Regarding sleep stages, what are consumer devices less accurate at distinguishing based on input?

Distinct stages of Non-REM and REM sleep.

Research comparing consumer trackers to the clinical gold standard, PSG, shows that while non-invasive sensors are very proficient at capturing broad metrics, they struggle with granularity. Consumer devices excel at basic duration metrics, specifically determining Total Sleep Time (how long you were actually asleep) and identifying fragmentation events like Wake After Sleep Onset (WASO). However, the algorithms based primarily on actigraphy and heart rate often fail to reliably differentiate the specific neurological qualities defining the boundaries and durations of deep Non-REM stages versus REM sleep, which requires reading specific brainwave patterns.