Biology subjects

Perslev, M.

Publications and source records attributed to Perslev, M..

3 recordsLinked to original sources

bioRxiv2025

Optimizing automated sleep stage scoring of 5-second mini-epochs: a transfer learning study

Study objectiveConventional sleep staging relies on 30-second epochs, potentially concealing transient sleep stage intrusion and reducing precision. Building on our previous study of mini-epochs, we investigated whether U-Sleep, an existing automatic deep learning-based sleep staging model with high performance in epochs, could be optimized to similar performance level in 5-second mini-epoch scoring, thereby enabling more detailed sleep characterization. MethodsWe created a dataset of 48,000 human-scored 5-second mini-epochs from 100 PSGs. We compared mini-epochs to human-scored epochs before U-Sleep was optimized using transfer learning and evaluated on a test set. Model performance was assessed using F1-scores, confusion matrices, stage distributions and transition rates comparing scorings of the original U-Sleep before, and the optimized U-Sleep after transfer learning to human-scored mini-epochs. ResultsCompared to human-scored epochs, human-scored mini-epochs captured significantly more transitions (1.70/minute vs. 0.21/minute, p<0.001), and significantly more wake (8.4% versus 5.4%), N1 (7.2% versus 5.4%), and N2 (51.8% versus 40.9%), less N3 (15.4% versus 25.2%) and REM sleep (16.7% versus 23.0%) (all p<0.001). Optimizing U-Sleep improved its performance significantly from F1=0.74 to F1=0.81 (p<0.05) and gave increased transition rates in the test set (original U-Sleep: 1.06/minute, optimized U-Sleep: 1.34/minute, human-scored miniepochs: 1.70/minute). Stage distributions did not differ between optimized U-Sleeps scorings and human-scored mini-epochs. ConclusionsAfter optimization, U-Sleep performance in mini-epochs matched the high performance levels previously reported in both human and automated 30-second epoch scoring. This demonstrates the feasibility of precise, automated high resolution sleep staging. Future work should include external validation and application to full-night recordings. Statement of significanceConventional 30-second epochs limit temporal resolution in sleep staging and may conceal transient intrusions of wake or sleep stages. However, no validated methods are available for highresolution scoring. In this study, we trained and validated the state-of-the-art deep learning model U-Sleep for accurate automatic 5-second mini-epoch scoring using a large dataset of humanscored mini-epochs. The optimized model achieved a high performance, matching levels from previously reported automatic and human epoch scoring. Compared to epoch scoring, miniepochs captured significantly more stage transitions, supporting their ability to uncover sleep dynamics that are otherwise lost. Our findings show the potential of high-resolution sleep staging for more detailed characterization of sleep architecture and demonstrate the feasibility of precise, automatic mini-epoch scoring.

bioengineering↗

bioRxiv2024

Membrane curvature association of amphipathic helix 8 drives constitutive GPCR endocytosis

Cellular signaling relies on the activity of transmembrane receptors and their presentation on the cellular surface. Their continuous insertion in the plasma membrane is balanced by constitutive and activity dependent internalization, which is orchestrated by adaptor proteins recognizing semi-specific motifs within the receptors intracellular regions. Here we describe a complementary and evolutionary conserved and refined trafficking mechanism for G-protein coupled receptors (GPCR). This mechanism relies on the insertion of their amphipathic helix 8 into the inner leaflet of lipid membranes, orthogonal to the transmembrane helices. These amphipathic helices dictate subcellular localization of the receptors and autonomously drive their endocytosis by cooperative assembly and association with areas of high membrane curvature. The strength of helix 8 membrane insertion propensity quantitatively predicts the rate of constitutive internalization of GPCRs. This discovery advances our understanding of membrane protein trafficking and highlights a new principle of receptor-lipid interactions that may have broader implications for cellular signaling and therapeutic targeting. One-Sentence SummaryReceptor proteins navigate cellular membranes by interacting with their curvature using an evolutionary conserved mechanism that relies on amphipathic helices and complements direct coupling to the endocytic protein machinery.

cell biology↗

bioRxiv2023

Validation of the sleep EEG headband ZMax

Polysomnography (PSG) is the gold standard for recording sleep. However, the standard PSG systems are bulky, expensive, and often confined to lab environments. These systems are also time-consuming in electrode placement and sleep scoring. Such limitations render standard PSG systems less suitable for large-scale or longitudinal studies of sleep. Recent advances in electronics and artificial intelligence enabled wearable PSG systems. Here, we present a study aimed at validating the performance of ZMax, a widely-used wearable PSG that includes frontal electroencephalography (EEG) and actigraphy but no submental electromyography (EMG). We analyzed 135 nights with simultaneous ZMax and standard PSG recordings amounting to over 900 hours from four different datasets, and evaluated the performance of the headbands proprietary automatic sleep scoring (ZLab) alongside our open-source algorithm (DreamentoScorer) in comparison with human sleep scoring. ZLab and DreamentoScorer compared to human scorers with moderate and substantial agreement and Cohens kappa scores of 59.61% and 72.18%, respectively. We further analyzed the competence of these algorithms in determining sleep assessment metrics, as well as shedding more lights on the bandpower computation, and morphological analysis of sleep microstructural features between ZMax and standard PSG. Relative bandpower computed by ZMax implied an error of 5.5% (delta), 4.5% (theta), 1.6% (alpha), 0.5% (sigma), 0.8% (beta), and 0.2% (gamma), compared to standard PSG. In addition, the microstructural features detected in ZMax did not represent exactly the same characteristics as in standard PSG. Besides similarities and discrepancies between ZMax and standard PSG, we measured and discussed the technology acceptance rate, feasibility of data collection with ZMax, and highlighted essential factors for utilizing ZMax as a reliable tool for both monitoring and modulating sleep.

neuroscience↗