New Sleep.ai Study Demonstrates AI’s Role in Transforming Preventive Healthcare

Study of 68K Users and 2.7M Nights Validates AI Models for Sleep Apnea and Insomnia; With Further Validation in Progress

 Sleep.ai, the company behind the world’s most validated sleep intelligence platform, presents results of its machine-learning R&D, demonstrating that smartphone-measured sleep stage data may be used to screen for risk of obstructive sleep apnea and insomnia. The analyses highlight the potential for Sleep.ai’s scalable models to drive major breakthroughs in population health. The conference abstract results will be published in the supplementary issue of Sleep Medicine, the official journal of the World Sleep Society.

Research
The research, “Machine learning-based prediction of sleep apnea and insomnia using objective sleep data from a contactless smartphone application,” was recently presented at the World Sleep Congress in Singapore. Models were trained on 2.7 million nights of objective sleep data from over 68,000 users collected through Sleep.ai’s PSG-validated smartphone app, which uses sonar signals to stage sleep without the need for wearables or external devices.

Also Read: AiThority Interview with Tim Morrs, CEO at SpeakUp

Results

• Obstructive Sleep Apnea (OSA): Sleep.ai’s model, which used objective sleep data collected from the smartphone app and no questions and answers, exhibited an AUC of 0.80, outperforming some commonly used screening approaches.
• Insomnia: The insomnia model, also using only sleep stage data from the smartphone application, approximated the accuracy of some screening methods with an AUC of 0.73.

Why It Matters
Rates of sleep apnea and insomnia going undiagnosed remain unacceptably high, with nearly 1 billion people globally remaining undiagnosed and untreated due to problems in the entire clinical pipeline, from education and knowledge through to screening, diagnosis, and access to medical and therapeutic services. By adopting smartphone-based screening technologies, Sleep.ai offers a simpler way to enable screening and subsequent treatment.

“Eliminating friction is key to improving population sleep health. These models were deployed on ubiquitous smartphones and may be well-positioned to scale triaging at-risk users for confirmatory testing, potential diagnosis, and timely treatment,” said Elie Gottlieb, PhD, Head of Applied Sleep Science at Sleep.ai.

Market POV
Most consumer health solutions today rely on wearables, focus narrowly on apnea, or remain limited to early-stage validation. Sleep.ai sets a new benchmark with a smartphone-only, dual-disorder screening model, assessed in real-world conditions, with further clinical validation ongoing. By combining scientific rigor with unmatched scalability, Sleep.ai advances its mission to make research-backed sleep intelligence universally accessible. “We know that people around the world are searching about sleep at a rate higher than any other health topic – we are laser-focused on liberating the data on their phones to drive better sleep health for the purposes of helping them and our partners who serve them,” said Colin Lawlor, CEO at Sleep.ai.

Also Read: Cognitive Product Design: Empowering Non-Technical Users Through Natural Language Interaction With AI-Native PLM