Sleep keeps showing up as one of the clearest predictors of chronic disease, cognitive decline, and burnout. Still, in everyday medicine, it gets measured unevenly, defined in fuzzy terms, and too often left out of real care planning.
At the same time, healthcare is sprinting toward a setup built on constant streams of information, prediction, and AI-supported choices. Remote monitoring, digital twins and therapeutics, virtual care platforms, all of it runs on the assumption that more data will translate into better outcomes. The problem is that a lot of these tools are expanding fast without a dependable grasp of one of the most basic drivers of health: sleep.
Sleep threads through cardiovascular function, metabolism, immune activity, and mental health. It happens nightly, which means it can produce a steady flow of information that, measured well, could tell us a lot about near-term changes and long-range risk.
Yet it still isn’t treated like a vital sign.
A foundational signal that’s still missing
Unlike blood pressure or glucose, sleep rarely gets measured with the same regularity or discipline outside of clinic-based contexts. It’s commonly self-reported, checked only now and then, or guessed through consumer tools whose accuracy can swing widely. For a system that says it wants prevention and earlier intervention, that inconsistency leaves a major blind spot.
AI is often presented as the fix for messy, scattered health data. In reality, it can only work with what it’s fed. When key inputs are missing, uneven, or not well understood, AI tends to multiply the mess instead of cleaning it up. Sleep is an easy place to see that pattern.
Wearables have made sleep tracking available to huge numbers of people, but availability is not the same as reliability. Two devices can give noticeably different sleep results for the same person on the same night. Total sleep time, deep sleep, wake-after-sleep-onset, these numbers change depending on the device, the measurement method, and the assumptions baked into the algorithm. That variation matters because it changes what people think is happening, and what a system might recommend doing next.
Once those inconsistent signals start driving clinical interpretation or AI-generated recommendations, variability becomes a safety issue, not just a technical quirk.
Then there’s adherence. Plenty of users don’t wear their devices every night, or they stop wearing them for stretches of time. That creates holes that make it harder to spot patterns. In clinical settings where continuity is the whole point, missing data weakens confidence in whatever conclusions follow.
When AI scales the wrong signals
More and more, AI is being laid over these sleep signals to spit out insights and next steps. On paper, that sounds like real progress, continuous data paired with intelligent interpretation. In practice, it can create a different problem: precision that looks real but isn’t.
AI can produce confident, detailed outputs even when the underlying data isn’t solid enough to support them. With sleep, that can mean advice that feels personal while resting on shaky measurement and thin, non-validated history. The old rule still applies, garbage in, garbage out, even when the “out” comes wrapped in polished AI language.
In healthcare, data-driven decisions have consequences. If the data are patchy or inconsistent, and the system interpreting them doesn’t seriously account for uncertainty, you end up with confidence that outpaces accuracy. That mismatch can mean missed warning signs, interventions that come too late, or reassurance that shouldn’t have been given.
Sleep makes this problem hard to ignore because it’s universal and continuous. If health AI can’t read something this basic with care, it pushes a bigger question to the surface: how well will these systems handle signals that are more complicated, less direct, or harder to observe?
There’s also a widening gap between “personalized” as marketing and “personalized” as true insight. Many products present outputs as tailored to an individual, but they’re often built on broad patterns instead of deeply contextual, long-term data. The guidance sounds specific, but ends up generic.
As AI gets woven into care delivery, that difference stops being academic and starts being crucial.
What it would take to treat sleep like a vital sign
If sleep is going to operate as an actual vital sign, the field has to rethink measurement, validation, and how sleep information is used in care.
First, measurement has to shift away from convenience alone and toward consistency and clinical usefulness. Not every signal captured in a consumer setting is fit for clinical decision-making. If sleep is going to be part of care pathways, clearer standards and steadier measurement practices will be needed.
Second, sleep has to be read over time. A single night, or even a short run of nights, doesn’t say much. The value shows up in patterns, in trends, in variability, in how sleep responds when something changes. Systems built for outcomes, not just engagement, have to reflect that.
Third, interpretation has to stay tied to context. Sleep isn’t a standalone phenomenon. It shifts with physical health, mental health, environment, and behavior. If you treat sleep data as an isolated number instead of connecting it to what’s happening in a person’s life and body, you miss the point.
Finally, the healthcare environment needs to get serious about which signals are so fundamental they deserve deeper investment, and which ones can be approximated without much harm. Sleep is looking more and more like it belongs in the first group.
Calling sleep a “missing vital sign” isn’t new, but the urgency has changed because AI is accelerating how data gets pulled into care. That speed is also exposing just how fragile some of the underlying inputs are. Sleep sits right in the middle of the tension: everyone agrees it matters, but it’s still measured unevenly and folded into care poorly.
Closing the gap won’t be simple. It means better measurement, tougher validation, and a more disciplined approach to turning data into insight. But if healthcare genuinely wants better long-term outcomes, it can’t keep treating sleep as an afterthought. Weak foundational signals make everything built on top of them weaker too.
Photo: viridian1, Getty Images
Colin Lawlor is a globally recognized leader in sleep science and digital health, with over 30 years of executive experience, 15 of them dedicated to redefining how we measure, understand, and improve sleep. As founder and CEO of Sleep.ai and former executive at ResMed, Colin has pioneered some of the most scientifically validated sleep technologies in both consumer and clinical markets, including the first contactless, sonar-based sleep tracking systems.
Under his leadership, Sleep.ai has transformed over 800 million hours of consumer sleep data into AI-driven health insights and coaching that help partners across health, wellness, and technology deliver personalized, science-backed sleep solutions. His team’s innovations have powered more than 100 scientific publications and led to the first-ever government-reimbursed digital sleep intervention in Germany. A passionate advocate for using sleep to prevent chronic conditions like heart disease, dementia, and diabetes, Colin connects the dots between science, technology, and real health impact.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
