For decades, the public health advice on sleep has been a single number: seven to eight hours. But a new study by Ultrahuman, developed with Stanford University’s Snyder Lab, Bangor University’s Hans-Peter Kubis, and La Trobe University’s Driller Lab, suggests that might only be half the story.
The bigger lever might be consistency — going to bed at roughly the same time each night, with the study highlighting potential metabolic benefits that could offer a myriad of health and longevity benefits.
The study is currently undergoing peer review on medRxiv.
The study
The Ultrahuman study analyzed 227,860 nights of data from 5,859 adults across 100 countries — one of the largest real-world, free-living sleep and metabolism studies ever conducted.
What makes the dataset unusual is that sleep and glucose were captured simultaneously, in everyday life, on the same people. Most sleep studies rely on a single sensor or take place in a lab. This study combined sleep architecture from the Ultrahuman Ring with continuous glucose data from the M1 CGM — one of the longest-running consumer CGM programmes in existence.
The work builds on earlier Ultrahuman studies in Nature, with further research underway with Mayo Clinic, Harvard, Pfizer and Unilever.
What the data showed
When the research team applied machine learning to identify what most separated the metabolically healthy from those approaching dysfunction, one signal stood out: sleep timing consistency.
A difference of just 10–15% in day-to-day variability in bedtime was enough to separate users with elite, athlete-level glucose control from those tracking toward pre-diabetes.
Poor sleep revealed a pattern of signals associated with a body under stress:
- Overnight glucose averaged 6.4 mg/dL higher
- Time spent in a healthy glucose range fell by 13.9%
- Sleep heart rate ran 9 bpm higher
- HRV dropped by 7ms
- Minimum nocturnal heart rate was 6 bpm elevated
Despite identifying as well, the overnight physiology of users with the poorest sleep consistency drifted toward pre-diabetic territory — and the strongest predictor was simply how irregular their bedtime was.
Why this matters
Chronically elevated blood sugar — even below the diabetic threshold — is linked to impaired cognition, disrupted appetite regulation, and a higher long-term risk of type 2 diabetes and cardiovascular disease. Those are two of the leading causes of preventable death globally.
What this dataset suggests is that sleep timing irregularity may be quietly pushing otherwise healthy people in that direction, years before anything would show up in a clinical test.
The incredible thing is that people don’t need a prescription, gym membership, or major lifestyle changes to see a huge benefit in their health. They just need to go to bed at roughly the same time each night.
How health ecosystems can help
This study was only possible because two sensors were measuring the same person at the same time. That’s the underlying argument for an integrated health ecosystem — being able to see how systems in the body interact, rather than looking at sleep, glucose, heart rate and recovery in isolation.
It also opens up a second possibility: once the relationship between sleep patterns and metabolic markers is mapped at population scale, it becomes possible to infer metabolic insights for people who don’t wear a CGM. The sensor data trains the model; the model can then read signal from the sensors people already have.
Regulators are moving in the same direction. The FDA is explicitly targeting connected health devices as tools for managing chronic disease, including pre-diabetes.
Conclusion
“We’ve long suspected that circadian consistency was upstream of metabolic health — but the scale of this dataset allows us to see it with unusual clarity for the first time,” said Ultrahuman CEO Mohit Kumar. “What’s striking is how accessible the intervention is. The data shows that going to bed at roughly the same time each night is one of the most powerful metabolic levers most people aren’t using.”
