In the countdown to my second Pfizer dose, I covered my quantified self in a blanket of warm data. Here’s what happened when I pulled apart the threads.
Early this past February, I was nervous. I was about to get my second shot of the Pfizer coronavirus vaccine, and I was hoping for a serious reaction. I was planning to monitor my body’s reaction to it and, I hoped, contribute to science. But things weren’t quite working out as planned.
My Oura ring, the health-tracking device I’d worn for five years, was dying, shifting from unreliable to “hospice care.” I had to charge it every day, versus the recommended every-other-day. My Neuroverse monitor, which measures brain waves as I sleep, wouldn’t hold a charge reliably through the night. My Levels/Freestyle Libre blood sugar reports had become increasingly random.
And then there was the underlying question: What sort of reaction would I have to the vaccine? Would I even have one at all? All I got from the first dose was a slightly sore arm, and lots of people barely have a reaction, even the second time around.
This past year my trail as a self tracker ran squarely through the coronavirus pandemic. This is the story of how I made the best of it as a science experiment.
Tracking and detecting COVID-19
I have always been a tracker. When I was five and my brother was three, we both had chicken pox. He walked around the house moaning (sorry, George!), while I busily counted my blisters and scabs, and applied the ancient art of favoring analysis over direct experience; it hurts less if you observe rather than experience a trauma. After I skipped fourth grade, I became extremely conscious of my age and prized my status as “the youngest” of whatever group I was in. I started swimming every day when I was 18. At first I counted laps, but that became too distracting, so I started counting minutes.
Much later, with the advent of email, I discovered the single most reliable marker of my frame of mind: the size of my inbox, which indicates the length of my to-do list and accordingly, my stress level. Later still, I joined Ernest Ramirez, Steve Dean, and many others in the quantified self movement—an eclectic bunch of techies, do-it-yourself medical types, environmentalists and others who track everything from their bowel movements to how often their roommates do the dishes. We used to get together in person before the pandemic (and now do so online) to argue about precision and accuracy. But the fundamental question is purpose. What practical lessons can you glean from what you know? What theories can you construct and test about yourself and others? And does such self awareness help, or merely distract you from more important matters?
We have asked ourselves these same questions for years and shared our data in hopes of finding answers. We are in the second decade of the age of wearable devices. Over time and overlapping, I have tried Fitbits, Jawbones, Misfit Shines, and the ill-fated Basis watch, which was sold to Intel and then vanished. All those pioneering brands were mostly motion trackers, but I also wore to bed an early electroencephalogram (EEG) headband called “the Zeo”—though I have no idea how accurate it was.
Nevertheless I have monitored and recorded my sleep patterns for years—first with the Zeo, and continuing with the current generation of sleep trackers, which can track heart rate and even temperature and respiration, and do a good job of assessing sleep quality. They also feed the data into algorithms that try to identify sleep stages, though with less accurate results, as far as I can tell.
To be blunt, all the trackers I use currently disagree on the details. For the record, they are SleepScore (formerly ResMed), Oura, Whoop, and Neuroverse. I’ve also used Circadia and Dreem, both of which have shifted focus from consumer to clinical markets. And my glucose monitor also records a regular overnight dip in blood sugar. But in many ways the fact that they disagree doesn’t matter. The point is not absolute agreement, but some kind of consonance in response to outside changes—whether that’s an overnight flight or some longer-term behavior change such as no more evening meals or switching from swimming to rowing in my bedroom (courtesy of COVID-19). Sleep trackers reveal the state of the whole body, and they can register your body’s response to things like pathogens, food poisoning, injuries, too many late nights out, simple noise… or a vaccine. All of those will register as deviations from your baseline sleep patterns.
And generally, all of the trackers show those deviations. In the end, I like the Oura the best. Since 2016 it has been feeding back my overall sleep quality, as far as I can determine, better than any other tracker. And I received a new and noticeably improved ring, with more accurate deep sleep results vis a vis the Neuroverse, just in time for my second Pfizer dose.
At last—scientific truth
Unknowingly, I began to prepare for my N of 1 study with the COVID-19 vaccine and how it affects my sleep almost two years ago, when I met the team from Neuroverse. They have a pre-market version of an intriguing EEG device with a lot of science and some patented algorithms to interpret the data it records, along with some neurofeedback interventions. The device sticks to your forehead, records your brain waves and movement (including which side you slept on!), and scientifically determines which of the four sleep stages you are in throughout the night. It provides a hypnogram record of those stages plus other factors such as restlessness and the intensity of your brain waves.
At last, I thought, the scientific truth! Now I would know whether my pathetically low deep sleep numbers from Oura were for real, or just an artifact of an algorithm that doesn’t quite match my particular way of sleeping deep. After all, the four stages of sleep are determined by looking at actual brain waves and not just by detecting clever correlations and using algorithms.
Yes, deep sleep and REM (dream) sleep are the prized slumber states in the sense they help your body and mind recover. However, sleep stages are defined by a number of different wave patterns (spindles, K-complexes, intensity ratios and the like), and they don’t always match neatly. It’s the kind of thing that currently takes brain scientists hours to determine for each night. Neuroverse is developing algorithms to do it automatically, and my data along with others’ is being used for this worthy task.
Yet it turns out that “scientifically” segmenting your night into sleep stages is not the whole picture after all. Indeed, the Neuroverse plus the Oura and the SleepScore (which measures sounds and deduces sleep based on breathing) pretty much constitute a DIY, pieced-together “polysomnography” of the kind you might get in a sleep lab (minus the folks in white coats).
Here’s the thing: You may indeed be in a period of deep sleep, for example, but it can still be “bad” sleep. Your resting heart rate may be high or your heart rate variability low; you may be breathing too fast or your body may be too hot, if not actually feverish. The brain manages these conditions too, of course, but mostly independently of sleep stages.
My sleep stages are not the final truth but more like fractured glimpses of a much more complex picture.
As it turns out, I’m very restless—I move a lot—even when I’m in deep sleep. I wake up frequently, and go back to sleep easily. Neuroverse notes my movements as a sleep quality factor but appropriately ignores them when determining whether I’m in deep sleep, which affects the staging. (This is subtle, but staging is about the type of sleep itself, in the brain, versus the state of the whole sleeping body.). By contrast, I suspect, Oura’s algorithm may use restlessness to interpret some of my deep sleep or REM as light sleep, since it has no access to my brain waves.
So after years of following my sleep readings in various iterations, I was eager to see how they would register my response to the vaccine.
Meanwhile, what does the Oura ring have to say? Oura reports two primary metrics for sleep: readiness and sleep score. The readiness score is an overall measure of your recovery. It signals your capacity to perform at your mental, emotional, and physical best, based on body signals, including resting heart rate, heart rate variability, respiratory rate and temperature, in the context of recent sleep and activity patterns, to determine how well-rested you are and whether you’re ready for a challenge. Your sleep score is focused on the sleep itself—the total duration, the stages, the timing, and the sequence of it all, using your heart rate as a key metric for the algorithm that assesses quality and determines sleep stages.
A touch of COVID-19?
But first, did I already have COVID-19 last year? I was traveling in February, as usual, and came down with what I thought was a bad cold—something that’s unusual for me. In fact, I never got a good read on my temperature because I do not even own a thermometer (and have never missed a day of work for illness). I was in San Diego with my work team and ended up going to bed before dinner because I simply couldn’t stay awake. Overall, I felt crappy and was coughing, but nothing earth-shaking.
As you can see from my data above, my normal average resting heart rate is around 45, and my readiness as measured by Oura is usually in the 80s (out of 100, I assume). That week a year ago I managed to swim every day as usual, though I started to feel sick and went to bed around 5:00 pm on Thursday, February 5—a week after the World Health Organization declared the novel coronavirus a public health emergency of international concern. I got extremely chilled in the pool the next morning and took a long time to stop shivering. COVID-19 was not yet really in the news except for a few cases in Seattle linked to a traveler from China. Yes, I travel a lot around the world and frequent international air lounges in the United States, but it seemed far off and exotic.
I kept feeling crummy. On February 11 my readiness per Oura dipped to 44 and my temperature was three degrees above normal. I finally acquiesced to the pleas of my friends to see a doctor before my next trip a few days later to Amsterdam. The doctor tested me for influenza—negative—and then prescribed an antibiotic against a sinus infection and pneumonia. Whether it was that treatment or my natural recovery—it certainly wasn’t my overnight flight to Amsterdam—two weeks later I was feeling normal again.
At the time, there were no tests for COVID-19 antibodies. By the time there were and I managed to take one and then another in May, I tested negative—but too late to be sure. I may never know whether I actually had the disease.
Dry runs—Shingrix and flu vaccines
Last fall, I had two dry runs for observing how my biological signals might respond to the COVID-19 vaccination. First was receiving my second dose of the Shingrix shingles vaccine on September 4 (I did have a bad case of chicken pox as a kid, you will recall). Not much action there: My average nightly resting heart rate increased to 49, my readiness dropped to 56, and my body temperature was 1.8 degrees above normal—but overall it wasn’t too bad.
Next I got my regular annual flu vaccine on Halloween day, went to bed early that night, and slept poorly but stayed in bed for 10.5 hours. This seemed like more of a dry run for what I could expect with the coronavirus vaccine a few months later.
That night I had an elevated average resting heart rate and saw my heart rate variability reduced. I felt bad the next day, but I recovered by the third night (see chart above). My qualitative assessment was that the vaccine had a noticeable effect, to be sure, but it was hardly a scary event.
Overall, after the flu shot, my sleep and readiness scores were both visibly below average, and my temperature was up by three degrees. But apart from the temperature, that’s not unheard of for me. Prior to the pandemic, when I regularly flew overnight every month or so, I frequently had a night of lower than normal scores, but I usually recovered the next night. (Backing up that observation is the fact that my best sleep scores ever were in the months of March to April 2020, when I stopped traveling altogether and slept better on average because of it.)
After these dry runs, I was finally ready for the grand finale: the coronavirus vaccine. I received my first dose on January 19, and it passed without much quantitative fanfare. My readiness score was 68, and my sleep score was 69—both slightly lower than the normal 80-ish.
Indeed (see chart below), I felt much worse two weeks later, on February 3—bad enough that I went to get a COVID-19 test not so much because I thought I had COVID-19 (it felt like a bad cold) but because I wanted to feel free to keep up my normal routine without infecting anyone. Probably it was just a mild cold. Anyway, it disappeared.
I showed up to take dose #2 of the Pfizer COVID-19 vaccine on February 13 in the same Mount Sinai facility on First Avenue in New York City as the first time, close enough to walk there and back. The second time, I felt like an old hand—or arm.
I was hoping for a fairly serious reaction—something that would confirm that the vaccine was inducing a reaction. After a tough Valentine’s Day night of vaccine malaise during which I slept poorly, I rebounded nicely the next night.
The second dose of COVID-19 vaccine seemed to wear me out and require more sleep to recover. I slept a lot and actually increased my readiness (despite a slightly increased heart rate), as measured by the Oura ring. But my Neuroverse data showed one delayed night of notably worse staging. In short, my body demanded and actually got more sleep, but the actual duration of deep and REM sleep seemed to be impaired—presumably by the vaccine response—the second night in particular.
There are no Neuroverse results included in the 2020 graphs because I did not carry the device with me while traveling in February 2020 (the charging process made that inconvenient), and because I was waiting for a device upgrade in November 2020. So I have the results only for the third graph. All I can say is that I look forward to getting sick and to traveling overnight again, so I can continue to study the reaction of my brain waves in particular. The experiment continues.
The whole point of writing this piece was to learn something. In the end, I didn’t learn much from the data. But I did learn a lot from looking at it—how difficult it is to piece together a coherent narrative. The complexity is the truth; simplicity is deceptive. Indeed, the more I discovered… the less I knew.
Through a tracker darkly
Over a lifetime of tracking, I have learned a great deal about my body, but I have also learned how much more there is to know. Scientists refer to living things in general as complex adaptive systems: An individual organism is not just the sum of its parts but a bigger, emergently different whole. It’s the way you can tell the difference between liquid water and frozen ice not by looking at the individual molecules of water, but by watching how they interact. The molecules are the same, but their dynamics are very different.
In the end, the data are not a diagnosis. They are an alert.
What I’m coming to learn is that my sleep stages are not the final truth but more like fractured glimpses of a much more complex picture. These various scores reflect the reality that sleep is just one part of what happens to you overnight. What I got around my vaccination was the usual: “Something is going on…” but no real explanation, just as I used to get for a night spent trying to sleep in the window seat of an airplane. If you have a likely cause, that explanation is probably obvious. But in the end, the data are not a diagnosis. They are an alert.
Maybe that’s the point. As I worked my way through the text of this article and tried to match it to the tiny screenshots from my phone, I got more and more frustrated with my inability to see the bigger picture. Then I considered the elephant in the room—literally—the allegory of the six blind men and the elephant. Each man touches a different part of the beast and imagines an entirely different animal.
Perhaps detecting some subtle, specific immune reaction or accurately diagnosing some disease state isn’t what sleep trackers are good for at all. The trackers are like the blind men tugging at the tail, patting the trunk, or feeling the floppy ears. They capture some small part of it but never the whole thing. They see the water molecules but not the liquid or solid they form.
In the end, I was able to get good, usable data that provided useful insights, but nothing that gave a clear signal, no earth-shaking discoveries. That’s how science works sometimes.
Maybe the point of these trackers is to help us build a subtle, data-driven resilience against disease, rather than an analysis of disease. “How are you?” versus “How is your disease?” There is no single number that tells it all, but rather just a collection of moving parts pointing to a whole we never see. In my case, wearing the devices and checking the numbers has given me the social courage to carve out time for a swim, to beg off early for bed without apology, to say things like “No dessert, thank you!” and to steer my body in a healthier direction.
That’s why I wear these things. The data are there, and it’s up to me to interpret them in the context of what I know. In the end, I am my own elephant.
Disclosures: I’m an investor in Circadia, which monitors respiratory rates for FDA-approved use in hospitals but is not really suitable for consumer use. I’m also an investor in Levels, which tracks blood glucose using the Freestyle Libre device from Abbott. And I am a beta tester of the Neuroverse device mentioned in this article and an informal advisor to the company.
Editor’s note: This story was updated on April 25, 2021 to correct what the Neuroverse data showed in response to the second COVID-19 vaccine dose.