Let’s Talk About That Sleep Score
I have a patient who I’ve been treating for chronic insomnia and low energy during they day. The treatment is going well and they have been feeling better—more rested, more clear-headed, more resilient during the day. But at our evaluation lat week about their treatment plan, they said to me:
“I’ve been feeling great… but my Oura ring still says my sleep is terrible.”
And they are not the first person to tell me that their experience of sleep is so much better but their app is telling them that its not. Sometimes it’s the opposite. For some people, their sleep score is high, but they wake up groggy or crash by mid-afternoon. But in either case, they’ve been trained to trust the app over their own experience. And when the app doesn’t validate how they feel, they start to second-guess the progress we’re making. Or worse—they feel discouraged, like their body isn’t cooperating despite their efforts.
Sleep monitors like Oura and Fitbit are everywhere now, worn as badges of health optimization and self-awareness. But when they start to dictate how we feel about our own bodies, it’s worth asking: What are these devices actually measuring? How accurate are they? And what’s the cost of outsourcing our internal sense of rest to a commercial algorithm?
Let’s take a closer look.
What Are Sleep Trackers Measuring, Really?
Sleep, as it's commonly understood in modern science, is divided into several stages: light sleep, deep sleep (also known as slow-wave or N3 sleep), and REM sleep. These stages are primarily defined by patterns of brain activity measured through electroencephalography (EEG). Light sleep is a transitional state, where brainwaves begin to slow, muscles relax, and awareness of the outside world starts to fade. Deep sleep is marked by delta waves—slow, high-amplitude brain activity associated with physical restoration, tissue repair, and immune function. REM sleep, on the other hand, is paradoxical: the brain becomes highly active, almost wake-like, while the body remains effectively paralyzed. This is when dreaming occurs, and it is thought to be key for memory consolidation and emotional processing.
This structure gives researchers and clinicians a way to analyze sleep in measurable terms. EEG provides a visual record of what the brain is doing moment by moment, and from this we’ve built a framework that assigns meaning to each phase. That framework is valuable. It gives us a shared language to discuss sleep quality, to compare states across time and populations, and to identify patterns associated with dysfunction.
More fundamentally, the idea that sleep can be fully understood through a chart of waveforms may itself be too narrow. EEG can tell us what kind of electrical activity is present in the brain, but it cannot tell us how that sleep is lived. It doesn’t know if you felt safe, or if your dreams left a residue of sadness. It doesn’t know if you woke feeling held or fractured, restored or restless. These are human experiences, and they are part of sleep, too. Nonetheless, EEG and its clinical assessment partner polysomnography (PSG) are the biomedical tools we have and they are the assessment metrics that sleep tracker devices are trying to mimic.
When it comes to the at-home sleep monitors like Oura and Fitbit, those devices rely on a small suite of biometrics, mostly collected through sensors on your finger or wrist. Here’s what they’re actually recording during the night:
Heart Rate (HR): Tracks how many times your heart beats per minute. Typically lower during deep sleep.
Heart Rate Variability (HRV): Measures the variation in time between heartbeats, associated with parasympathetic nervous system activity (rest and recovery).
Movement (Actigraphy): Uses accelerometers to detect tossing, turning, and stillness, often interpreted as sleep depth.
Skin Temperature: Relative changes in peripheral body temperature, often correlated with circadian phase and recovery.
Respiratory Rate: Estimated from subtle physical signals like heart rate patterns and temperature variation, but not actually able to read the movement of air into and out of lungs.
Blood Oxygen (SpO₂): On some devices, measures how well your blood is carrying oxygen during the night.
These are the ingredients. From there, proprietary algorithms infer sleep stages (light, deep, REM, wake) and calculate a composite “sleep score” meant to represent your sleep quality on a scale from poor to excellent. Note the word proprietary here because none of the companies in this space will tell you exactly how they calculate your sleep score — “trade secrets” and what not…
Notable though. when sleep trackers attempt to recreate the EEG system using proxies like the ones listed above, they step even further away from the source. A Fitbit-type device cannot measure brainwaves directly. It can only infer, based on surface-level data, which stage a person might be in. And while those inferences are sometimes correct, they are also often wrong—particularly when it comes to deep sleep and REM. The body may be still during both, but the internal experiences are radically different. To a wristband, they can look the same.
When patients come in saying they feel rested, but their device disagrees, we often encourage them to believe their body. That doesn’t mean we ignore data—but it means we don’t let it override lived experience. Sleep, like health more broadly, is not just what we can measure. It’s also what we can feel.
But here’s the critical thing: none of these devices are reading your brain.
In a sleep lab, clinicians use polysomnography (PSG) to track brain waves, eye movements, muscle tone, breathing, heart activity, and more. This is how sleep stages are defined—not by heart rate or motion, but by patterns of electrical activity in the brain.
By contrast, commercial devices are just guessing at those stages based on peripheral proxies. Their accuracy in doing so is… mixed.
Validation Studies: The Data Doesn’t Quite Match
Several independent studies* have compared consumer sleep trackers like the Oura Ring and Fitbit to clinical-grade sleep testing(PSG). These studies generally show that while wearables can track general trends, they fall short when it comes to accurate sleep staging and nuanced interpretation.
The most common issues include:
Total sleep time: Both Oura and Fitbit show high agreement with clinical measurements here. Most devices can detect when you are generally asleep versus awake.
REM sleep: Oura's Gen 3 ring performs reasonably well at estimating REM duration, though its accuracy varies by user.
Light vs. deep sleep: Both devices tend to overestimate light sleep and underestimate deep sleep, with Fitbit in particular detecting deep sleep correctly only about 50% of the time compared to EEG measurements.
Wake after sleep onset (WASO): This important insomnia metric is routinely underestimated by both devices.
Overall staging of sleep: The further a metric gets from total time asleep, the more the accuracy drops.
These discrepancies have real-world effects. When a patient comes in saying their Oura Ring reported zero deep sleep, but they woke up feeling clear and rested, they’re often confused. And understandably so. The device’s conclusion doesn’t match their experience. In cases like this, the tracker probably missed the signal. These tools rely on proxies like heart rate, HRV, and movement—not the brain activity that actually defines sleep stages.
Deep sleep, for example, is defined by slow-wave EEG patterns, something a wearable cannot directly detect. REM sleep involves high brain activity and vivid dreams while the body remains still. A wristband can’t see what the eyes are doing. It can’t monitor brainwaves. It can only measure what’s happening on the surface—and then infer what’s happening inside.
So the algorithm does its best. It guesses. And often, it guesses wrong.
The danger is not just in technical inaccuracy. It's in the impact those guesses have on the person wearing the device. We’ve seen patients feel discouraged by a low score, even when their subjective experience of sleep was good. Others spiral into anxiety after one “bad night,” despite no change in energy, cognition, or mood. In some cases, people have begun to structure their entire day around the number on their screen—eating differently, canceling plans, or even skipping exercise based on what the ring told them.
In that context, the data has usurped human experience and understanding, and we have ceded our agency to a machine that is wrong 25-40% of the time. Once we hit that point, the abidication of our relationship to our bodies is not so much a technological problem as a psychological one.
The Psychological Toll of Quantifying the Unconscious
Sleep is one of the last frontiers of health that happens entirely without conscious effort. It’s mysterious and restorative, an internal process that unfolds beyond our control. And perhaps because of that, there’s a particular vulnerability in trying to pin it down with numbers.
When a wearable device offers a neat little score each morning, it’s tempting to take that as truth. Over time, people begin to check their app before they check in with themselves. They start to wonder not “how do I feel this morning?” but “what does the ring say?” As we’ve discussed, a bad score can color the whole day—even if the night felt restful. A good score can override a lingering sense of fatigue.
This gap between subjective experience and algorithmic output creates a subtle but profound shift: people begin to trust the data over their own bodies. We’ve seen patients who are sleeping better by every meaningful measure—more energy, fewer night wakings, calmer mornings—start to question that progress because their sleep score hasn’t budged. Others have reported a sense of deflation after a single “bad” score, as if their own sense of rest had been invalidated.
There’s even a name for this phenomenon: orthosomnia—a condition where obsession with sleep data creates anxiety and, paradoxically, worsens sleep. The numbers become not a support, but a source of stress. In trying to optimize the unconscious, we end up bringing more tension into the very system we’re trying to soothe.
This isn’t just about the accuracy of the data. It’s about how we relate to it—and whether we still allow ourselves to trust the quieter signals of our own physiology.
Who Benefits from This System?
When a patient’s sense of healing begins to unravel because of a number on their phone, it becomes important to ask a harder question: who, exactly, is this system serving?
Technology companies like Oura and Fitbit present themselves as partners in health. They suggest that wearing a device will deepen self-awareness, encourage better habits, and provide meaningful insight into the body’s rhythms. The branding is subtle but persuasive—language that frames the product as a mirror, as if these companies are simply helping you see yourself more clearly.
But clarity is not the real commodity here. Engagement is.
These devices are designed to encourage daily interaction. Sleep scores, readiness rings, temperature trends, recovery insights—each one is a prompt to return to the app, to re-enter the loop, to keep checking. That behavior is not accidental; it is engineered. The more you interact with the app, the more valuable you become as a user. Not because you are getting healthier, but because your attention, your habits, and your data are monetizable assets.
For example, Oura requires a paid monthly membership to access detailed metrics, including many that the ring is already collecting. Fitbit offers a “premium” tier that unlocks expanded analysis and coaching tools. In both cases, your own body’s data is held behind a paywall. What you are buying is not just a health device—it is ongoing access to information about yourself, packaged and interpreted by someone else’s algorithm.
Beyond subscriptions, there are additional layers of value extraction. Your biometric data—whether or not it is personally identifiable—can be aggregated, analyzed, and repurposed. These data streams are useful for machine learning development, corporate wellness programs, health research partnerships, and future product rollouts. In this economy, the body is not sacred. It is a resource to be mined.
It is easy to forget that these companies do not exist to support your healing. They exist to generate returns for investors. That is not a cynical interpretation of their efforts; it is the legal obligation of a company to make money for investors. Publicly traded companies are required to prioritize shareholder value, and their boards are bound by a legal fiduciary responsiblity to work in the best interests of their shareholders financial welfare, which means decisions around design, data use, and product development are ultimately driven by profitability—not by human wellbeing.
The impact of this model is not always visible right away. It reveals itself slowly, in the gradual erosion of trust in your own sensations. Over time, the app becomes the authority. If it tells you that your sleep was poor, you may start to doubt how rested you feel. If it gives you a high readiness score, you may push through fatigue that your body was asking you to respect. Each time that shift happens—each time an external score overrides an internal signal—the technology becomes a little more central, and you become a little less sovereign in your own experience.
This is the deeper concern. Not that the metrics are imperfect, though they are. Not even that the business model relies on dependence, though it does. The concern is that, under the guise of empowerment, we are being conditioned to hand over the most basic elements of bodily wisdom to systems that do not know us, do not care for us, and are not designed to support healing in any real or relational way.
The more we rely on these devices to tell us how we feel, the more difficult it becomes to hear what the body is saying on its own.
Returning to the Wisdom of the Body
There is a kind of medicine that lives outside of screens. It does not need scores or apps or predictive algorithms to tell you how you are doing. It begins with something much older, and much more intimate: your own felt sense of being alive.
This is not romantic nostalgia. It is a call to remember that your body has its own language—one that can be learned, listened to, and trusted. Every day, your system offers signals about what it needs and how it’s doing: the texture of your energy, the steadiness of your breath, the clarity of your thoughts, the quality of your rest. These signals are not noise. They are the foundation of real self-knowledge.
At Root & Branch, this is where we begin. We practice a form of medicine that does not extract you from your experience but guides you more deeply into it. Chinese medicine has always been rooted in observation—not just by the practitioner, but by the patient. The pulse, the breath, the sleep, the dreams—these are meaningful data points, but they are not reduced to numbers. They are read in context, with care and curiosity.
When patients come to us, we are not trying to optimize them. We are trying to help them feel at home in their bodies again. Sometimes that means sleeping more deeply. Sometimes it means waking with clarity. Sometimes it means understanding what fatigue is trying to say, rather than overriding it. We do not measure success by how well someone fits a norm. We measure it by how clearly they can hear themselves—and how gently they are able to respond.
There is a place for technology, but it should never replace your own inner sense of knowing. You do not need permission from a device to trust how you feel. You do not need a score to validate your rest. What you need is space to reconnect—with your breath, with your rhythms, with the signals your body is offering every day.
If you are ready to come back to yourself, we are here. Root & Branch is a place where your experience matters. Where we listen. Where we help you learn to listen, too.
Sources*
Brigham and Women’s Hospital. “Wearable Sleep Trackers Put to the Test: Oura Ring, Apple Watch, Fitbit Compared to Gold-Standard Sleep Study.” Sleep Review, 7 May 2024, https://sleepreviewmag.com/sleep-diagnostics/consumer-sleep-tracking/wearable-sleep-trackers/oura-ring-apple-watch-fitbit-face-off-sleep-accuracy-study/.
Jeon, M., et al. “Validation of Fitbit Inspire 2 for Sleep Staging Against Polysomnography.” Journal of Sleep Medicine, vol. 21, no. 1, Jan. 2024, pp. 15–22. https://pmc.ncbi.nlm.nih.gov/articles/PMC9985403/.
Lee, S.Y., et al. “Accuracy of Fitbit Sleep Staging Compared With Polysomnography in Healthy Adults.” Journal of Sleep Medicine, vol. 18, no. 3, Mar. 2022, pp. 131–139. https://e-jsm.org/journal/view.php?doi=10.13078/jsm.220019.
Oura Health. “Oura Ring Validated Against Polysomnography by University of Tokyo Researchers.” Oura Blog, 18 Oct. 2023, https://ouraring.com/blog/oura-ring-accuracy-validation-study-university-of-tokyo/.