Whoop vs. Oura Ring vs. Apple Watch for Optimizing Sleep and Workout Performance with Continuous Glucose Monitoring and One Meal a Day: Part 1
Wearables are all the rage these days. However, beyond their fancy and comprehensive marketing campaigns, I’m curious - from a scientific perspective - how helpful, reliable, and accurate are these products?
I’m skeptical, of course, but the future looks promising. While these current products are far from perfect, they hint at an exciting vision for improving our health.
Those of you who have been following my work for a while know that I’m keenly interested in the science that should inform our personal decisions regarding diet, sleep, and workouts. I’ve written extensively on ketogenic science, fat metabolism, shrinking fat cells, hydration, salt & electrolytes, “healthy fats”, sleep, muscle growth & flexibility, etc… to help lay a scientific foundation for readers to discuss a practical plan with their medical professionals for how to live.
Everyone is different. An inherent problem with reading health and wellness blogs is that you either get vague references (e.g. the classic “studies show” comment with no links to primary research, or links to not-applicable research), or you get N=1 experiments where the author is unfortunately trying to - often with good intentions - be broadly prescriptive.
My goal in this multi-part article series is to evaluate Whoop, Oura Ring, and Apple Watch for sleep and workout optimization in the context of Continuous Glucose Monitoring (CGM) with a Dexcom G6 (arguably the gold standard for CGM). If I’m successful, then you’ll learn about my perspective of the strengths and weaknesses of these products (and types of products) so you’ll be better informed about whether you want to give motion, heart rate, and glucose tracking a go.
One meal a day
While I’ve been experimenting with ketogenic diets for almost two decades now, and I strongly believe they are an important tool in our wellness toolbox, since October of last year I’ve personally enjoyed daily time-window eating (ad libitum regarding quantity and macros) between roughly 5:30pm and 7:30pm. This is often referred to as OMAD (one meal a day) or “intermittent fasting”, although I think the term “fasting” should be referred to a state where significant autophagy and ketogenesis is occurring, and a ~22hr fast isn’t quite long enough for that in most cases.
As an important caveat/disclaimer, I’m not a medical doctor. My goal is to focus on information transfer only (not advice or prescriptions) so you can make your own decisions, but - within my own story and context - I’ve personally found OMAD to be amazing. When I first started seriously and consistently doing ~22hr daily fasts in October 2019, I found that my standard 3-mile runs before dinner easily turned into 5-mile runs, and I was able to focus on work more easily throughout the day with highly optimized attention and energy levels.
I’m pretty sure that the reason why OMAD works so well for me, however, is because of the decades of experimentation with ketogenic diets, fasting, and time-window eating. I have likely pushed the “metabolic range” of my endocrine system into a state of rapidly being able to switch between deriving energy from more than just glucose and glycogen stores. I’m excited for a future where this specific form of metabolic flexibility can be measured, but until then we have rather “dumb” instruments at the general consumer level to glean insights for ourselves.
So - back to the topic at hand - in this series I’m going to walk through 30 days of N=1 experimentation I did this spring (in COVID-19 quarantine with the rest of the world) with GCM, Oura Ring, Whoop, and an Apple Watch. In part one here I’ll focus on the experimental setup and the type of data that can be acquired.
The experimental setup
First, here are the details of the equipment:
- Apple Watch Series 5.0
- Whoop Strap 3.0
- Oura Ring 2.0
- Dexcom G6
- Precision Xtra (for manual blood BHB and glucose measurement)
- Contour Next EZ (for manual glucose measurement)
- TrueMetrix (for manual glucose measurement, which ended up not being nearly as accurate as the Precision Xtra or Contour Next)
- A Walgreens-branded home blood-pressure device
Regarding the accuracy of manual glucose meters, this comprehensive study showed Contour Next to be the most accurate, so I used that as my gold standard in the 30 days of experiments. Thankfully, the glucose readings from my Precision Xtra matched up to the Contour Next remarkably well.
Second, here are the notable types of data these various devices claim to show me:
- Heart Rate & Resting Heart Rate (bpm) - Apple Watch, Oura Ring, and Whoop
- Heart Rate Variability - Apple Watch, Oura Ring, and Whoop, however there are some VERY IMPORTANT differences between the sample frequency of these three devices. Read this article from Harvard Medical School’s blog if you are not up to speed on HRV and why it’s important. A Whoop only measures your HRV right before you wake up and uses that single measurement as a key variable in it’s recovery score (which we’ll talk about in my next post in this series). An Apple Watch takes HRV recordings automatically once every couple of hours at night, and an Oura Ring takes continuous HRV measurements (I’m guessing every ~5min). The Oura’s HRV graph overnight - assuming you can keep a good fit (which I’ll talk about) - is by far my favorite feature of an Oura Ring compared to Whoop or an Apple Watch.
- Sleep Length & Stages - I’ll show example output below from Oura Ring, Whoop, and a few Apple Watch apps I tested.
- Blood Glucose Levels (mg/dL) - Dexcom G6 (every 5min) and the manual meters listed above. Despite Dexom’s marketing, and my experience with it in summer 2019, this go around the Dexcom had trouble being accurate without calibration.
- Blood Pressure (standard mmHg systolic/diastolic from my Walgreens branded device)
- Blood BetaHydroxybutyrate (BHB, in mM/L) from the Precision Xtra
Example data output
A typical day for me while doing ~22hr fasts - where dinner is my only meal of the day - looks like this from a blood glucose perspective:
This is the default output/screenshot from the Dexcom G6 app. I’ll share more commentary on these levels in future posts in this series.
A 30min interval training workout (moderate level of intensity) looks like this from the Whoop and Apple Watch, respectively (wearing each, one on each wrist, at the same time):
Unfortunately the Oura Ring doesn’t give you much data throughout the day on activity beyond this (below), so I don’t wear it when I’m just sitting/standing while working:
During the course of the experiment I would wear the Oura Ring during workouts - along with the Whoop and Apple Watch - to help with Oura’s activity/rest scoring algorithm.
Now, regarding sleep, here is what the Whoop output looks like when you click on their four stages (awake, light, REM, and deep):
And the Apple Watch app SleepWatch outputs data like this (the same night as above), which I appreciate (it doesn’t even try to guess REM):
And the Pillow app outputs stages like this (below, also the same night), where the pink is their attempt at REM (and Orange is awake):
The Oura Ring output for the same night looks like this:
And here is the heart rate overnight graph and RHR, avg. HRV, body temperature (unique to Oura Ring), and respiratory rate (which the Whoop has but Apple Watch does not).
And finally, the most impressive output from the Oura Ring is this HRV chart overnight:
I’d be interested to see other people’s HRV charts to compare what a “normal” night’s sleep looks like. I’m wondering if HRV, for example, goes up during or around periods of REM sleep. I’ve tried lining up the various graphs, however, and nothing notable pops out.
Regardless, you can see why just using a single HRV data point for a “recovery” algorithm is potentially problematic.
Regarding manual blood glucose measurements, the TrueMetrix seems to always be inaccurately high (here are 5 different readings in the same 5min window):
Whereas the Precision Xtra and Contour Next EZ always either matched or were off by 1 mg/dL from each other (below are readings from two different days)
A quick list of things to be aware of regarding these various devices are the following:
- To get accurate heart rate data from Whoop, I found that I had to wear it very tightly. If I had it even somewhat loose, the heart rate data became very slow compared to the Apple Watch or Oura.
- The Dexcom G6 most often needs calibrating, even if they claim it doesn’t. I wore one for 30 days last summer while coming in and out of ketosis (which I’ll write up in another post/series at some point) and it was more accurate then - right out of the box - than this time (Spring 2020) when it often would start out inaccurately low.
- I found it extremely difficult to get accurate readings from the Oura Ring, and I tried multiple fingers and positions on the finger. I likely wouldn’t have even known I was getting inaccurate readings if I wasn’t also wearing an Apple Watch and a Whoop at the same time. I finally figured out I could wear it on my right index finger - with the ring positioned very tightly up close to my knuckle - in order to get comparable data to the Apple Watch and Whoop overnight.
- I found that I had to calibrate a 10-day session with a G6 sensor roughly 2x (i.e. the beginning and mid-way through).
Obviously there are a ton of trackers out there that measure all sorts of things. (I’m interested, for example, in what is rumored to be blood pressure and blood oxygen level measurements in upcoming versions of the Apple Watch). And I’d expect to see a ton of Dexcom competitors arise this decade for CGM. But for now, comparing the data between Whoop, Apple Watch, and Oura ring while monitoring glucose with a Dexcom G6 seems like a useful exercise, as most of my hard-core wearables and athlete friends tend to use one or more of these devices over something like a Fitbit.
So, for final notes from an experimental setup perspective, with my goal being to optimize sleep and workout performance with ~22hr daily fasts, I did try to establish metrics while eating two meals a day (I couldn't imagine eating three!). I also experimented with eating low-carb or high-carb, and - as I’ll describe in future posts - those didn’t result in great sleep or workouts in the context of how I was performing OMAD. I’m sure there is a way to optimize sleep and workouts in a low or high carb OMAD context, but that is outside the scope of this N=1 experiment.
Author’s note: In my next post in this series we’ll dive into exact food/drink intake, workout choices, and examine the data from the various devices described above. Subscribe to my newsletter to stay posted when I publish the next article. Have questions/comments? Hit me up on Twitter (@wclittle) or email me at firstname.lastname@example.org. Thanks!