Counting steps that go up

Stairs change every part of the walk a hand-worn step counter reads. The step is shorter and harder, the cadence is slower, the trunk rises and falls with each one, and a hand may leave the swing altogether to take the rail. None of it is inferable from a result measured on flat ground, which is why the expert consensus on step-count validation asks for stairs to be tested in its own right rather than assumed from level walking.

Across complete flights the Ultrahuman Ring came within 6.5% of the hand count, and three in four of them landed within a tenth of it. That sits inside the 10% to 15% error the same consensus treats as acceptable for general use. The direction is the one the ring shows on a treadmill too: it reads low, by about four steps in every hundred, rather than inventing steps that were never taken.

Where stairs sit against ordinary walking is the useful part. They come in below a treadmill walk at 95.2% and above a walk outdoors at 91.8%. A stairwell is an indoor, controlled surface — the riser height is fixed and the flight repeats — and that is the same property that puts every indoor condition on this page above its outdoor counterpart. What it costs against a belt is the landing at every turn, and the hand that reaches for the rail.

The published work on stairs is thin, and what there is does not agree with itself. One study put eight trackers on a staircase against a video count of every step, found no general difference between them there, and closed by advising caution in reading any step count taken on stairs; another, testing five sensor placements on the body, found that none of them recorded stair climbing accurately. What the field does agree on is that stairs has to be measured rather than carried over from level ground.