Blood oxygen (SpO₂) across devices — Apple, Garmin and Oura compared
SpO₂ is where a units quirk and a sampling difference get mistaken for one device being more accurate than another.
The units mismatch
Apple stores SpO₂ as a fraction between zero and one — a reading shown as ninety-seven percent is stored as the fraction just under one — while Garmin, Oura and a pulse oximeter report a whole-number percent. Line the two up without converting and one series sits near one and the other near a hundred, a hundredfold apart, which looks like catastrophic disagreement and is only a scale. The fix is conversion, not distrust.
The sampling mismatch
Even in the same units, devices sample SpO₂ differently. Some take occasional spot checks when you hold still; others measure continuously through the night. So one device gives you a handful of daytime points and another a dense overnight curve, and their averages describe different populations of readings. Comparing a spot-check average to an overnight average is comparing two different things, not grading two sensors.
What comparison holds
Within one device and one sampling mode, SpO₂ trends and overnight dips are meaningful. Across devices, convert the units first, then compare like sampling to like — overnight to overnight — and treat consumer SpO₂ as a coarse signal on all of them, useful for change rather than for a precise number.