Nearly 96,000 people wore a wrist device for a week, then went home and got on with their lives. Nine years later, researchers checked in on what had happened to their health, and found something that goes well beyond the usual advice to just get more sleep: which stage of sleep people spent their nights in mattered for 83 separate diseases.

The study, led by Jingsong Luo at Peking University and published September 17, 2026, in the open-access journal PLOS Medicine, drew on 95,559 participants from the UK Biobank, all of whom wore accelerometers capable of distinguishing REM sleep, deep sleep, and light sleep across seven consecutive nights. Researchers then tracked those participants' health records for a median of roughly nine years, cross-referencing sleep architecture against more than 1,000 distinct diseases.

What actually separates REM from deep sleep

Most public health messaging around sleep focuses on duration, get seven to nine hours, avoid all-nighters, and call it done. This study cuts into finer detail, examining what happens during sleep rather than just how long it lasts. REM sleep, the stage associated with vivid dreaming and rapid eye movement, and deep sleep, also called slow-wave sleep, serve different physiological functions and cycle through the night in a predictable pattern that shifts as sleep progresses.

Distinguishing between them typically requires either a sleep lab or, as in this study, a wearable device sophisticated enough to detect the movement and physiological signatures unique to each stage. That's part of what makes a dataset of this size notable: getting objective, stage-specific sleep data from nearly 96,000 people, rather than relying on self-reported sleep quality, is a logistical feat most sleep research simply can't match.

The scale of the REM sleep finding

The headline result was stark. People who spent more time in REM sleep had a lower risk of 83 different diseases, according to ScienceDaily's coverage of the findings, spanning conditions as varied as heart failure, dementia, Parkinson's disease, Alzheimer's disease, atrial fibrillation, and multiple sclerosis. That's not a narrow cluster of related conditions; it cuts across cardiovascular, neurological, and autoimmune categories simultaneously, according to Popular Science's reporting, which noted the 83 diseases spanned 12 distinct disease categories.

Deep sleep showed a more targeted, though still meaningful, protective association. More deep sleep was tied to lower risk of seven conditions specifically, including type 2 diabetes, major depressive disorder, and sleep apnea, according to HealthDay's coverage of the study. The distinction between REM's broad protective signal and deep sleep's narrower one suggests the two sleep stages may be doing genuinely different physiological work, rather than simply serving as two markers of the same underlying "good sleeper" trait.

Why five hours turns out to be a meaningful cutoff

Total sleep duration still mattered independently of sleep architecture. The six-to-eight-hour range emerged as the sweet spot across the analysis, with people sleeping less than five hours facing sharply elevated risk. According to Popular Science's reporting, people getting under five hours nightly showed 37 of the 41 significant health risks identified in the full analysis when compared against the six-to-eight-hour reference group, including heart failure, type 2 diabetes, and chronic obstructive pulmonary disease.

Sleeping more than eight hours wasn't a clean win either. That group showed fewer of the elevated risks associated with short sleep, but carried a higher risk specifically for major depressive disorder, according to Popular Science, a reminder that the relationship between sleep duration and health doesn't move in one direction indefinitely; more isn't uniformly better past a certain point.

What happened with fragmented, irregular sleep

Beyond how long people slept and which stages they favored, the researchers also examined sleep consistency and interruption. People whose sleep duration varied more from night to night, rather than following a stable, predictable pattern, showed higher risks of anxiety and major depressive disorder, according to Popular Science's reporting on the findings.

Nighttime wakefulness carried its own distinct risk profile. More time spent awake after initially falling asleep was associated with higher risks of several conditions, including substance dependence and osteoarthritis. That's a notably wide-ranging pair of associations for a single sleep metric, one metabolic and inflammatory, the other behavioral, suggesting fragmented sleep may stress the body through more than one physiological pathway at once.

Why this matters more in China specifically, but not only there

Luo's research team operates within a national context where sleep health has become an explicit public health priority. According to ScienceDaily's coverage, sleep quantity and quality is a stated focus of China's National Health Commission, and the most recent 2026 China Sleep Health White Paper found that the average surveyed Chinese person sleeps just 6.97 hours per night, with a meaningful share of the population falling short of recommended durations.

That national policy backdrop helps explain why a Chinese-led research team took on a study of this scale using UK Biobank data, a globally recognized, richly annotated dataset that allows for exactly this kind of large-scale, long-follow-up disease association work. The findings themselves, though, aren't specific to any one country's sleep habits; the underlying biology of REM and deep sleep functions similarly across populations, even if average sleep duration and quality vary by region.

What this means for how people think about sleep

The practical takeaway from a study like this isn't simply "sleep more," a message most people have already absorbed without much behavior change to show for it. It's that the internal structure of a night's sleep, how much time is spent cycling through REM versus deep sleep, and how consistently that pattern repeats night to night, carries its own independent health signal separate from raw duration.

That's a harder message to act on directly, since most people can't consciously control how much REM or deep sleep they get the way they can, in theory, control what time they go to bed. But it does help explain a stubborn puzzle sleep researchers have long grappled with: why two people who each sleep seven hours a night can end up with meaningfully different long-term health outcomes. The answer, at least according to this study's nine-year follow-up of nearly 96,000 people, may lie less in the clock and more in what's actually happening inside those hours.