Deep into Sleep Ep207 - Sleep Data Revolution: Using AI to Detect Patterns, Health Risks, and Improve Sleep Quality with Mikael Kågebäck, PhD
Aug 29, 2025
Mikael Kågebäck, Chief Technology Officer at Sleep Cycle and AI expert, dives into the transformative role of artificial intelligence in sleep health. He reveals how audio data can unveil global sleep patterns and identify health risks like sleep apnea. Mikael discusses the groundbreaking 'Cough Radar' feature, which uses cough detection to predict respiratory illness trends. He also shares insights on the ethical management of user data, the future of personalized sleep coaching, and how these advancements could revolutionize sleep medicine.
AI Snips
Chapters
Transcript
Episode notes
On-Device Processing Protects Privacy
- Sleep Cycle processes audio on-device to protect user privacy and allow offline tracking.
- Short user-consented clips are anonymized and annotated to train robust global models.
Global Patterns and Pandemic Effects
- Aggregated sleep metrics reveal cross-country differences like shorter sleep in East Asia versus Europe.
- Longitudinal data enabled measuring pandemic-related increases in sleep from working at home.
Track Sleep Over Time, Not Single Nights
- Use nightly sleep architecture and regularity to monitor long-term health signals, not single nights.
- Track sleep efficiency and staging over weeks to spot meaningful changes.