Jaap F van der Aar, Merel M van Gilst, Daan A van den Ende, Pedro Fonseca, Bregje N J van Wetten, Hennie C J P Janssen, Elisabetta Peri, Sebastiaan Overeem
{"title":"利用迁移学习优化异质睡眠障碍人群中的可穿戴单通道脑电图睡眠分期。","authors":"Jaap F van der Aar, Merel M van Gilst, Daan A van den Ende, Pedro Fonseca, Bregje N J van Wetten, Hennie C J P Janssen, Elisabetta Peri, Sebastiaan Overeem","doi":"10.5664/jcsm.11380","DOIUrl":null,"url":null,"abstract":"<p><strong>Study objectives: </strong>Although various wearable electroencephalography devices have been developed, performance evaluation of the devices and their associated automated sleep stage classification models is mostly limited to healthy participants. A major barrier for applying automated wearable electroencephalography sleep staging in clinical populations is the need for large-scale data for model training. We therefore investigated transfer learning as a strategy to overcome limited data availability and optimize automated single-channel electroencephalography sleep staging in people with sleep disorders.</p><p><strong>Methods: </strong>We acquired 52 single-channel frontopolar headband electroencephalography recordings from a heterogeneous sleep-disordered population with concurrent polysomnography (PSG). We compared 3 model training strategies: \"pretraining\" (ie, training on a larger dataset of 901 conventional PSGs), \"training-from-scratch\" (ie, training on wearable headband recordings), and \"fine-tuning\" (ie, training on conventional PSGs, followed by training on headband recordings). Performance was evaluated on all headband recordings using 10-fold cross-validation.</p><p><strong>Results: </strong>Highest performance for 5-stage classification was achieved with fine-tuning (κ = .778), significantly higher than with pretraining (κ = .769) and with training-from-scratch (κ = .733). No significant differences or systematic biases were observed with clinically relevant sleep parameters derived from PSG. All sleep disorder categories showed comparable performance.</p><p><strong>Conclusions: </strong>This study emphasizes the importance of leveraging larger available datasets through deep transfer learning to optimize performance with limited data availability. Findings indicate strong similarity in data characteristics between conventional PSG and headband recordings. Altogether, results suggest the combination of the headband, classification model, and training methodology can be viable for sleep monitoring in the heterogeneous clinical population.</p><p><strong>Citation: </strong>van der Aar JF, van Gilst MM, van den Ende DA, et al. Optimizing wearable single-channel electroencephalography sleep staging in a heterogeneous sleep-disordered population using transfer learning. <i>J Clin Sleep Med.</i> 2025;21(2):315-323.</p>","PeriodicalId":50233,"journal":{"name":"Journal of Clinical Sleep Medicine","volume":" ","pages":"315-323"},"PeriodicalIF":3.5000,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Optimizing wearable single-channel electroencephalography sleep staging in a heterogeneous sleep-disordered population using transfer learning.\",\"authors\":\"Jaap F van der Aar, Merel M van Gilst, Daan A van den Ende, Pedro Fonseca, Bregje N J van Wetten, Hennie C J P Janssen, Elisabetta Peri, Sebastiaan Overeem\",\"doi\":\"10.5664/jcsm.11380\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Study objectives: </strong>Although various wearable electroencephalography devices have been developed, performance evaluation of the devices and their associated automated sleep stage classification models is mostly limited to healthy participants. A major barrier for applying automated wearable electroencephalography sleep staging in clinical populations is the need for large-scale data for model training. We therefore investigated transfer learning as a strategy to overcome limited data availability and optimize automated single-channel electroencephalography sleep staging in people with sleep disorders.</p><p><strong>Methods: </strong>We acquired 52 single-channel frontopolar headband electroencephalography recordings from a heterogeneous sleep-disordered population with concurrent polysomnography (PSG). We compared 3 model training strategies: \\\"pretraining\\\" (ie, training on a larger dataset of 901 conventional PSGs), \\\"training-from-scratch\\\" (ie, training on wearable headband recordings), and \\\"fine-tuning\\\" (ie, training on conventional PSGs, followed by training on headband recordings). Performance was evaluated on all headband recordings using 10-fold cross-validation.</p><p><strong>Results: </strong>Highest performance for 5-stage classification was achieved with fine-tuning (κ = .778), significantly higher than with pretraining (κ = .769) and with training-from-scratch (κ = .733). No significant differences or systematic biases were observed with clinically relevant sleep parameters derived from PSG. All sleep disorder categories showed comparable performance.</p><p><strong>Conclusions: </strong>This study emphasizes the importance of leveraging larger available datasets through deep transfer learning to optimize performance with limited data availability. Findings indicate strong similarity in data characteristics between conventional PSG and headband recordings. Altogether, results suggest the combination of the headband, classification model, and training methodology can be viable for sleep monitoring in the heterogeneous clinical population.</p><p><strong>Citation: </strong>van der Aar JF, van Gilst MM, van den Ende DA, et al. Optimizing wearable single-channel electroencephalography sleep staging in a heterogeneous sleep-disordered population using transfer learning. <i>J Clin Sleep Med.</i> 2025;21(2):315-323.</p>\",\"PeriodicalId\":50233,\"journal\":{\"name\":\"Journal of Clinical Sleep Medicine\",\"volume\":\" \",\"pages\":\"315-323\"},\"PeriodicalIF\":3.5000,\"publicationDate\":\"2025-02-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Clinical Sleep Medicine\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.5664/jcsm.11380\",\"RegionNum\":3,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"CLINICAL NEUROLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Clinical Sleep Medicine","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.5664/jcsm.11380","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CLINICAL NEUROLOGY","Score":null,"Total":0}
Optimizing wearable single-channel electroencephalography sleep staging in a heterogeneous sleep-disordered population using transfer learning.
Study objectives: Although various wearable electroencephalography devices have been developed, performance evaluation of the devices and their associated automated sleep stage classification models is mostly limited to healthy participants. A major barrier for applying automated wearable electroencephalography sleep staging in clinical populations is the need for large-scale data for model training. We therefore investigated transfer learning as a strategy to overcome limited data availability and optimize automated single-channel electroencephalography sleep staging in people with sleep disorders.
Methods: We acquired 52 single-channel frontopolar headband electroencephalography recordings from a heterogeneous sleep-disordered population with concurrent polysomnography (PSG). We compared 3 model training strategies: "pretraining" (ie, training on a larger dataset of 901 conventional PSGs), "training-from-scratch" (ie, training on wearable headband recordings), and "fine-tuning" (ie, training on conventional PSGs, followed by training on headband recordings). Performance was evaluated on all headband recordings using 10-fold cross-validation.
Results: Highest performance for 5-stage classification was achieved with fine-tuning (κ = .778), significantly higher than with pretraining (κ = .769) and with training-from-scratch (κ = .733). No significant differences or systematic biases were observed with clinically relevant sleep parameters derived from PSG. All sleep disorder categories showed comparable performance.
Conclusions: This study emphasizes the importance of leveraging larger available datasets through deep transfer learning to optimize performance with limited data availability. Findings indicate strong similarity in data characteristics between conventional PSG and headband recordings. Altogether, results suggest the combination of the headband, classification model, and training methodology can be viable for sleep monitoring in the heterogeneous clinical population.
Citation: van der Aar JF, van Gilst MM, van den Ende DA, et al. Optimizing wearable single-channel electroencephalography sleep staging in a heterogeneous sleep-disordered population using transfer learning. J Clin Sleep Med. 2025;21(2):315-323.
期刊介绍:
Journal of Clinical Sleep Medicine focuses on clinical sleep medicine. Its emphasis is publication of papers with direct applicability and/or relevance to the clinical practice of sleep medicine. This includes clinical trials, clinical reviews, clinical commentary and debate, medical economic/practice perspectives, case series and novel/interesting case reports. In addition, the journal will publish proceedings from conferences, workshops and symposia sponsored by the American Academy of Sleep Medicine or other organizations related to improving the practice of sleep medicine.