Wearables

Wearable devices — watches, rings, bands and the increasingly varied set of sensors people attach to themselves — occupy an interesting position: genuinely useful for some things, systematically overinterpreted for others, and marketed in a way that blurs the line.

What they measure well is movement and timing. Step counts, activity duration, workout detection and heart rate during steady exercise are reasonably accurate on current devices, and the trends over weeks are meaningful. Sleep timing — when you fell asleep and woke — is also decent.

What they estimate rather than measure includes energy expenditure, sleep stages, stress, and the composite readiness or recovery scores that many devices now lead with. These are models built on the signals above, and they can be substantially wrong for an individual. Treating an estimated calorie figure as a budget, or letting a readiness score determine whether you train, is putting more weight on the number than it can carry. The trend over a month is informative; a single day's score is not.

There is a specific failure mode worth naming, because it is common: the device increases anxiety rather than health. Someone sleeps fine, sees a poor score, and feels tired. Someone with a tendency toward health anxiety acquires a continuous stream of ambiguous physiological data. If a wearable is making you worry more than it is helping you act, taking it off is a legitimate and underused option.

The medical-adjacent features are the exception and deserve credit. Irregular rhythm notification in particular has a genuine evidence base and has led to real diagnoses. Blood oxygen and temperature sensing are more variable. The framing to hold is that these features can prompt you to see a doctor; they do not substitute for one, and none of them diagnose anything.

Practical buying considerations: battery life determines whether you actually wear it overnight, which determines whether the sleep data exists at all; and data portability determines whether years of history is yours or the vendor's. Check the export path before committing to an ecosystem.

Health data is among the most sensitive there is. Read what is shared, with whom, and whether an insurance or employer programme is involved.