Recovery Scores Feel Like Vitals. The Lab Never Sent Them Home.

The cultural name for the earlier, nerdier version of this habit is dated. Gary Wolf and Kevin Kelly founded Quantified Self in 2007, a meetup culture of people who logged weight, mood, location, and sleep in spreadsheets and early gadgets, aiming at 'self-knowledge through numbers.' That project was voluntary, small, and often self-skeptical. The home version that followed is default software on a watch. You do not join a meetup. You wake up to a score you did not ask a clinician to interpret.

This is a different argument from the one about the sleep-product market. Mattresses, melatonin, and smart beds are a retail category. The quantified-self problem is epistemological: why a proprietary index is treated as a measurement of the body, and what the lab-to-wrist swap actually discarded.

Validation studies keep finding the same split. In 2022, Miller, Sargent, and Roach put 53 healthy adults in a sleep laboratory for one night with six consumer devices and with polysomnography. For the two-state question, asleep or awake, agreement with the lab ran from 86 to 89 percent. For the multi-state question, which stage, or wake, agreement fell to 50 to 65 percent, with Cohen's kappa values that the authors read as slight to moderate. Their conclusion was specific: the devices were reasonable for field estimates of when sleep happened and how long it lasted, and not yet adequate for staging.

A 2024 Brigham and Women's study in Sensors compared Oura Ring Gen3, Fitbit Sense 2, and Apple Watch Series 8 against polysomnography in 35 adults without a diagnosed sleep disorder. Sleep-versus-wake sensitivity was 95 percent or higher for all three. Stage sensitivity ranged from 50 to 86 percent depending on device and stage. Oura was not statistically different from the lab on wake, light, deep, or REM duration in that sample. Fitbit and Apple Watch were, in opposite directions on light and deep sleep. The paper discloses funding from Oura Ring Inc. That does not make the measurements fake. It does mean the head-to-head should not be read as an independent ranking of brands, which this article will not provide.

Recovery and readiness scores add a second compression. They typically mix overnight heart-rate variability, resting heart rate, and sleep duration or staging, then weight those inputs in a way the company does not publish as a clinical formula. Heart-rate variability from a good optical sensor can track in the same direction as ECG-derived HRV for many people, which is useful for noticing a trend after a hard week. It is not a diagnosis of overtraining, illness, or 'recovered enough to race.' WHOOP built a business on that index as a membership. Oura and Apple display adjacent scores. The cultural move is the same in each case: several noisy signals become one authoritative integer.

An integer can be reliable in the statistical sense of repeating itself. Wear the same ring, sleep in the same room, get 82 three days running. That reliability is what makes the number feel scientific. Clinical meaning is a different test: does 82 correspond to a construct a sleep physician would recognize, in a population that includes insomnia, apnea, and shift work, not only healthy volunteers in a lab? The validation papers above were run in healthy adults. They do not license the score as a fact about disease.

The regulatory gap is not a loophole someone forgot to close. It is written policy. In 2016, section 3060 of the 21st Century Cures Act removed certain software functions intended only to maintain or encourage a healthy lifestyle, unrelated to diagnosis or treatment, from the definition of a device. The same year, FDA's device center said it did not intend to examine low-risk general wellness products for device compliance. The 2019 guidance restated the two-part test: general wellness intended use, and low risk (not invasive, not implanted, no inherently dangerous energy). In 2026 the agency updated the policy again to address wearables that estimate physiologic parameters such as blood pressure, oxygen saturation, glucose, and heart-rate variability. Those products can still be wellness products if they stay on the wellness side of the line: not for diagnosis, not a substitute for a cleared device, not for medical management.

That 2026 update matters because it is the opposite of a crackdown. As sensors got closer to clinical quantities, FDA clarified that some of those quantities can still be sold as lifestyle numbers. The score culture did not outrun the regulator by accident. It grew inside a category the regulator defined.

Population sleep, measured the old way, did not become a solved problem in the years the scores proliferated. The CDC's National Health Interview Survey found that in 2024, 30.5 percent of US adults averaged less than seven hours of sleep in a 24-hour period, the American Academy of Sleep Medicine's recommended minimum. Only 54.8 percent said they woke feeling well-rested most days or every day. Those are survey answers, not ring data. They are also a reminder that a culture of nightly scores can expand while the share of people who sleep enough does not move in any dramatic way.

In 2017, Kelly Baron's group at Rush and Northwestern described patients who arrived at clinic because a tracker said their sleep was inadequate, including people whose lab studies did not match the device's story. They called the pattern orthosomnia, on the model of orthorexia: a perfectionist quest for 'correct' sleep. It is not an official diagnosis. It is a clinical observation that the score can become the symptom. The quantified self at home is that observation at population scale. The lab never sent a 0-100 recovery index home. The market did, under a wellness policy that asked the score not to claim it was medicine, and did not ask it to stop looking like medicine.