[hnuser123456]

Maybe we can require people take selfies in sunlight with a known camera sensor and white balance and exposure to accurately gauge their natural skin color, then have a corrective calibration factor sent to the watch's oximeter.

[etrautmann]

I think there are different issues between cameras and pulse ox. I would guess that the SNR on a pulse ox is decreased by the additional attenuation of the LED illumination being sent in and scattered back out. It may be possible to compensate for this with increased LED brightness, but it would depend on the total degree of signal attenuation whether this is feasible given the power budget.

For an imaging sensor, the sensors are designed to be sensitive at different wavelengths and in different dynamic ranges, with some historical set points determined by major manufacturers being located in Japan. While darker tones do imply less collected light, this seems like a very solvable problem with in-camera image processing.

Another interesting question is around non-invasive neural sensors. Many EEG electrodes work relatively poorly in thick or curly hair, and there's been several efforts to develop EEG electrodes appropriate for natural hair [0]

[0] https://pubmed.ncbi.nlm.nih.gov/33019375/