Sleep Cycle, the Gothenburg-based sleep tracking company listed on Nasdaq Stockholm, has spent the past year running a clinical validation study for an AI tool that screens for obstructive sleep apnea using nothing but the microphone already built into a user’s iPhone. No mask, no at-home polysomnography kit, no wearable required. The company says the technology is built on more than 3 billion analyzed sleep sessions collected through its existing app and is now approaching the performance bar needed for regulatory clearance.
A Massive, Mostly Invisible Problem
The scale of the condition Sleep Cycle is targeting is staggering: an estimated 950 million adults aged 30 to 69 worldwide have obstructive sleep apnea, and nearly 80% of those cases go undiagnosed, according to figures the company cites from existing epidemiological research. Untreated sleep apnea is linked to elevated risk of hypertension, heart disease, stroke, and daytime cognitive impairment, but the traditional diagnostic pathway — a referral to a sleep clinic followed by an overnight polysomnography study or a take-home testing kit — is expensive, slow, and inaccessible for huge swaths of the population, which is precisely the gap Sleep Cycle is trying to fill with a phone that most people already own.
How the AI Actually Works
Dr. Mikael Kågebäck, Sleep Cycle’s Chief Technology Officer and a PhD in AI and machine learning, has described the system as built on “a proprietary sound model trained to identify subtle sleep-related breathing patterns” picked up by the phone’s microphone overnight. Rather than requiring a dedicated sensor, the model analyzes acoustic signatures — breathing rhythm changes, pauses, and other audio markers correlated with airway obstruction — that it has learned from Sleep Cycle’s enormous existing dataset of recorded sleep sessions, then flags users whose patterns suggest elevated OSA risk.
Inside the Clinical Study
The company’s pivotal validation study launched in June 2025 in partnership with Sleep Testing Australia, running out of clinical sites in Brisbane and Adelaide with roughly 700 participants, overseen by Dr. Mike Gradisar as principal investigator. Sleep Cycle has also been in FDA pre-submission dialogue in the US as it maps out a regulatory pathway, a sign the company intends to position the tool as a legitimate screening device rather than a wellness novelty. CEO Erik Jivmark has framed the effort in explicitly strategic terms: “Sleep is the next frontier in preventive health, and we’re here to lead it,” he said, positioning the apnea tool as the leading edge of a broader push into diagnostic-grade consumer sleep technology.
Consumer Appetite Is Already There
Even before the product has cleared regulatory review, consumer interest has been substantial: more than 150,000 people have registered interest in the forthcoming consumer sleep apnea screening service, according to the company. That kind of pre-launch demand reflects both the sheer size of the undiagnosed population and a broader cultural shift in 2026 toward sleep tracking as a mainstream wellness habit, alongside competing efforts from device makers like Eight Sleep, which has also been exploring whether its smart mattress sensor data can support health-condition detection beyond sleep optimization alone.
Reasons for Caution
Sleep medicine specialists, including Mason Vozzo, a diagnostic services manager at Sleep Testing Australia involved in evaluating the technology, have generally welcomed the application of machine learning to sleep diagnostics, but the broader clinical community remains cautious about audio-only screening tools. Acoustic analysis can be confounded by ambient noise, bed partners, pets, or shared sleeping environments, and a screening flag is categorically different from a diagnosis — false positives could send a wave of worried users to already-overburdened sleep clinics, while false negatives could offer dangerous reassurance to people who truly have the condition. Sleep Cycle has been careful to frame the tool as a screening aid rather than a diagnostic replacement, intended to identify at-risk individuals who should then pursue formal testing, not to substitute for a clinical workup.
What Happens After the Study
Sleep Cycle’s full clinical study was expected to wrap by September 2026, with the company stating its algorithm has already reached, in internal testing, the performance level it believes is required for regulatory approval. The next milestones to watch are whether the company files for and receives FDA clearance, how the tool performs once exposed to the messiness of real-world bedrooms rather than controlled clinical settings, and whether health systems and insurers are willing to treat a phone-based screen as a legitimate first step in the sleep apnea diagnostic pathway — a question with real consequences given how many of the 950 million affected adults currently never get diagnosed at all.
Photo: congerdesign / PIXABAY via Pixabay
