The Quantified Self

I've recently subscribed to the Oura ring and 23andMe. The goal is to learn a little bit more about myself. I would describe the results as mostly, "Confirming stuff I already knew and getting almost no actionable information." 

The Oura ring tracks a few vital signs, such as heart rate, respiratory rate, pulse ox, and body temperature. It also tracks movement, so it can tell if you're moving around or not. My interest was in tracking my sleep quality. The Oura ring guesses what phase of sleep you're in based on the various bio markers it can measure. You actually need the proper equipment (as in electrodes connected to the head) and brain-wave detection to measure which phase of sleep you're in, but the Oura ring makes a best guess based on movement, heart rate, etc. Its estimates leave room for improvement. See this paper comparing the ring to a proper polysomnography: 

From EBE analysis, ŌURA ring had a 96% sensitivity to detect sleep,and agreement of 65%, 51%, and 61%, in detecting “light sleep” (N1+N2), “deep sleep” (N3),and REM sleep, respectively.

That's not great. See also this video by a professional sleep researcher, who compared the measurements from his ring to what he got from professional-grade equipment. He got an overall accuracy just shy of 60% on phase-of-sleep detection. My own measurements usually show that I get around one hour of deep sleep. But sometimes it (alarmingly) tells me I only got about 20 minutes worth. I tend to average 1.5-2 hours of REM sleep, the remainder being light sleep. But I don't really believe it. My subjective well-being after a night's sleep is pretty uncorrelated with my Oura ring's sleep score. I wish it gave a lot more caveats and warnings to the user that "this is just an estimate", perhaps giving some kind of confidence interval. It was easy enough to find online reviews of the Oura ring, but I would worry about some users becoming needlessly anxious about their sleep quality based on the ring's inaccuracy. (See also this review by a doctor, who was not impressed.) 

The one thing that I did find useful is that it shows how your heart rate decreases over the course of the night. I have a resting heart rate in the 60s, which is normal. I previously had no way of knowing what my heart rate was while I slept. I now see that it tends to fall over the night and dips down into the 50s or even the high 40s. It also may be useful to know if your sleeping heart rate stabilizes early or late during your sleep. If it stabilizes late, that could mean you're over-training and need to take a break from physical activity. 

I'll give the Oura ring credit for highlighting a recurring pattern in my sleep. I routinely wake up feeling like garbage on Wednesday morning (specifically Wednesdays). My best guess is that it's due to eating a late meal Tuesday nights. I go to jiujitsu Tuesday evenings and get home at around 8:30, usually eating around 9 or later. If you eat a large meal too close to bed time, it can affect your sleep and lead to a late stabilization of your heart rate. I hadn't put this together, but I may experiment with smaller meals Tuesday evenings.

The other handy thing is tracking my average oxygen saturation. I sometimes snore, and I wanted to know if this was just annoying or if it could be full-blown sleep apnea. It runs in my family, so I have a reason to be worried.  I'm in better shape than some of my snoring family members, so maybe it's not a problem. (I don't snore as often or as loudly as my parents or brother, so something that I'm doing is probably working.) My Oura ring almost never detects a problem here. It's reading consistently at 98%. On an occasional night of cruddy sleep, I get a reading of 97% (never lower than that) and a message saying that it detected interruptions in breathing. Flawed as the Oura ring is, I think it's useful to flag potential issues. If it flags too many nights as having had breathing interruptions, I'll do a proper sleep study. (Likewise, if it fails to detect these, but my wife tells me my snoring is getting worse or I'm suffering excessive daytime sleepiness, I'll throw out my Oura ring and get a proper sleep study done.)

The Oura ring also tracks movement, so it can tell you if you move around a lot at night. This is interesting, but not really actionable. I get varying stats on different measures of my sleep score (time in deep sleep, REM sleep, etc.), but I consistently get poor marks for "restlessness." I do have to wake up an reposition myself a few times during the night, but the Oura ring is tracking a lot of tiny movements that don't rise to this level. I don't know what I'm supposed to do with this information. If I went to bed earlier and gave myself longer time to sleep, presumably I'd still be a restless sleeper. 

The information I got from 23andMe was likewise interesting but unsurprising. And none of it was actionable (though in theory it could have been). You get various reports telling you that you are more or less likely than average to experience IBS, sleep apnea, restless leg syndrome, and a host of other conditions. My issue with most of these is that you know from your subjective experience whether you're suffering from them or not. The information that your genes predispose you to it isn't useful. (Perhaps their test counter-indicates the condition, though you have it anyway. A false negative test on a question you already know the answer to.) I think many of these are polygenic tests. That is, they aren't testing for one or two specific genes that vastly change your probability of having the condition. They're based on an analysis of hundreds or thousands of genes, detecting correlations to various conditions. But there are also tests for certain "variants", genes that do have a well-understood connection to various diseases. (Examples are Cystic Fibrosis, Tay-Sachs Disease, and a bunch of conditions I've never heard of.) I was negative for all variants, so it didn't affect me. But I could see how someone would find this information actionable. You'd at least want to follow up with a doctor or work out some kind of monitoring plan. If you and your partner both tested positive for a given variant, it could spark a conversation with a genetics counselor about the potential affect on children. 

23andMe also provides some limited information about ancestry. My ancestry composition was pretty boring and predictable for an American of European ancestry. It pegs me as northwestern European, mostly English/Irish and German. My surname is German, my mother's maiden name is Irish, and both sides have a lot of Irish and German surnames going back generations. I also have a Swedish great great grandmother, though my 23andMe test pegs me as ~2% Finnish and 0% for the other Nordic states. I don't know if the genetic test is inaccurate or if perhaps my Swedish ancestors were recent transplant from Finland before they came to the United States. The breakdown of the British/Irish ancestry was leaning toward the English side, whereas I'd always thought of my ancestry as being more Irish. Maybe my Irish ancestors were also recent transplants from elsewhere in the UK? There was nothing I'd call "exotic". It apparently detected a small trace of Ashkenazi Jewish ancestry, which was intriguing to me. It reads as 0.1%, which comes out to one relative about ten generations ago. (I desperately wish to know more about this person, assuming s/he is real!) But it's hard to know what this means. Should I round it off to zero because it's within some margin of error? But then, why is the test yielding zero percent exactly for Italian, Spanish, East Asian, South Asian, etc., rather than a smattering of 0.1%s all over the place? 

23andMe can also connect you to other people who use the service, though you have to opt in to make yourself "discoverable." It pointed me to a number of distant cousins, only one of whose name I recognized. And it incorrectly marked him as a second cousin, whereas he's really a first cousin, once removed. It's still impressive, given that it's inferring these relationships based on a small market penetration. If, say, 10-20% of the US market used 23andMe, the family trees would be fuller and the connections would be made with more accuracy and greater confidence. 

A neighbor of mine was adopted as a baby. He recently discovered some of his birth siblings because his daughter was a 23andMe subscriber. Apparently they didn't know about him, the birth mother having died without revealing any details, so it's hard to imagine they would have met each other any other way. I think it's wonderful that a consumer service can make such discoveries for us. I don't know exactly what I was hoping for. Like, some forgotten branch of the family tree that nobody knows about. A "Grampa's little secret". (My paternal grandfather did have an out-of-wedlock son, but it's not a family secret.) 

I have heard people say they don't want to do a genetic test because it might conceivably affect their insurance rates. Perhaps the 23andMe data becomes discoverable by insurance underwriters, and they use the information to price or deny disability, life, or health insurance. I actually think this is nuts. You should want as much information as you can get that's relevant to your long-term planning. If there's information in your genes that suggests you're only going to live to, say your mid 50s, it's worth knowing that for planning your future. You probably won't want to have a kid in your 40s, for example. (Though you still may! Maybe you're a raging natalist, bless you, and you care about your hypothetical child's future whether you're around to enjoy it or not.) You might want to save more of your earnings to take care of your spouse and children, or (alternatively) you may decide to spend down your existing savings knowing that you won't be needing it. You may decide to drop some of those nagging health habits that will only pay off in your 70s and 80s. A lot of these decisions could actually go either way. What's weird to me is making this pact with your future insurers: "I won't peek if you don't." For most kinds of insurance, it's irrelevant anyway. Typically you get health insurance through your employer. They basically have to issue you the policy, they can't deny or change rates based on health status. So unless you're buying health insurance on the open market (hardly a "market"), you're fine. For life insurance, typically you'd have a term policy with a rate that was locked in when you bought it (hopefully when you were young). Say you buy a 30-year term life policy when you're 30. It's ten years later, you're curious about what you'll find in a genetic test, so you do 23andMe. Well, your term life policy is locked in. They can't cancel based on new information. It'll cover you until you're 60. At that point, maybe you want another policy and there's some adverse information in your genetic test. (Why, though? A life insurance policy is meant to cover the lost earnings of a breadwinner. By age 60, you've had time to accumulate savings and are usually an empty-nester, so this is less of an issue.) Okay, but if you get the genetic test at age 40 you still have 20 years to plan for that. That's 20 years worth of savings and compounding interest. If I'm genetically predisposed to dying early, I think I'd rather have that information and be able to plan with it, rather than preserving this option of buying cheaper life insurance at age 60. (And wouldn't the life insurer's underwriting discover whatever conditions you have by then? Wouldn't such a condition have started to manifest and leave a trace in your medical history?)