itself, but that's useless for diagnosing people who don't know they have a disease, which would be the medical breakthrough.

No presuming necessary. In the article, it describes a test involving smelling t-shirts in boxes.

This would be totally necessary for a robust conclusion-trust me, the social cues of having PD are not subtle ;)

She was able to detect a person who they thought was a false positive and turned out several months later did have Parkinson's. She detected her husbands before he had signs as well.

They also found the molecule she was smelling which was expressed with sebum. This isn't entirely unfounded as they already knew about dogs smelling cancers and other various diseases.

But what if, for example, she's just sensitive to the smell of fecal matter and these people tend to have loser poops? Now is she detecting parkinson or is she detecting IBS or someone that ate something spicy or drank milk while being lactose intolerant.

The issue is the one of the false positive and bayesian statistics. If she's detecting something that has a bunch of common causes then it's not really helpful to run a suite of tests to find an underlying problem on everyone that smells the same.

A fever can be a sign of cancer, but it's also a sign of the flu. Should we check everyone with a fever for cancer?

I'm not an expert, but a very quick search showed a meta analysis[1] which considers the false positives of using volatile biomarkers as a diagnosis. The original paper[2], of which Joy is co-author has a much smaller sample size, but also has a control group to measure false positives.

Again, I'm not an expert, but from personal experience I know that Parkinson's can be hard to diagnose definitively until there are serious symptoms. This test may be relatively poor but still be useful as a piece of evidence.

[1]: https://www.sciencedirect.com/science/article/abs/pii/S03038...

[2]: https://pubs.acs.org/doi/full/10.1021/acscentsci.8b00879

The science direct link isn't working for me.

But here's the problem as I see it. Parkinsons has an occurrence rate in the population of 0.1%. If there are conditions which cause the same smells as Parkinsons and they are more common in the population (1%, 5%, 10%) then this test all the sudden becomes very not useful because even at 1% occurrence rate in the population it's already 10x more likely that you have that condition rather than Parkinsons. That's the confounding problem. And a different comment here pointed out there are conditions that also seem to have exhibited the same smells.

Who knows, perhaps this is still worth it, but for an n=30 study, this is basically nothing to consider. The group size is simply way too small.

BTW, Medical media reporters really should have a "No reporting on studies with n < 500" rule. These sensational studies are always preliminary on really low population groups. I'd love to see the meta analysis to know how many studies it's lumped in and how big those are, though.

If it helps, the meta analysis is called: Volatile organic compounds analysis as promising biomarkers for Parkinson’s disease diagnosis: A systematic review and meta

Absolutely, it may not be useful as a screening test on the general population, but it may be useful as a piece of evidence for diagnosis alongside other pieces of evidence. Even for a test with a lot of false positives, the P(Parkinson's|positive) > P(Parkinson's|negative).

I generally agree that a lot of media doesn't accurately portray uncertainty in medical advancements, buts it's not as simple as having a sample size threshold. It really depends on the significance and the strength of the effect. Also, I want to hear about the exciting preliminary stuff, provided that it's properly caveated. There's just a lot of incentives to sensationalise.

That might be just an imprecise formulation by the journalist.

Smelling tuberculosis seems plausible to me, it is a disease of the lungs.

General cancer probably no, but I have repeatedly read comments by oncologists that sarcoma, specifically, has a distinct smell.

Makes me realise there is no real way to notate a smell.

How do people even get a handle on what a smell is, and how personal an experience is it.

This is a great observation! Indeed, most languages have very few words for smells.

- https://www.theatlantic.com/science/archive/2015/11/the-voca... - https://www.wired.com/2014/11/whats-up-with-that-smells-lang...

Aroma and smell training kits exist for people into wine, beer and so on: https://aromaster.com/

Yes, but is there a notation, like there is for music?

There's a whole pyramid scheme for perfumery, of course.

Notes are for frequency and there are marks for intensity, but you still have to write down what instrument is used.

In general in the vocabulary of smells you try first to understand the instrument (certain fruit, manure, sea, etc) and then the intensity of it.

I'm not very familiar how the intensity and variations are described but I presume that it is not as formal as in the case of music.

But it has no help if you can't smell it (the frequency is out of the hearing limit) or you can't recognize the instrument from the noise.

My sister worked in urology and could smell bladder cancer

Would it be so surprising if she just has an unusually adept sense of smell?

I think the surprising thing to a lot of people, even doctors, was the fact that a disease like Parkinson’s even had a “smell” in the first place. It sounds like now we understand a lot more about this as a result of these individuals ability to smell it!

Advanced lung cancer has a strong, very distinctive smell.

Reply on news.ycombinator.com