Four-stage sleep from wrist PPG and accelerometry
Our transformer model with Hidden Markov Model postprocessing enforces plausible stage transitions, reaching up to 90.0% accuracy and a Cohen’s kappa of 0.831 against polysomnography (EMBC 2026).
Sleep analytics you can defend to reviewers.
We estimate sleep from raw wearable signals, not only vendor sleep scores, and analyze it night by night alongside activity, HRV, smartphone sensing and EMAs.
What we deliver
Sleep analytics, end to end.
Standard and research metrics
Total sleep time, sleep efficiency, sleep onset latency, wake after sleep onset, sleep timing and regularity, and nocturnal heart rate and HRV.
Device validation studies
Epoch-by-epoch comparison of consumer devices against polysomnography, actigraphy or ECG, as in our Oura ring and smartwatch validation studies.
Night-to-night patterns
Variability, weekday and weekend differences, and trends across weeks or months, for example across pregnancy.
Sleep and daytime behavior
How sleep relates to activity, stress and mood, including mediation and causal analysis.
Nightly metrics for JITAI
Validated nightly metrics that can feed just-in-time adaptive intervention rules and EMA triggers in Centralive.
How we analyze it
- Deep learning on PPG, respiration and accelerometry
- Hidden Markov Model temporal correction
- Epoch-by-epoch agreement, sensitivity, specificity and Cohen’s kappa
- Bland-Altman analysis for continuous metrics
- Mixed-effects models for repeated nights
What you receive
- Per-night and per-participant sleep tables
- Hypnograms and summary figures
- A methods section and validation statistics for your paper
- Documented, reproducible code
Evidence
Selected peer-reviewed work by our team.
- Kazemi et al. Improved sleep stage predictions by deep learning of photoplethysmogram and respiration patterns. Computers in Biology and Medicine, 2024. doi:10.1016/j.compbiomed.2024.108679
- Kazemi et al. Hybrid sleep stage detection using Hidden Markov Chains and Transformers. IEEE EMBC, 2026 (forthcoming).
- Cao et al. Accuracy assessment of Oura ring nocturnal heart rate and heart rate variability in comparison with electrocardiography in time and frequency domains: comprehensive analysis. Journal of Medical Internet Research, 2022. doi:10.2196/27487
- Asgari Mehrabadi et al. Sleep tracking of a commercially available smart ring and smartwatch against medical-grade actigraphy in everyday settings: instrument validation study. JMIR mHealth and uHealth, 2020. doi:10.2196/20465
- Azimi et al. Personalized maternal sleep quality assessment: an objective IoT-based longitudinal study. IEEE Access, 2019. doi:10.1109/ACCESS.2019.2927781
- Kazemi et al. Can sleep quality attributes be predicted from physical activity in everyday settings?. IEEE EMBC, 2023. doi:10.1109/EMBC40787.2023.10340421
- Khatibi et al. Impact of COVID-19 pandemic on sleep including HRV and physical activity as mediators: a causal ML approach. IEEE BSN, 2023. doi:10.1109/BSN58485.2023.10331423
Read more on our blog: How raw multimodal signals beat the black-box sleep score.