MambaSleepCVD for prediction of long term cardiovascular outcomes from polysomnography
- 1University of Brescia
Abstract
Long-term cardiovascular disease (CVD) risk prediction remains a major global clinical priority. This study investigates how much prognostic information full-night polysomnography (PSG) carries after explicitly controlling demographic confounding, and whether combining specialized unimodal representations with continuous fusion mechanisms can effectively leverage it. We adopt a modular framework based on Coupled Mamba for simultaneous cross-modal and temporal fusion, capturing long-range dependencies and dynamic interactions between physiological modalities throughout sleep. With this capability we assess whether specialized supervised representations from task-specific unimodal encoders offer prognostic value comparable to task-agnostic multimodal self-supervised pre-training. Validated on the Sleep Heart Health Study (SHHS) dataset, our approach achieves performance comparable to or better than current state-of-the-art architectures in the fully supervised, data-abundant regime, although the cardiac channel alone accounts for most of the discriminative signal. More importantly, we identify a significant age-related bias in the reference dataset that, if left unaddressed, creates a shortcut leading to inflated risk predictions, accounting for a large part of the literatures reported performance. Ultimately, our findings suggest that supervised unimodal experts integrated through a fusion engine provide a flexible pathway for prognostic risk ranking from sleep recordings, while reliable CVD screening will require independent external cohorts and explicit confounder control.
Related
- A Frailty-integrated Prognostic Prediction Tool for Long-term Outcomes after Shunt Surgery in Idiopathic Normal Pressure Hydrocephalus: Development and Prospective Validation
- Long-Term Cardiovascular Care After Cytokine Release Syndrome
- Long-Term Major Adverse Kidney and Cardiovascular Events in Predialysis Chronic Kidney Disease: A Real-World Cohort Study
- Serum free eicosapentaenoic acid-to-arachidonic acid ratio and long-term risk of major adverse cardiovascular events in patients with coronary artery disease
- KT-CBHNET: a spatio-temporal deep learning framework for interpretable ECG signal classification for cardiovascular disease prediction
BibTeX
@misc{calzoni2026mambasleepcvd,
title = {{MambaSleepCVD for prediction of long term cardiovascular outcomes from polysomnography}},
author = {A. Calzoni and A. Dei Rossi and G. Monachino and B. Zanchi and L. Fiorillo and M. Savardi and F. D. Faraci and A. Signoroni},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.08.26361626},
url = {https://www.medrxiv.org/content/10.64898/2026.09.08.26361626v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - MambaSleepCVD for prediction of long term cardiovascular outcomes from polysomnography
AU - A. Calzoni
AU - A. Dei Rossi
AU - G. Monachino
AU - B. Zanchi
AU - L. Fiorillo
AU - M. Savardi
AU - F. D. Faraci
AU - A. Signoroni
PY - 2026
DA - 2026/09/08
PB - medRxiv
DO - 10.64898/2026.09.08.26361626
UR - https://www.medrxiv.org/content/10.64898/2026.09.08.26361626v1
AB - Long-term cardiovascular disease (CVD) risk prediction remains a major global clinical priority. This study investigates how much prognostic information full-night polysomnography (PSG) carries after explicitly controlling demographic confounding, and whether combining specialized unimodal representations with continuous fusion mechanisms can effectively leverage it. We adopt a modular framework based on Coupled Mamba for simultaneous cross-modal and temporal fusion, capturing long-range dependencies and dynamic interactions between physiological modalities throughout sleep. With this capability we assess whether specialized supervised representations from task-specific unimodal encoders offer prognostic value comparable to task-agnostic multimodal self-supervised pre-training. Validated on the Sleep Heart Health Study (SHHS) dataset, our approach achieves performance comparable to or better than current state-of-the-art architectures in the fully supervised, data-abundant regime, although the cardiac channel alone accounts for most of the discriminative signal. More importantly, we identify a significant age-related bias in the reference dataset that, if left unaddressed, creates a shortcut leading to inflated risk predictions, accounting for a large part of the literatures reported performance. Ultimately, our findings suggest that supervised unimodal experts integrated through a fusion engine provide a flexible pathway for prognostic risk ranking from sleep recordings, while reliable CVD screening will require independent external cohorts and explicit confounder control.
ER -