Poul Jennum
Professor
Make your next move.
Prepare your application
Review drafts before you send them.
Explore related searches
Articles (4)
Sleep is essential for physical and mental health, yet large–scale assessment of sleep stages and sleep apnea is limited by the cost and burden of clinical polysomnography. Wrist accelerometry provides a scalable lower–fidelity alternative, but its usefulness is highly reliant on modeling choices. Generative self–supervised learning has remained unexplored for this purpose. To address this, we developed and pretrained physiology–aware masked autoencoders to capture pulse– and respiration–related motion using ~700,000 days of wrist accelerometry from 108,904 recordings in the UK Biobank cohort. This effort culminated in AcceleRest, a transformer model pretrained using a respiratory amplification objective. Performance was validated against polysomnography across 478 recordings from six cohorts and devices including independent external test cohorts. AcceleRest feature vectors enabled linear wake–NREM–REM sleep staging with a macro F1 score of 0.69 and respiratory event detection with a macro F1 of 0.56. The combined model outputs enabled sleep apnea severity evaluation with a 67% sensitivity and 96% specificity for severe apnea. Overall agreement between polysomnography and AcceleRest showed a bias of 0.8 min for total sleep duration, with 95% limits of agreement (LoA) of −101.6 to 103.2 min, and 32.5 min for REM sleep duration and 95% LoA of −68.7 to 133.6 min. These findings demonstrate that physiology-aware pretraining can enable robust and clinically meaningful sleep phenotyping from wrist accelerometers, supporting scalable screening and longitudinal monitoring of sleep health. To the best of our knowledge, AcceleRest represents the first wrist accelerometry model for joint sleep stage and apnea evaluation. All code and models will be made available upon final peer-reviewed publication.
Year:
2026
Study Objectives This study investigates the association between non-rapid eye movement parasomnias (NREMp) and health and socioeconomic outcomes in adults. Methods We estimated health and socioeconomic consequences of NREMp in the 7 and 4 years preceding diagnosis, respectively, and 2 years following diagnosis. The retrospective cohort included 159 Danish adults diagnosed with NREMp after age 18 years compared with 636 controls (matched for age, gender, marital status, and residential municipality at index year). Diagnoses, healthcare costs, and socioeconomic data were obtained from Danish nationwide health and administrative registers. Results Patients with NREMp had significantly more healthcare contacts and higher prevalence of comorbidity across several major disease groups: certain infectious and parasitic diseases; mental and behavioral disorders; diseases of the nervous system; diseases of the eye and adnexa; diseases of the digestive system; diseases of the respiratory system; diseases of the genitourinary system; and symptoms, signs, and abnormal clinical and laboratory findings, injury, poisoning, and external causes. Their healthcare costs were also higher than those of the matched controls. However, no significant differences in mortality were observed during the 2-year post-diagnosis follow-up. In addition, educational attainment and employment rates were comparable with the matched-control population. Conclusion Adults diagnosed with NREMp showed increased comorbidity and healthcare utilization in the years preceding diagnosis compared to matched controls, without corresponding differences in education or employment. Statement of Significance This study provides novel insight into the health, social, and economic patterns in the 7 years preceding and 2 years following a diagnosis of non-rapid eye movement parasomnias (NREMp). The findings indicate that individuals later diagnosed with NREMp had a significantly higher prevalence of comorbidities across a broad range of disease categories, while no substantial differences were observed in mortality rates, education, employment, or income compared to controls. These results contribute to a better understanding of NREMp while the condition is linked to increased healthcare utilization and comorbid disease burden, it may not be associated with adverse outcomes in educational attainment or employment.
Year:
2025
Study Objectives The manual annotation of polysomnography (PSG) hypnograms is difficult and time-consuming. U-Sleep is an alternative, fast, and publicly available, automated sleep staging system evaluated in adult PSGs. In this study, we compare the staging done by sleep experts and U-sleep in a pediatric sample. Methods PSGs from 56 children aged 6–17 years old (healthy or with a chronic disease) were compared manually annotated with the result of U-sleep. The two outcomes were compared using F1 overlap scores, accuracy, Cohen’s kappa, and correlation coefficients. A qualitative analysis of the most significant systematic differences between the manual and automated scoring was performed. Results U-sleep matched the manually scored hypnograms with an overall mean F1 score (predicted performance) of 0.75 and reached an accuracy of 83.9% and an overall kappa value of 0.77. The stage-wise F1 scores, U-sleep achieved an F1 score of 0.79 in stage wake, 0.40 in N1, 0.86 in N2, 0.84 in N3, and 0.86 in REM. The correlation between U-sleep and the manual scorer was moderately or very strong in all sleep stages (r = .57–.81). Conclusions Overall, there is a high degree of agreement between manual and automatic scoring. This suggests that U-sleep is a valid and effective method for identifying sleep stages based on normal PSGs in a pediatric population. The disagreement was within what is expected for interscorer variation. Further evaluation needs of AI sleep-scoring models include analysis of outliers and pathological sleep staging—which is also a challenge in manual annotation.
Year:
2025
The waste-clearing flow of cerebrospinal fluid through the brain is driven by cerebrovascular, respiratory, and cardiac forces and is thought to be regulated by homeostatic sleep mechanisms. In a circadian-controlled sleep and sleep deprivation study, we investigated how heightened sleep pressure and slow-wave-rich sleep affect physiological brain oscillations in humans, as measured with accelerated neuroimaging. Our study also included a randomised, double-blind, placebo-controlled, cross-over administration of the adrenergic antagonist carvedilol, which allowed us to examine the effect of modulating cerebrovascular pulsatility. We report that sleep deprivation increases low frequency brain oscillations (LFOs) more than sleep does, with LFOs during sleep correlating with cognitive measures of sleep pressure. Conversely, slow-wave-rich NREM sleep (stages N2 and N3) enhances respiration- and cardiac-driven brain pulsations – particularly within grey and white matter. The strength of these brain pulsations escalates with sleep depth (N3 > N2) and correlates with EEG delta power. Carvedilol dampens LFOs, supporting that these reflect cerebrovascular oscillations. Altogether, our findings indicate that sleep pressure promotes cerebrovascular oscillations, while sleep slow waves synchronise with respiration- and cardiac-driven brain pulsations. Graphical Abstract
Year:
2024
Collaborators (9)
Maiken Nedergaard
University of Copenhagen
Anders Stevnhoved Olsen
Rigshospitalet
Pia Weikop
University of Copenhagen
Emmanuel Mignot
Stanford University School of Medicine
Gitte Moos Knudsen
Center director
Rigshospitalet
Vesa Kiviniemi
Professor of Functional Neuroimaging
University of Oulu
Sebastian Camillo Holst
Copenhagen University Hospital
Brice Maxime Hugues Ozenne
Associate Professor
University of Copenhagen
Andreas Brink-Kjær
Technical University of Denmark
Make your next move.
Prepare your application
Review drafts before you send them.