Every sleep tracker eventually gives you the same advice: make your room cooler, ditch the screens, stick to a bedtime, drink less caffeine and alcohol, and get more light in the morning.
It is good advice. And for most people, it moves the number.
But for many people who take insulin, it doesn't. They do everything the app suggests and yet they still wake up with a low score and a body that feels like it never fully went under.
The reason is not on the sleep-hygiene checklist. It’s in their glucose.
The link between glucose and sleep architecture is not a marketing idea. It shows up in the literature, and it is worth being precise about what that literature does and does not say.
A pediatric study — and it is important that it was children, 15 with type 1 diabetes and 15 controls — found that a rapid drop in glucose, faster than about 25 mg/dL per hour, was associated with awakenings, and that profound hypoglycemia increased slow-wave sleep and EEG delta activity (Pillar G et al., J Pediatr 2003;142(2):163-168, PMID 12584538).
The body isn’t resting during these events. It is responding to them.
Across both pediatric and adult type 1 diabetes, higher A1c has been associated with less slow-wave, or deep, sleep, and overnight hyperglycemia above roughly 154 mg/dL has been linked to lower overnight melatonin (Farabi SS, Diabetes Spectrum 2016;29(1):10-13, PMID 26912959).
And in a study of 67 insulin-treated women with type 2 diabetes, nocturnal glucose indices independently predicted next-day heart rate variability power, explaining roughly half the variation (R²=0.51, p=0.008; Klimontov VV et al., SpringerPlus 2016;5:337, PMID 27066358).
It would be easy to assume all of this is due to “high glucose is bad for sleep.” However, that isn’t quite the finding, and the distinction matters.
What disrupts the night is glycemic variability — the rate of change, the swings — more than any single number held steady. Instability is what keeps the autonomic nervous system working when it should be recovering. A moving glucose is a body that is responding. And recovery doesn't happen during a response.
This is why the sleep-hygiene checklist falls short for this group. A blackout curtain does not flatten a 2 a.m. swing, and no amount of morning sunlight changes the rate of an overnight decline.
Here is the part our own research made vivid. Among the patients we call High-Concern Strugglers, 81% purposefully go to bed high to guard against an overnight low — and that same group is the most likely to wake above 150 mg/dL (DQ&A PWD Segmentation and JTBD Report, January 2026, Slide 52).
Read that as a human decision, not a clinical error. These are people choosing a high night on purpose because a high night feels survivable and a low night feels dangerous. They are trading deep sleep for safety. The literature above explains, mechanistically, why that trade shows up the next morning as a wrecked sleep score. They are not doing sleep wrong. They are managing a real risk the only way the tools allow them to.
I want to be careful here, because the honest version is more useful than the convenient one. No published study shows that automating overnight glucose restores sleep architecture. We cannot claim Luna fixes deep sleep, because no one has earned that claim yet.
What we can say is directional and grounded: glycemic instability disrupts the autonomic recovery that sleep is for, and a meaningful number of insulin users are deliberately accepting worse nights to stay safe.
Sleep-only automation, automated insulin delivery that works while you sleep and steps back when you wake, is built for exactly that window and exactly that trade. The aspiration is to make the safe and restful choices the same choice, instead of forcing people to pick. Whether it improves the number on the ring is a hypothesis we intend to test, not a result we get to announce.
Some sleep problems can be fixed by a cooler room. This isn’t one of them.
Nights protected. Days yours.