A week of wrist data tied REM sleep to 83 diseases. It can’t show why.

A week of motion data, then years of records

A study published September 17 in PLOS Medicine tracked sleep in 95,559 UK Biobank adults using wrist accelerometers worn for seven days. Researchers applied an algorithm to estimate sleep stages, then compared those measures with hospital diagnoses across 1,049 conditions over a median 8.9 years. journals +1

The strongest signal was for REM, or rapid-eye-movement sleep. Each 47.6-minute increase—the study’s interquartile range—was associated with lower incidence of 83 diseases across 12 categories. The modeled hazard was 26% lower for heart failure, 46% lower for dementia and 80% lower for parkinsonism. Those are relative associations in a population, not changes in any one person’s absolute odds. journals +1

Sleep architecture may matter, but so does duration

REM was not simply a proxy for spending longer in bed: when researchers examined REM as a share of total sleep, a higher share was associated with lower risk of 54 conditions after stringent statistical correction. More deep sleep was associated with lower risk of seven conditions, while irregular sleep and more waking after sleep onset tracked with higher risk for several others. journals +1

Duration still mattered. For 69 of 86 conditions with nonlinear duration patterns, the study’s modeled lowest-risk point fell between six and eight hours. Participants sleeping under five hours had higher associations with 37 conditions compared with the six-to-eight-hour reference group. These results do not establish a new universal sleep prescription; they describe patterns in this cohort. journals +1

Why “more REM” isn’t a health prescription

The evidence has important limits. Sleep was sampled for one week, and the researchers inferred sleep stages from movement rather than measuring brain waves in a sleep laboratory. The paper reports a moderate F1 score of 0.49 for its three-class sleep-stage classifier when benchmarked against polysomnography, the clinical sleep test. Its UK Biobank participants averaged 56 years old and were 96.9% White, narrowing how confidently the results apply to everyone. journals +1

Most importantly, an observational study can’t show whether more REM protects health, illness disrupts sleep before diagnosis, or other factors influence both. Medical News Today notes that reverse causation is plausible for some outcomes, such as depression, where disrupted sleep can itself be a symptom. medicalnewstoday People usually have more REM sleep later in the night, the National Heart, Lung, and Blood Institute explains, so sleep duration and stage mix can overlap. nhlbi The study offers a research lead—not a tracker score to chase or a way to diagnose disease.