Hacker News new | comments | ask | show | jobs | submit login
Quantified Self: investigating Vitamin D & sleep with a Zeo (gwern.net)
79 points by gwern on Feb 14, 2012 | hide | past | web | favorite | 30 comments



I've pretty much given up on experiments like this. There are just too many variables to control for; maybe you're sleeping poorly because you're stressed out rather than because of the Vitamin D pill you happened to take today. I've used a Zeo and can draw conclusions between what the chart shows and how I feel, but I haven't been able to do anything to control what the chart looks like.

Honestly, the deepest effect I've been able to determine through personal experimentation is that coffee contains some sort of stimulant. Anything else is too subtle for me to be confident about.


> There are just too many variables to control for; maybe you're sleeping poorly because you're stressed out rather than because of the Vitamin D pill you happened to take today.

The answer to variables and problems is better setups (more blinding, more randomization, more data), and not counsels of despair. I agree that there's a lot of variability in sleep scores, which is why I run things for as long as I have patience for (in this case, ~43 days; with melatonin, I could run for many months because it wasn't hurting my sleep but helping it).


It is worth it to point out that Vitamin D has been shown to prevent colds...but the "recommended" dose unfortunately still leaves people deficient.

A nutritionist recommended I take at least 5000IU a day (with regular blood tests to confirm I wasn't going over the limit.) My blood test level started out at 18 (extremely low, despite the fact that I lived in San Diego at the time) and is now a more respectable mid-50s.

Most Americans are chronically Vitamin D-deficient (sources: http://www.scientificamerican.com/article.cfm?id=vitamin-d-d... and http://planetgreen.discovery.com/food-health/vitamin-d-defic...) and at this point, it's pretty much safe to assume that you're not getting enough of it and to consider taking a 2000-5000IU daily supplement.

Anecdotally, I can say it's made such a huge difference in me getting less sick on a regular basis (and is SO cheap) that I've been recommending it to pretty much everyone I know.

I've also become a huge fan of NOT wearing sunscreen, especially if you're going to be out in the sun for less than an hour. With proper Vitamin D levels, you actually burn less. Try it for yourself and see--it's been a massive difference in my life.

Vitamin D I use (Only $8 for 4 month's worth! Disclaimer--Amazon affiliate link, but it is actually what I use) http://amzn.to/zJYS1X


Important information is missing here, you age, gender, and race. (I am not asking you to provide this information, I'm suggesting people see medical advise before arbitrarily starting a supplement)

Your skin's pigment density is an important aspect of how you handle sunlight. Your ability to process these chemicals is a function of all three, lifestyle, and other health factors.

http://www.medscape.com/viewarticle/715417 http://en.wikipedia.org/wiki/Vitamin_D


FYI: the non-affiliate link version: http://www.amazon.com/gp/product/B002EA99HE


Have you considered sublingual vitamin D? Just curious since I'd always thought it to be the preferred form but see more people caps.


I have, but the ones I got tended to crumble into tiny pieces before I could even get them in my mouth, so I went back to the capsules.


I currently use liquid sublingual but wonder about shelf life. Going to give the caps a try.


gwern, after reading the rest of that page, especially the melatonin section, it looks like your next sleep experiment should be a double-blind experiment on getting rid of the cat. I'm not sure how you'd double-blind that, but extrapolating linearly from your description of how you double-blinded the Vitamin D capsules I would imagine it looks something like having a housemate randomly stuff the cat into one of two bags and you randomly opening precisely one each night via some sort of complicated-but-scientific mechanical contraption. (Perhaps based on the random decay of a radioactive atom for some extra classical flavor.) I would definitely expect some sort of statistically significant effects on sleep to arise from such an experimental setup. In fact, it would probably be so statistically significant that even the others sharing your house would experience statistically significant effects on their sleep, and that's a whole heaping helping of significance.


> I would imagine it looks something like having a housemate randomly stuff the cat into one of two bags and you randomly opening precisely one each night via some sort of complicated-but-scientific mechanical contraption.

I dasn't - I'd be accused of being a sociopathic animal abuser (again). On the other hand, there must be some way to rig the door to open at, say, 1 AM with 1:1 odds. Hm...


This guy is either a genius or he's completely crazy: http://www.gwern.net/Nootropics


I am most definitely not a genius (http://www.gwern.net/Links#iq), so I guess I am just crazy. Oh well - hopefully I'm the entertaining kind of crazy and not the unpleasant kind.


Gwern has an addictive element of thoroughness, candidness, and sheer information density in his posts.


I'm not a statistics expert, but I would argue that using a t-test is not appropriate in this setting where you are alternating melatonin and non-melatonin days. There are two potential reasons for this:

1. One night's sleep is likely to influence the next night, and this cannot really be controlled for in the current design.

2. I think there is 'information' or statistical power available in the scenario where you sleep poorly on a melatonin off day after sleeping well on a melatonin on day, because this supports the hypothesis that melatonin helps sleep rather than the null. The key to making n=1 experiments more relevant is harnessing this fact statistically. Not sure how one would do this however.


> I'm not a statistics expert, but I would argue that using a t-test is not appropriate in this setting where you are alternating melatonin and non-melatonin days.

I'm not an expert either; I just do the best I can, and using a t-test is better than not using any test, y'know?

> 1. One night's sleep is likely to influence the next night, and this cannot really be controlled for in the current design.

Seth Roberts actually said something like that about my vitamin D setup, arguing there could be spillover effects to the next day, which is why I spent any time looking at the 'lagged nights' to see if there was any odd behavior in the following nights. His suggestion was to randomize blocks of nights or even weeks, rather than individual days. Of course, it was a bit late to start over and I didn't really feel like running another experiment...


He really needs a positive control, i.e. something he knows for sure improves his sleep in order to quantify the effect. Say his data shows a small effect, and it is even statistically significant. He still won't have learned anything without having some reference for the magnitude of the change.


> He still won't have learned anything without having some reference for the magnitude of the change.

I don't think that's true. This is what the concept of 'effect size' (http://en.wikipedia.org/wiki/Effect_size) was developed for, quantifying how big a divergence we're talking about. I didn't calculate it here because I literally just starting using R a few days ago and I don't know how.

A quick and dirty way is to look at the difference in the 2 means and compare it to what is the 'normal range' (intuitively something like a standard deviation or two), which in this case, my ZQ ranges normally from the mid-80s to the very low 100s (a range of ~20 points), and the difference in ZQ means is 4 points. So that's 1/5 the normal variability - that's nothing to sneeze at! (By a strange coincidence, this is almost the same sort of numbers as for grades in class, so we could analogize: not taking vitamin D at night would improve my sleep grades by one letter. When I was in school, I would not have turned my nose up at such an intervention.)

So, your comment is wrong as stated, I think. But as it happens, I do have something which could serve as your 'positive control': my previous (unblinded non-randomized) experiment with melatonin over half a year or so, which found a shift in mean ZQs of ~8 points. I was really impressed with melatonin when I first started (as have many people been), so to find vitamin D is half as hurtful as melatonin was helpful... I had no plans to take vitamin D in the evening before, but I sure as heck don't now!


nice! It's nothing more that an N-1 experiment, but I've really enjoyed using the Zeo to try and quantify what my sleep is doing.

A couple years ago I wrote a little lucid dreaming app for the zeo you might be interested in.

http://blog.myzeo.com/forum/zeo-raw-data-library/lucid-dream...


oh, also, in terms of replacing the pads. Every couple months I put a little dab of EEG conductive paste on the pads and they seem to last forever. I have needed to change the headband, though.


Really, that's a very interesting strategy. So would that paste be something like http://www.amazon.com/Parker-Labs-Tensive-Conductive-Adhesiv... ?

(Also, I do plan to look into programs that work with the serial cable - more for meditation than lucid dreaming - but I haven't bought the cables and connectors yet.)


yup, except that you don't actually want the adhesive. I use Ten20 conductive gel. http://www.amazon.com/Ten20-EEG-Conductive-Paste-Tube/dp/B00...


Your link seems broken, BTW. (Zeo moved their forums at some point in the past few months and broke every link.)


oh shoot, thanks for noticing! Here's the fixed link: http://www.myzeo.com/sleep/node/714


Have you noticed Seth Robert's more recent postings about time of day being critical to Vitamin-D effect? It sounds as though you were taking it just before bed - which is the exact opposite of what he now recommends.


As I say in the first sentence, those anecdotes were why I started in the first place.


You say: "Seth Roberts has speculated that vitamin D, despite its myriads of other benefits, may harm sleep based on some anecdotes (with 2 null results)"

My point is that he's collected a lot more anecdotes since then, and there seems to be some indication that Vitamin-D helps sleep when taken at sunrise rather than at night. I was curious about why you didn't mention this. I'm guessing that the posting just pre-dates his newer observations.


I got tired of collecting anecdotes; after a little bit, one more anecdote 'well I kind of sort of felt better although you know I didn't collect any kind of data whatsoever' is just pointless to link to. Sleep is variable, you can see any pattern you want to in a few days, especially with a dab of selection bias. (If I had stopped after 10 days, the means would have looked more impressive; but the p-values tell the truth.)


Have you considered taking Vitamin-D at sunrise as Seth suggests?


Gwern wrote in his experimental design that he took Vitamin D supplements in the mornings, when he hadn't taken them the night before.


Enough people have asked that I've decided to test out morning vitamin D. Still pondering the exact protocol for randomizing & blinding.




Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: