Inst. for Sleep
Devices > Wearables

Wearable Validation vs. PSG

We aggregate validation studies comparing commercial sleep trackers against the gold-standard Polysomnography (PSG). Data indicates high reliability for Total Sleep Time (TST), but significant failure rates in multi-stage classification (Deep vs. REM).

The Limitations of Optical Sensors (PPG)

Consumer devices rely on actigraphy (movement) and photoplethysmography (PPG for heart rate). They do not measure brainwaves (EEG), which is the clinical requirement for staging sleep. Therefore, all "Deep Sleep" and "REM" scores on wearables are algorithmic inferences based on autonomic nervous system proxies (HRV and movement).

Aggregate Accuracy Findings (2023-2024 Studies)

  • Sleep/Wake Detection: 88-92% accurate. Wearables are generally good at knowing if you are asleep.
  • Total Sleep Time (TST): Tends to overestimate TST by 15-30 minutes due to counting motionless awake time as sleep.
  • REM Staging: ~60-70% agreement with PSG. Often confuses REM with Light Sleep (N1/N2).
  • Deep Sleep (N3) Staging: ~50-60% agreement with PSG. Highly variable among individuals.

Clinical Consensus

Do not alter medical treatment or diagnose insomnia based on wearable sleep staging data. Use these devices strictly for macro-trends (e.g., consistency of bedtime, total hours in bed).