REM, rapid eye movement, is literally the name of the sleep stage this story is about
xlibberCC BY 2.0
Illustrative. REM stands for rapid eye movement, the eye activity this study's deep learning model used to help identify that sleep stage, not a photograph of any study participant.
A 95,559 Person UK Biobank Study Found More REM Sleep Tracks With Lower Risk of 83 Diseases
Most sleep advice still boils down to one instruction: get more hours. A new study using data from the UK Biobank suggests the real story is more specific than that. It is not simply how much you sleep, it is which stage of sleep you are actually getting, and REM sleep in particular may matter far more than anyone realized.
- A UK Biobank study of 95,559 adults wore wrist accelerometers for a week, then researchers tracked health records for a median of 8.9 years, using deep learning to break each night into light, deep, and REM sleep.
- More REM sleep tracked with lower risk across 83 of the diseases studied, with the strongest links for dementia (46 percent lower risk) and Parkinson's disease (80 percent lower risk) per typical increase in REM minutes.
- Deep sleep mattered too, but for far fewer conditions, just 7. Sleeping under 5 hours total raised risk of 37 conditions, and 6 to 8 hours total sleep was the sweet spot for most outcomes.
- This is one large observational study. It shows an association, not proof that adding REM sleep prevents disease, and no trial has yet tested deliberately increasing it.
How researchers tracked sleep architecture in 95,559 people
Researchers led by Shengzhi Sun at Capital Medical University in Beijing used data from 95,559 UK Biobank participants who wore wrist accelerometers for seven consecutive days. Instead of relying on people's own memory of how they slept, a deep learning model classified each night into light sleep, deep (slow wave) sleep, and REM sleep, minute by minute. The team then followed everyone's health records for a median of 8.9 years and checked how sleep architecture at the start of the study lined up with who later developed any of dozens of diseases.
Why REM sleep specifically stood out
For every interquartile range increase in REM sleep, about 47.6 minutes, participants had a lower risk across 83 separate disease outcomes. The strongest associations were for two conditions people worry about most as they age: dementia (46 percent lower risk) and Parkinson's disease (80 percent lower risk). Deep sleep mattered too, it was linked to lower risk of 7 conditions, but REM sleep's reach across the disease list was far wider. Total sleep duration still counted on its own terms: sleeping under 5 hours a night raised risk of 37 conditions, and 6 to 8 hours total appeared to be the sweet spot for most outcomes studied.
Honest caveat
This is a single, very large, well designed observational study, association, not proven causation. UK Biobank volunteers tend to be healthier, more often white, and higher income than the general population, so the exact risk numbers may not generalize to everyone. Disease diagnoses were recorded after the week of sleep tracking, so it is possible that early, undiagnosed disease was already disrupting people's sleep, rather than poor sleep causing the disease, even with years of follow up. No randomized trial has tested whether deliberately increasing REM sleep lowers disease risk, so treat this as REM sleep being linked to lower risk, not REM sleep preventing disease.
What this means for you
You cannot easily choose to have more REM sleep the way you can choose to go to bed earlier, but you can influence the conditions that shape it. REM sleep concentrates in the final sleep cycles of the night, closer to morning, so cutting your sleep short on the back end costs you REM specifically, not just total time in bed. Alcohol late in the evening is also a well established REM suppressant, and a consistent sleep and wake schedule protects the full architecture of the night, REM included, rather than just the hours you log. None of that guarantees a lower disease risk on its own, but it is a more specific target than just aiming for eight hours.
It is not simply how much you sleep, it is which stage of sleep you are actually getting
Primary sources
Common questions
Does this mean I should try to get more REM sleep specifically?
The study only shows an association, not a proven way to change your risk. There is no approved way to directly add REM sleep the way you can add total sleep time, and no randomized trial has tested whether deliberately increasing it lowers disease risk. What you can influence are the conditions that support a full night's architecture: a consistent sleep and wake schedule, avoiding alcohol close to bedtime, and not cutting the final hours of sleep short, since REM concentrates later in the night.
Why did REM sleep matter for so many more diseases than deep sleep?
The study does not fully explain why, only that the pattern showed up across a broad list of 83 outcomes for REM sleep versus 7 for deep sleep. REM sleep is linked to memory processing and emotional regulation, and separate research has tied it to brain health specifically, which may be part of why dementia and Parkinson's disease showed the strongest associations. This study measured a statistical pattern, not the underlying biology.
Could poor sleep be an early symptom of these diseases rather than a cause?
That is one of the honest limitations here. Diagnoses were recorded after the week of sleep tracking, but conditions like dementia and Parkinson's disease can disrupt sleep years before diagnosis, so reverse causation cannot be fully ruled out even with a median follow up of almost 9 years. The researchers note this limitation themselves.
Stay Curious.
Better health isn't found in a single article. It's built through curiosity, evidence, and small discoveries over time. Join thousands exploring longevity, gut health, recovery, nutrition, and the tools that truly make a difference.