What Scientists Just Discovered About REM Sleep
A large new study has linked more REM sleep with a lower risk of 83 diseases—including dementia, heart failure and Parkinson’s disease. The finding is striking, but the headline needs one important qualification: the study found an association, not proof that REM sleep directly prevents 83 diseases.
Researchers analyzed sleep and health data from 95,559 UK Biobank participants. The adults wore wrist accelerometers for seven days, and researchers used a deep-learning system to estimate sleep stages, total sleep time, sleep irregularity and wakefulness after falling asleep. Participants were then followed for a median of 8.9 years.
The result offers a more detailed picture of sleep than simply asking whether someone slept for seven or eight hours. It suggests that sleep architecture—the pattern of light, deep and REM sleep—may be connected to long-term health in different ways.
What the study found
The researchers examined associations between sleep patterns and 1,049 disease outcomes. Across the analysis, variations in sleep were linked with 156 disease incidences.
The most widely reported finding was that greater REM sleep was associated with a lower risk of 83 diseases. For every interquartile-range increase in REM sleep—about 47.6 minutes—the study reported a lower risk of conditions including heart failure, dementia and Parkinson’s disease. The reported hazard ratios were 0.74 for heart failure, 0.54 for dementia and 0.20 for Parkinson’s disease.
The findings were not limited to REM sleep. More deep sleep was associated with a lower risk of seven conditions, including type 2 diabetes and major depressive disorder. Greater sleep irregularity and more wakefulness after sleep onset were associated with higher risks of several conditions, including anxiety and substance-use disorders.
Total sleep duration also showed a non-linear pattern. For many outcomes, the lowest-risk range was concentrated between six and eight hours. Participants sleeping fewer than five hours showed the broadest pattern of elevated vulnerability, with higher risk reported for 37 conditions compared with the six-to-eight-hour reference group.
These results do not mean that everyone should try to force exactly 47 extra minutes of REM sleep. They also do not mean that an individual’s disease risk can be calculated from a wearable’s sleep-stage estimate.
Why REM sleep matters
REM stands for rapid eye movement. During this stage, brain activity is relatively active, breathing can become irregular and vivid dreaming is common. REM sleep usually becomes more prominent later in the night, which is one reason cutting sleep short may remove a disproportionate amount of later-night REM sleep.
Sleep is not a single inactive state. The body moves through repeating cycles that include non-REM stages and REM sleep. The National Heart, Lung, and Blood Institute notes that sleep supports brain function, memory, metabolism, immune activity and cardiovascular health. It also explains that heart rate and blood pressure change across sleep stages, including increases during REM sleep and awakening.
Scientists are still working out exactly what each stage contributes. REM sleep has been linked in previous research to memory processing, emotional regulation and brain development, but a biological role does not automatically explain every association seen in a population study.
The study’s biggest limitation
The new research was observational. Researchers measured participants’ sleep patterns and later compared health outcomes. They did not assign people to sleep more or less REM sleep, and they did not manipulate sleep in a controlled experiment.
That means the study cannot establish that more REM sleep caused the lower disease risk. It is possible that health affects sleep rather than the other way around. This is known as reverse causation.
For example, an early stage of a neurological or cardiovascular condition might disrupt sleep before the condition is formally diagnosed. The researchers excluded disease events occurring during the first six months after sleep measurement to reduce this problem, but that step cannot eliminate it entirely.
Other factors may also influence both sleep and health. Age, medications, mental health, work schedules, socioeconomic conditions, physical activity, alcohol use, smoking, obesity and existing sleep disorders can all matter. Researchers adjusted for many demographic, lifestyle and environmental variables, but statistical adjustment cannot account perfectly for every hidden difference between participants.
The sleep-stage estimates were not the same as a sleep-lab test
The participants did not undergo full polysomnography, the laboratory test that records brain waves, eye movements, muscle activity, breathing and other signals. They wore wrist accelerometers instead.
The study used a deep-learning system called SleepNet to estimate REM, non-REM and wake periods from movement data. The system had been trained and tested against laboratory sleep studies, but its agreement with polysomnography remained moderate. The paper reports an F1 score of 0.49 for the REM/non-REM/wake classification and notes systematic differences between accelerometer-derived estimates and laboratory measurements.
This does not make the study useless. Wrist-based measurement allowed researchers to study nearly 100,000 people in everyday conditions, which would be extremely difficult with repeated laboratory sleep tests. But it does mean that the study’s REM figures should be understood as algorithmic estimates, not exact measurements of brain-defined REM sleep.
What the “83 diseases” number really means
The number is impressive partly because the researchers tested a very large set of outcomes. They examined more than 1,000 disease phenotypes across multiple body systems and looked for statistically significant associations after correcting for multiple comparisons.
That approach can reveal patterns that a study focused on one disease might miss. It can also produce a headline number that sounds more definitive than the underlying evidence.
“Associated with a lower risk of 83 diseases” does not mean that REM sleep was proven to prevent each condition. It does not mean that all 83 associations are equally strong. It does not mean that increasing REM sleep would produce the same benefit in every person. And it does not mean that REM sleep replaces exercise, medical treatment, vaccination, blood-pressure control or other established ways to reduce disease risk.
The best interpretation is more measured: in this cohort, people with higher estimated REM sleep tended to experience fewer diagnoses across a broad group of outcomes during follow-up. That pattern deserves further study.
Should you try to increase your REM sleep?
The study does not provide a simple REM-sleep exercise. People cannot reliably command themselves to spend a specific number of minutes in REM sleep, and consumer sleep trackers may estimate sleep stages differently from one another.
The more useful message is to protect the conditions that support normal sleep:
•Give yourself enough time in bed instead of repeatedly cutting the final hours short.
•Keep your sleep and wake times reasonably consistent.
•Make the bedroom quiet, dark and comfortable.
•Avoid treating alcohol as a sleep aid; it can disrupt sleep architecture even when it makes you feel drowsy.
•Limit caffeine late in the day if it interferes with falling asleep.
•Seek medical advice for loud snoring, gasping, persistent insomnia, severe daytime sleepiness or unusual movements during sleep.
The National Heart, Lung, and Blood Institute emphasizes that inadequate sleep over time can affect thinking, memory, metabolism, immune function and cardiovascular health. Those established effects are already enough reason to take sleep seriously.
The real discovery is that sleep quality is multidimensional
The most valuable part of this research may not be the number 83. It is the reminder that sleep duration alone does not describe the entire night.
Two people may both report sleeping seven hours, while one has relatively stable sleep and the other spends much of the night awake. One may have a different distribution of REM and deep sleep. A wearable may not measure those stages perfectly, but the distinction is scientifically important.
Sleep health is not about chasing a perfect score. It is about creating a regular opportunity for the brain and body to complete their normal cycles. The new study suggests that REM sleep may be one part of that picture, but it does not prove that REM sleep is a shield against dozens of diseases.
So yes, your sleep may be helping protect your long-term health. But the evidence does not say that REM sleep alone can prevent 83 diseases. What scientists have discovered is more useful than a miracle claim: healthy sleep appears to be a whole system, and both its quantity and its architecture may matter.
