Search

Rapid eye movement (REM) sleep behavior disorder (RBD) is parasomnia and a prodromal manifestation of Parkinson's disease. The current diagnostic method relies on manual scoring of polysomnograms (PSGs), a procedure that is time and effort intensive, subject to interscorer variability, and requires high level of expertise. Here, we present an automatic and interpretable diagnostic tool for RBD that analyzes PSGs using end-to-end deep neural networks. We optimized hierarchical attention networks in a 5-fold cross validation directly to classify RBD from PSG data recorded in 143 participants with RBD and 147 age-and sex-matched controls. An ensemble model using logistic regression was implemented to fuse decisions from networks trained in various signal combinations. We interpreted the networks using gradient SHAP that attribute relevance of input signals to model decisions. The ensemble model achieved a sensitivity of 91.4 % and a specificity of 86.3 %. Interpretation showed that electroencephalography (EEG) and leg electromyography (EMG) exhibited most patterns with high relevance. This study validates a robust diagnostic tool for RBD and proposes an interpretable and fully automatic framework for end-to-end modeling of other sleep disorders from PSG data. Clinical relevance- This study presents a novel diagnostic tool for RBD that considers neurophysiologic biomarkers in multiple modalities.

The aim of this study was to design a new deep learning framework for end-to-end processing of polysomnograms. This framework can be trained to analyze whole-night polysomnograms without the limitations of and bias towards clinical scoring guidelines. We validated the framework by predicting the age of subjects. We designed a hierarchical attention network architecture, which can be pre-trained to predict labels based on 5-minute epochs of data and fine-tuned to predict based on whole-night polysomnography recordings. The model was trained on 511 recordings from the Cleveland Family study and tested on 146 test subjects aged between 6 to 88 years. The proposed network achieved a mean absolute error of 7.36 years and a correlation to true age of 0.857. Sleep can be analyzed using our end-to-end deep learning framework, which we expect can generalize to learning other subject-specific labels such as sleep disorders. The difference in the predicted and chronological age is further proposed as an estimate of biological age.