Wearable readiness scores: how a proprietary index gets assembled from a few signals

What the device is physically measuring

Consumer wrist and ring devices measure a small number of physical quantities and infer everything else from them through modelling. The core measurements are typically pulse timing derived from optical sensors, movement from accelerometers, and skin temperature at the point of contact. Pulse timing at the skin is not identical to the electrical signal a clinical recording captures, because blood arrival is affected by vessel properties.

Skin temperature likewise is not core body temperature, since it responds to ambient conditions and to blood flow in the skin itself. Every derived quantity, including sleep staging and recovery status, is computed from that narrow set of primary signals rather than measured independently.

Why the score is relative rather than absolute

Devices establish a personal baseline over an initial period and then express each day as a deviation from that individual baseline. This design is sensible, because the underlying quantities vary enormously between people for reasons that have nothing to do with health. It also means a score cannot be compared between two people, since each number refers to a different reference point.

A baseline computed over a rolling window drifts as behaviour changes, so a stable score can coexist with a gradual real change. Understanding the baseline logic explains most of the apparently strange behaviour users notice after a holiday or an illness.

The undisclosed weighting

The step from measured signals to a single score involves weighting those signals through a formula that each manufacturer treats as proprietary information. Because the formula is undisclosed, a user cannot know whether a low score was driven by heart rate, by temperature, or by sleep duration. Manufacturers also revise these formulas, which means a score today and a score from an earlier software version are not strictly comparable.

Independent evaluation is possible only against whatever the device chooses to expose, so validation covers components rather than the composite. This is not deception so much as ordinary commercial practice, and it does place a firm limit on interpretation.

Validation and what it covers

Comparisons against laboratory reference methods have been published for several devices, with better agreement for some measures than others. Agreement on total sleep time is generally reasonable, while agreement on individual sleep stages is substantially weaker across devices. Heart rate accuracy at rest tends to be good and degrades during movement, which is a known limitation of optical measurement.

The composite scores themselves are validated much less often, because there is no external standard for readiness to compare against. A number with no external reference standard cannot be assessed for accuracy in the usual sense at all, only for consistency with its own past output.

Using the output sensibly

The most defensible use is watching trends over weeks rather than reacting to any individual morning value. Single-day changes fall well within the noise of the underlying measurements for most people under ordinary circumstances. There is also a documented tendency for tracking to increase anxiety about sleep in some users, which can affect the thing being tracked.

Devices are not diagnostic instruments and are not regulated as such, other than for specific features that have been separately cleared. Any persistent symptom, or a reading that genuinely worries someone, belongs with a clinician rather than with a customer support forum.