Performance of wearable light sensors for measuring photopic and melanopic illuminance under laboratory and free-living conditions

Abstract Wearable sensors are commonly used to study the effects of free-living light exposure on physiological outcomes, however rigorous validation of their performance has been limited. To address this gap, we quantified the accuracy and precision of light measurements from four commonly used wearable sensors (LYS, ActLumus, GENEActiv, and ActiGraph wGT3X-BT). Eight devices from each manufacturer (32 total) were compared to a NIST-calibrated spectrometer under LED-generated cool (peak ~460 nm) and warm (peak ~600 nm) white light at indoor intensities (2-1000 lux) and broad-spectrum sunlight (5 k-100 k lux). We found linear responses for all sensors between 50-20 k photopic lux, irrespective of spectrum, but detection ranges varied: ActLumus (2-100 k lux), LYS (2-60 k lux), GENEActiv (2-40 k lux), and ActiGraph (50-20 k lux). Accuracy – percent difference to criterion – also differed: ActLumus (+3.3 ± 13.4%, mean ± SD), LYS (+19.5 ± 42.0%), and below -20% for GENEActiv and ActiGraph. Inter-device variability was consistently lowest for ActLumus, and melanopic illuminance was less accurate with LYS than ActLumus, particularly in warm indoor light. Laboratory findings were compared to one-week free-living light measurements in 21 individuals concurrently wearing the most and least accurate sensors in randomized positions on the non-dominant wrist. The mean 24 h lux per participant was correlated (r = 0.91, p<.001) but lower for ActiGraph (77.3 ± 68.5 lux) than ActLumus (515.0 ± 436.0 lux; p<.001) with greatest differences ≤100 lux (p<.001), consistent with the laboratory results. Thus, differences in illuminance range and accuracy can lead to large disparities in free-living measures across manufacturers, suggesting a need for greater technical standardization.