{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-deep-learning-architecture-for-temporal","title":"A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series","arxiv_id":"1707.03321","date":"2017-07-05","proceeding":null,"authors":["Stanislas Chambon","Mathieu Galtier","Pierrick Arnal","Gilles Wainrib","Alexandre Gramfort"],"abstract":"Sleep stage classification constitutes an important preliminary exam in the\ndiagnosis of sleep disorders. It is traditionally performed by a sleep expert\nwho assigns to each 30s of signal a sleep stage, based on the visual inspection\nof signals such as electroencephalograms (EEG), electrooculograms (EOG),\nelectrocardiograms (ECG) and electromyograms (EMG). We introduce here the first\ndeep learning approach for sleep stage classification that learns end-to-end\nwithout computing spectrograms or extracting hand-crafted features, that\nexploits all multivariate and multimodal Polysomnography (PSG) signals (EEG,\nEMG and EOG), and that can exploit the temporal context of each 30s window of\ndata. For each modality the first layer learns linear spatial filters that\nexploit the array of sensors to increase the signal-to-noise ratio, and the\nlast layer feeds the learnt representation to a softmax classifier. Our model\nis compared to alternative automatic approaches based on convolutional networks\nor decisions trees. Results obtained on 61 publicly available PSG records with\nup to 20 EEG channels demonstrate that our network architecture yields\nstate-of-the-art performance. Our study reveals a number of insights on the\nspatio-temporal distribution of the signal of interest: a good trade-off for\noptimal classification performance measured with balanced accuracy is to use 6\nEEG with 2 EOG (left and right) and 3 EMG chin channels. Also exploiting one\nminute of data before and after each data segment offers the strongest\nimprovement when a limited number of channels is available. As sleep experts,\nour system exploits the multivariate and multimodal nature of PSG signals in\norder to deliver state-of-the-art classification performance with a small\ncomputational cost.","url_abs":"http://arxiv.org/abs/1707.03321v2","url_pdf":"http://arxiv.org/pdf/1707.03321v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"a-deep-learning-architecture-for-temporal","repo_url":"https://github.com/mcd4874/neurips_competition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"eeg-1","task_name":"EEG"},{"task_slug":"eeg","task_name":"Electroencephalogram (EEG)"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"time-series-1","task_name":"Time Series"},{"task_slug":"time-series","task_name":"Time Series Analysis"}],"methods":[{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1707.03321","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}