SleepMaMi: A Universal Sleep Foundation Model for Integrating Macro- and Micro-structures

Keondo Park, Younghoon Na, You Rim Choi, Hyunwoo Ryu, Hyun-Woo Shin, Hyung-Sin Kim
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:96917-96942, 2026.

Abstract

While the shift toward unified foundation models has revolutionized many deep learning domains, sleep medicine remains largely restricted to task-specific models that focus on localized micro-structure features. These approaches often neglect the rich, multi-modal context of Polysomnography (PSG) and fail to capture the global macro-structure of a full night’s sleep. To address this, we introduce SleepMaMi, a Sleep Foundation Model engineered to master both hour-long sleep architectures and fine-grained signal morphologies. Our framework utilizes a hierarchical dual-encoder design: a Macro-Encoder to model full-night temporal dependencies and a Micro-Encoder to capture short-term characteristics from biosignals. Macro-Encoder is trained via Demographic-Guided Contrastive Learning, which aligns overnight sleep patterns with objective subject metadata, such as age and sex, to refine global representations. Micro-Encoder is optimized via a hybrid Masked Autoencoder (MAE) and multi-modal contrastive objective. Pre-trained on a massive corpus of $>$20,000 PSG recordings (158K hours), SleepMaMi outperforms or matches existing foundation models across a diverse suite of downstream tasks, demonstrating superior generalizability and label-efficient adaptation for clinical sleep analysis.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-park26n, title = {{S}leep{M}a{M}i: A Universal Sleep Foundation Model for Integrating Macro- and Micro-structures}, author = {Park, Keondo and Na, Younghoon and Choi, You Rim and Ryu, Hyunwoo and Shin, Hyun-Woo and Kim, Hyung-Sin}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {96917--96942}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/park26n/park26n.pdf}, url = {https://proceedings.mlr.press/v306/park26n.html}, abstract = {While the shift toward unified foundation models has revolutionized many deep learning domains, sleep medicine remains largely restricted to task-specific models that focus on localized micro-structure features. These approaches often neglect the rich, multi-modal context of Polysomnography (PSG) and fail to capture the global macro-structure of a full night’s sleep. To address this, we introduce SleepMaMi, a Sleep Foundation Model engineered to master both hour-long sleep architectures and fine-grained signal morphologies. Our framework utilizes a hierarchical dual-encoder design: a Macro-Encoder to model full-night temporal dependencies and a Micro-Encoder to capture short-term characteristics from biosignals. Macro-Encoder is trained via Demographic-Guided Contrastive Learning, which aligns overnight sleep patterns with objective subject metadata, such as age and sex, to refine global representations. Micro-Encoder is optimized via a hybrid Masked Autoencoder (MAE) and multi-modal contrastive objective. Pre-trained on a massive corpus of $>$20,000 PSG recordings (158K hours), SleepMaMi outperforms or matches existing foundation models across a diverse suite of downstream tasks, demonstrating superior generalizability and label-efficient adaptation for clinical sleep analysis.} }
Endnote
%0 Conference Paper %T SleepMaMi: A Universal Sleep Foundation Model for Integrating Macro- and Micro-structures %A Keondo Park %A Younghoon Na %A You Rim Choi %A Hyunwoo Ryu %A Hyun-Woo Shin %A Hyung-Sin Kim %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-park26n %I PMLR %P 96917--96942 %U https://proceedings.mlr.press/v306/park26n.html %V 306 %X While the shift toward unified foundation models has revolutionized many deep learning domains, sleep medicine remains largely restricted to task-specific models that focus on localized micro-structure features. These approaches often neglect the rich, multi-modal context of Polysomnography (PSG) and fail to capture the global macro-structure of a full night’s sleep. To address this, we introduce SleepMaMi, a Sleep Foundation Model engineered to master both hour-long sleep architectures and fine-grained signal morphologies. Our framework utilizes a hierarchical dual-encoder design: a Macro-Encoder to model full-night temporal dependencies and a Micro-Encoder to capture short-term characteristics from biosignals. Macro-Encoder is trained via Demographic-Guided Contrastive Learning, which aligns overnight sleep patterns with objective subject metadata, such as age and sex, to refine global representations. Micro-Encoder is optimized via a hybrid Masked Autoencoder (MAE) and multi-modal contrastive objective. Pre-trained on a massive corpus of $>$20,000 PSG recordings (158K hours), SleepMaMi outperforms or matches existing foundation models across a diverse suite of downstream tasks, demonstrating superior generalizability and label-efficient adaptation for clinical sleep analysis.
APA
Park, K., Na, Y., Choi, Y.R., Ryu, H., Shin, H. & Kim, H.. (2026). SleepMaMi: A Universal Sleep Foundation Model for Integrating Macro- and Micro-structures. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:96917-96942 Available from https://proceedings.mlr.press/v306/park26n.html.

Related Material