A large prospective study published in PLOS Medicine analyzed wrist tracker data from over 95,000 UK Biobank participants to explore how real-world sleep patterns influence the risk of nearly 1,100 diseases. Using machine learning to estimate sleep stages from wrist accelerometers, researchers followed participants for a median of 8.9 years and found that sleep architecture — not just total sleep time — plays a major role in long-term health.

REM sleep emerged as the standout protector, with each ~48-minute increase linked to significantly lower risk across 83 diseases. Deep sleep also showed benefits, particularly for metabolic and mental health conditions. On the flip side, irregular sleep and frequent nighttime waking were tied to higher rates of anxiety, depression, and substance use disorders.

By the Numbers:

• REM sleep: lower risk for 83 diseases, including Parkinson's (HR 0.20), MS (HR 0.48), and dementia (HR 0.54)
• Deep sleep: lower risk for 7 conditions, including type 2 diabetes (HR 0.89) and major depressive disorder (HR 0.86)
• Less than 5 hours of sleep: increased risk across 37 conditions
• 6–8 hours/night: associated with lowest disease risk across 69 conditions

Why it matters: These findings reinforce that sleep quality is a modifiable health factor with wide-ranging consequences — and wearable technology may be a scalable tool for tracking it at a population level.