Arquitetura Hierárquica CNN-BiMamba para Identificação de Estágios de Sono em Polissonografia com Aplicação à Detecção de REM sem Atonia
Resumo
O estagiamento automático do sono é etapa fundamental para a quantificação de REM Sem Atonia (RSWA), principal biomarcador do Transtorno Comportamental do Sono REM (RBD), marcador precoce de doenças neurodegenerativas. Métodos de aprendizado profundo geralmente operam em épocas de 30 segundos, resolução insuficiente para os eventos musculares breves (0,5–5s) característicos de RSWA. Este trabalho propõe uma arquitetura hierárquica CNN-BiMamba para o estagiamento do sono a partir de sequências contínuas de mini-épocas de 3 segundos. Avaliado em 151 exames de PSG via validação cruzada GroupKFold, o modelo alcançou F1-macro de 0, 736±0, 021 e Kappa de Cohen de 0, 691±0, 027, constituindo uma etapa inicial viável para futura detecção e quantificação de RSWA.
Palavras-chave:
CNN-BiMamba, Estagiamento do Sono, REM sem Atonia (RSWA), Polissonografia
Referências
Alvarez-Estevez, D. and Rijsman, R. (2022). Haaglanden Medisch Centrum sleep staging database. PhysioNet. Version 1.1, DOI: 10.13026/t79q-fr32.
American Academy of Sleep Medicine (2023). The AASM Manual for the Scoring of Sleep and Associated Events: Terminology and Technical Specifications. Version 3.
Anderer, P., Ross, M., Cerny, A., Vasko, R., Shaw, E., and Fonseca, P. (2023). Overview of the hypnodensity approach to scoring sleep for polysomnography and home sleep testing. Front. Sleep, 2:1163477. DOI: 10.3389/frsle.2023.1163477.
Chambon, S., Galtier, M. N., Arnal, P. J., Wainrib, G., and Gramfort, A. (2018). A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26:758–769. DOI: 10.1109/TNSRE.2018.2813138.
Chen, X., Zhang, Y., Tang, Z., Wu, H., Chen, C., and Chen, W. (2025). MtRBD: Advancing iRBD Analysis With Multi-Task Learning for Joint Sleep Staging and RSWA Detection. IEEE J. Biomed. Health Inform., 29(12):8743–8750. DOI: 10.1109/JBHI.2025.3606944.
Frauscher, B., Iranzo, A., et al. (2012). Normative emg values during rem sleep for the diagnosis of rem sleep behavior disorder. Sleep, 35(6):835–847.
Goerttler, S., Wang, Y., Eldele, E., Wu, M., and He, F. (2025). MSA-CNN: A Lightweight Multi-Scale CNN with Attention for Sleep Stage Classification. DOI: 10.48550/arXiv.2501.02949.
Gu, A. and Dao, T. (2024). Linear-Time Sequence Modeling with Selective State Spaces. DOI: 10.48550/arXiv.2312.00752.
Jiahao, H., Ur Rahman, M. M., Al-Naffouri, T., and Laleg-Kirati, T.-M. (2025). Mamba-CAM-Sleep: A Mamba-based Channel Attention Model for Sleep Staging Classification. In Annual Int. Conf. of the IEEE EMBC, pages 1–6, Copenhagen, Denmark. IEEE. DOI: 10.1109/EMBC58623.2025.11251806.
Postuma, R. B., Iranzo, A., et al. (2019). Risk and predictors of dementia and parkinsonism in idiopathic REM sleep behaviour disorder: a multicentre study. Brain, 142(3):744–759. DOI: 10.1093/brain/awz030.
Schenck, C. H., Bundlie, S. R., Ettinger, M. G., and Mahowald, M. W. (1986). Chronic Behavioral Disorders of Human REM Sleep: A New Category of Parasomnia. Sleep, 9(2):293–308. DOI: 10.1093/sleep/9.2.293.
Sobreira-Neto, M. A., Stelzer, F. G., et al. (2023). REM sleep behavior disorder: update on diagnosis and management. Arquivos de Neuro-Psiquiatria, 81(12):1179–1194. DOI: 10.1055/s-0043-1777111.
Supratak, A., Dong, H., Wu, C., and Guo, Y. (2017). DeepSleepNet: A model for automatic sleep stage scoring based on raw single-channel eeg. IEEE TNSRE Journal, 25(11):1998–2008. DOI: 10.1109/TNSRE.2017.2721116.
Tsinalis, O., Matthews, P. M., and Guo, Y. (2016). Automatic sleep stage scoring using time-frequency analysis and stacked sparse autoencoders. Annals of Biomedical Engineering, 44:1587–1597. DOI: 10.1007/s10439-015-1444-y.
Zhang, A. H., He-Mo, A., et al. (2026). Mamba-based deep learning approach for sleep staging on a wireless multimodal wearable system without electroencephalography. SLEEP, 49(4):zsag022. DOI: 10.1093/sleep/zsag022.
Zhang, C., Cui, W., and Guo, J. (2024). MSSC-BiMamba: Multimodal Sleep Stage Classification and Early Diagnosis of Sleep Disorders with Bidirectional Mamba. DOI: 10.48550/arXiv.2405.20142.
Zhou, X., Han, Y., et al. (2024). BiT-MamSleep: Bidirectional temporal Mamba for eeg sleep staging. URL: [link].
American Academy of Sleep Medicine (2023). The AASM Manual for the Scoring of Sleep and Associated Events: Terminology and Technical Specifications. Version 3.
Anderer, P., Ross, M., Cerny, A., Vasko, R., Shaw, E., and Fonseca, P. (2023). Overview of the hypnodensity approach to scoring sleep for polysomnography and home sleep testing. Front. Sleep, 2:1163477. DOI: 10.3389/frsle.2023.1163477.
Chambon, S., Galtier, M. N., Arnal, P. J., Wainrib, G., and Gramfort, A. (2018). A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26:758–769. DOI: 10.1109/TNSRE.2018.2813138.
Chen, X., Zhang, Y., Tang, Z., Wu, H., Chen, C., and Chen, W. (2025). MtRBD: Advancing iRBD Analysis With Multi-Task Learning for Joint Sleep Staging and RSWA Detection. IEEE J. Biomed. Health Inform., 29(12):8743–8750. DOI: 10.1109/JBHI.2025.3606944.
Frauscher, B., Iranzo, A., et al. (2012). Normative emg values during rem sleep for the diagnosis of rem sleep behavior disorder. Sleep, 35(6):835–847.
Goerttler, S., Wang, Y., Eldele, E., Wu, M., and He, F. (2025). MSA-CNN: A Lightweight Multi-Scale CNN with Attention for Sleep Stage Classification. DOI: 10.48550/arXiv.2501.02949.
Gu, A. and Dao, T. (2024). Linear-Time Sequence Modeling with Selective State Spaces. DOI: 10.48550/arXiv.2312.00752.
Jiahao, H., Ur Rahman, M. M., Al-Naffouri, T., and Laleg-Kirati, T.-M. (2025). Mamba-CAM-Sleep: A Mamba-based Channel Attention Model for Sleep Staging Classification. In Annual Int. Conf. of the IEEE EMBC, pages 1–6, Copenhagen, Denmark. IEEE. DOI: 10.1109/EMBC58623.2025.11251806.
Postuma, R. B., Iranzo, A., et al. (2019). Risk and predictors of dementia and parkinsonism in idiopathic REM sleep behaviour disorder: a multicentre study. Brain, 142(3):744–759. DOI: 10.1093/brain/awz030.
Schenck, C. H., Bundlie, S. R., Ettinger, M. G., and Mahowald, M. W. (1986). Chronic Behavioral Disorders of Human REM Sleep: A New Category of Parasomnia. Sleep, 9(2):293–308. DOI: 10.1093/sleep/9.2.293.
Sobreira-Neto, M. A., Stelzer, F. G., et al. (2023). REM sleep behavior disorder: update on diagnosis and management. Arquivos de Neuro-Psiquiatria, 81(12):1179–1194. DOI: 10.1055/s-0043-1777111.
Supratak, A., Dong, H., Wu, C., and Guo, Y. (2017). DeepSleepNet: A model for automatic sleep stage scoring based on raw single-channel eeg. IEEE TNSRE Journal, 25(11):1998–2008. DOI: 10.1109/TNSRE.2017.2721116.
Tsinalis, O., Matthews, P. M., and Guo, Y. (2016). Automatic sleep stage scoring using time-frequency analysis and stacked sparse autoencoders. Annals of Biomedical Engineering, 44:1587–1597. DOI: 10.1007/s10439-015-1444-y.
Zhang, A. H., He-Mo, A., et al. (2026). Mamba-based deep learning approach for sleep staging on a wireless multimodal wearable system without electroencephalography. SLEEP, 49(4):zsag022. DOI: 10.1093/sleep/zsag022.
Zhang, C., Cui, W., and Guo, J. (2024). MSSC-BiMamba: Multimodal Sleep Stage Classification and Early Diagnosis of Sleep Disorders with Bidirectional Mamba. DOI: 10.48550/arXiv.2405.20142.
Zhou, X., Han, Y., et al. (2024). BiT-MamSleep: Bidirectional temporal Mamba for eeg sleep staging. URL: [link].
Publicado
08/09/2026
Como Citar
SOSTER, Fernando A.; SOSTER, Letícia A.; CAZZOLATO, Mirela T..
Arquitetura Hierárquica CNN-BiMamba para Identificação de Estágios de Sono em Polissonografia com Aplicação à Detecção de REM sem Atonia. In: DATA SCIENCE FOR SOCIAL GOOD BRAZILIAN WORKSHOP (DS4SG) - SIMPÓSIO BRASILEIRO DE BANCO DE DADOS (SBBD), 41. , 2026, São Carlos/SP.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 653-658.
DOI: https://doi.org/10.5753/sbbd_estendido.2026.249932.
