The following is a guest article by Colin Lawlor, CEO at Sleep.ai
Most digital mental health products slot sleep into the “bedtime content” bucket, creating meditation tracks, breathing exercises, or wind-down playlists. The premise is that sleep sits downstream from mood: calm someone down, and sleep arrives. But the evidence points the other direction, too.
A Journal of Affective Disorders meta-analysis reported that people who were not depressed but did have insomnia faced about double the risk of later developing depression. Affects this significant don’t show up often in psychiatric epidemiology, and this one has stayed consistent through years of replication.
A Signal That Keeps Getting Stronger
The relationship runs both ways, and it runs deep. Poor sleep predicts future mental health symptoms. Mental health symptoms disrupt future sleep. The mechanisms are physiological, not only psychological. Chronic sleep loss changes how the brain regulates emotion, how the body tolerates stress, and how the experiences of the day get consolidated during the night. Those changes turn into measurable risk for depression and anxiety.
This isn’t hanging on a single headline study. A meta-analysis of more than 172,000 people arrived at essentially the same takeaway and a 2024 umbrella review pooling 29 systematic reviews did too. An APA meta-analysis of 154 sleep-loss studies found a clear dose-response pattern: less sleep, worse mood, and higher anxiety, step by step along the curve. For anyone delivering or building mental health care, sleep is one of the most reliable indicators of how someone is doing day to day. Yet in many digital mental health products, it’s functionally invisible.
Why Mental Health Products Haven’t Incorporated Sleep
The teams building in mental health know the research, but have faced structural obstacles in implementation.
First comes the fact that sleep is hard, particularly the data. Patients are using apps, watches, rings, and even mattresses to track – but figuring out how to use the differing (often conflicting) data from each of these devices can be really challenging. Then you have accuracy issues – many of these devices have been trained on populations with no sleep issues, and they can be way off the mark when it comes to populations suffering from sleep issues.
Secondly, the engagement ceiling. In clinical trials of depression and anxiety apps, about a third of users are gone by follow-up, and real-world drop-off is often worse. One panel-based analysis found that only around 10% of users were still active a week after download. Conversation, content, and coaching can help significantly improve symptoms for those who stick with them, but that group shrinks quickly. If digital mental health is going to make a meaningful jump forward, it has to come from the product working smarter, not from trying to squeeze more usage out of people. That lever has already been pulled.
The third issue is design. Apps built around chat, lessons, and coaching haven’t had an easy path for bringing in an objective physiological input. Sleep tracking, for a long time, meant hardware that many members wouldn’t wear. Product teams responded in a predictable way: they treated sleep as nearby wellness material and kept it outside the core workflow that drives engagement and care.
What Integration Looks Like Now
Things are beginning to improve. A 2025 meta-analysis in the Journal of Clinical Sleep Medicine reported that current consumer sleep trackers have reached accuracy levels that clinicians and researchers can work with. Smartphone-based measurement, using motion, ambient audio, and on-device AI, has been improving alongside it. Measuring sleep through the phone someone already owns is now something you can actually ship.
Some digital mental health platforms are already building around this shift. Meru Health weaves heart rate variability biofeedback into its CBT-based depression and anxiety program, using a cardiac signal as an objective anchor alongside therapist conversation. HearMe is working sleep intelligence into a peer support experience. Different physiological inputs, same basic move: let objective measurement shape how the product responds to the user.
Payers are moving as well. German statutory insurers, covering about 25 million members, have begun reimbursing digital sleep coaching programs as preventive care, launching the first reimbursable digital sleep intervention at that scale. When payers start paying for sleep as prevention, it’s a sign the evidence has pushed sleep out of the “wellness extra” category.
Where the Bar Should be Set
As more teams try to integrate sleep, it helps to be clear on what counts as real integration versus a new feature label.
The first question is whether the sleep signal is genuinely objective. Mood journals and sleep diaries don’t line up well with measured physiological state. If a system depends on users remembering, estimating, and logging how they slept, it’s still self-report, just dressed up.
Next is validation against a clinical reference. Sleep tracking quality differs drastically from one device to the next. If sleep data is going to inform a mental health product, it should be benchmarked against polysomnography or a comparable standard, with peer-reviewed comparison data in the open. The distance between “we track sleep” and “we track sleep in a way a clinician would recognize” is larger than most marketing copy admits.
Then comes the practical test: does sleep change what the product does, or did it just earn a tab in the app? If sleep produces a dashboard but never alters coaching, prompting, or clinical escalation, it’s still adjacent content. Integration becomes real when the product behaves differently after a week of poor sleep than it does after a week of good sleep.
Digital mental health’s first decade produced meaningful science and real symptom improvement. The next stage needs physiological grounding, and sleep is the most replicated, workable signal to build around. The research has been settled for a long time. What wasn’t available was a credible way to measure sleep at the scale these products operate at. That gap is closing now.
The practical move for anyone building in this space is to look closely at where sleep sits in the mental health experience they’re responsible for. If it’s limited to a meditation track, ask whether that still fits what the evidence says. If it’s missing entirely, ask what’s blocking it. And if sleep data is already present, the real question is simple: when the signal says the user slept poorly, does anything the product actually does change? That’s the bar worth holding the whole category to.
Get Fresh Healthcare & IT Stories Delivered Daily
Join thousands of your healthcare & HealthIT peers who subscribe to our daily newsletter.
We respect your privacy and will never sell or give out your contact information