Modern deep learning holds a great potential to transform clinical studies of human sleep. Teaching a machine to carry out routine tasks would be a tremendous reduction in workload for clinicians. Sleep staging, a fundamental step in sleep practice, is a suitable task for this and will be the focus in this article. Recently, automatic sleep-staging systems have been trained to mimic manual scoring, leading to similar performance to human sleep experts, at least on scoring of healthy subjects. Despite tremendous progress, we have not seen automatic sleep scoring adopted widely in clinical environments. This review aims to provide the shared view of the authors on the most recent state-of-the-art developments in automatic sleep staging, the challenges that still need to be addressed, and the future directions needed for automatic sleep scoring to achieve clinical value.

Sleep is a key phenomenon to both understanding, diagnosing and treatment of many illnesses, as well as for studying health and well being in general. Today, the only widely accepted method for clinically monitoring sleep is the polysomnography (PSG), which is, however, both expensive to perform and influences the sleep. This has led to investigations into light weight electroencephalography (EEG) alternatives. However, there has been a substantial performance gap between proposed alternatives and PSG. Here we show results from an extensive study of 80 full night recordings of healthy participants wearing both PSG equipment and ear-EEG. We obtain automatic sleep scoring with an accuracy close to that achieved by manual scoring of scalp EEG (the current gold standard), using only ear-EEG as input, attaining an average Cohen’s kappa of 0.73. In addition, this high performance is present for all 20 subjects. Finally, 19/20 subjects found that the ear-EEG had little to no negative effect on their sleep, and subjects were generally able to apply the equipment without supervision. This finding marks a turning point on the road to clinical long term sleep monitoring: the question should no longer be whether ear-EEG could ever be used for clinical home sleep monitoring, but rather when it will be.

Human sleep is cyclical with a period of approximately 90 minutes, implying long temporal dependency in the sleep data. Yet, exploring this long-term dependency when developing sleep staging models has remained untouched. In this work, we show that while encoding the logic of a whole sleep cycle is crucial to improve sleep staging performance, the sequential modelling approach in existing state-of-the-art deep learning models are inefficient for that purpose. We thus introduce a method for efficient long sequence modelling and propose a new deep learning model, L-SeqSleepNet, which takes into account whole-cycle sleep information for sleep staging. Evaluating L-SeqSleepNet on four distinct databases of various sizes, we demonstrate state-of-the-art performance obtained by the model over three different EEG setups, including scalp EEG in conventional Polysomnography (PSG), in-ear EEG, and around-the-ear EEG (cEEGrid), even with a single EEG channel input. Our analyses also show that L-SeqSleepNet is able to alleviate the predominance of N2 sleep (the major class in terms of classification) to bring down errors in other sleep stages. Moreover the network becomes much more robust, meaning that for all subjects where the baseline method had exceptionally poor performance, their performance are improved significantly. Finally, the computation time only grows at a sub-linear rate when the sequence length increases.

Quantification of sleep is important for the diagnosis of sleep disorders and sleep research. However, the only widely accepted method to obtain sleep staging is by visual analysis of polysomnography (PSG), which is expensive and time consuming. Here, we investigate automated sleep scoring based on a low‐cost, mobile electroencephalogram (EEG) platform consisting of a lightweight EEG amplifier combined with flex‐printed cEEGrid electrodes placed around the ear, which can be implemented as a fully self‐applicable sleep system. However, cEEGrid signals have different amplitude characteristics to normal scalp PSG signals, which might be challenging for visual scoring. Therefore, this study evaluates the potential of automatic scoring of cEEGrid signals using a machine learning classifier (“random forests”) and compares its performance with manual scoring of standard PSG. In addition, the automatic scoring of cEEGrid signals is compared with manual annotation of the cEEGrid recording and with simultaneous actigraphy. Acceptable recordings were obtained in 15 healthy volunteers (aged 35 ± 14.3 years) during an extended nocturnal sleep opportunity, which induced disrupted sleep with a large inter‐individual variation in sleep parameters. The results demonstrate that machine‐learning‐based scoring of around‐the‐ear EEG outperforms actigraphy with respect to sleep onset and total sleep time assessments. The automated scoring outperforms human scoring of cEEGrid by standard criteria. The accuracy of machine‐learning‐based automated scoring of cEEGrid sleep recordings compared with manual scoring of standard PSG was satisfactory. The findings show that cEEGrid recordings combined with machine‐learning‐based scoring holds promise for large‐scale sleep studies.

Sleep and sleep quality assessment by means of sleep stage analysis is important for both scientific and clinical applications. Unfortunately, the presently preferred method, polysomnography (PSG), requires considerable expert assistance and significantly affects the sleep of the person under observation. A reliable, accurate and mobile alternative to the PSG would make sleep information much more readily available in a wide range of medical circumstances. Using an already proven method, ear-EEG, in which electrodes are placed inside the concha and ear canal, we measure cerebral activity and automatically score the sleep into up to five stages. These results are compared to manual scoring by trained clinicians, based on a simultaneously recorded PSG. The correspondence between manually scored sleep, based on the PSG, and the automatic labelling, based on ear-EEG data, was evaluated using Cohen’s kappa coefficient. Kappa values are in the range 0.5–0.8, making ear-EEG relevant for both scientific and clinical applications. Furthermore, a sleep-wake classifier with leave-one-out cross validation yielded specificity of 0.94 and sensitivity of 0.52 for the sleep stage. Ear-EEG based scoring has clear advantages when compared to both the PSG and other mobile solutions, such as actigraphs. It is far more mobile, and potentially cheaper than the PSG, and the information on sleep stages is far superior to a wrist-based actigraph, or other devices based solely on body movement. This study shows that ear-EEG recordings carry information about sleep stages, and indicates that automatic sleep staging based on ear-EEG can classify sleep stages with a level of accuracy that makes it relevant for both scientific and clinical sleep assessment.

Objective: Brain waves vary between people. This work aims to improve automatic sleep staging for longitudinal sleep monitoring via personalization of algorithms based on individual characteristics extracted from sleep data recorded during the first night. Approach: As data from a single night are very small, thereby making model training difficult, we propose a Kullback–Leibler (KL) divergence regularized transfer learning approach to address this problem. We employ the pretrained SeqSleepNet (i.e. the subject independent model) as a starting point and finetune it with the single-night personalization data to derive the personalized model. This is done by adding the KL divergence between the output of the subject independent model and it of the personalized model to the loss function during finetuning. In effect, KL-divergence regularization prevents the personalized model from overfitting to the single-night data and straying too far away from the subject independent model. Main results: Experimental results on the Sleep-EDF Expanded database consisting of 75 subjects show that sleep staging personalization with single-night data is possible with help of the proposed KL-divergence regularization. On average, we achieve a personalized sleep staging accuracy of 79.6%, a Cohen’s kappa of 0.706, a macro F1-score of 73.0%, a sensitivity of 71.8%, and a specificity of 94.2%. Significance: We find both that the approach is robust against overfitting and that it improves the accuracy by 4.5 percentage points compared to the baseline method without personalization and 2.2 percentage points compared to it with personalization but without regularization.