Sleep staging based on polysomnography (PSG) performed by human experts is the de facto “gold standard” for the objective measurement of sleep. PSG and manual sleep staging is, however, personnel-intensive and time-consuming and it is thus impractical to monitor a person’s sleep architecture over extended periods. Here, we present a novel, low-cost, automatized, deep learning alternative to PSG sleep staging that provides a reliable epoch-by-epoch four-class sleep staging approach (Wake, Light [N1 + N2], Deep, REM) based solely on inter-beat-interval (IBI) data. Having trained a multi-resolution convolutional neural network (MCNN) on the IBIs of 8898 full-night manually sleep-staged recordings, we tested the MCNN on sleep classification using the IBIs of two low-cost (<EUR 100) consumer wearables: an optical heart rate sensor (VS) and a breast belt (H10), both produced by POLAR®. The overall classification accuracy reached levels comparable to expert inter-rater reliability for both devices (VS: 81%, κ = 0.69; H10: 80.3%, κ = 0.69). In addition, we used the H10 and recorded daily ECG data from 49 participants with sleep complaints over the course of a digital CBT-I-based sleep training program implemented in the App NUKKUAA™. As proof of principle, we classified the IBIs extracted from H10 using the MCNN over the course of the training program and captured sleep-related changes. At the end of the program, participants reported significant improvements in subjective sleep quality and sleep onset latency. Similarly, objective sleep onset latency showed a trend toward improvement. Weekly sleep onset latency, wake time during sleep, and total sleep time also correlated significantly with the subjective reports. The combination of state-of-the-art machine learning with suitable wearables allows continuous and accurate monitoring of sleep in naturalistic settings with profound implications for answering basic and clinical research questions.

Classifying sleep stages is an important basis for neuroscience, health sciences, psychology and many other fields. However, the manual determination of sleep stages is tedious and time consuming. Therefore, the development of automatic sleep stage classifiers based on data collected with low-cost sensor systems is an important research area. This study aims to analyse the generalisability of different machine learning approaches for sleep stage classification. We train three different models (random forest, CNN-LSTM and seq2seq) for classifying three as well as four sleep stages, with the MESA data set. For validation, we use a fivefold cross-validation and further validate the models with one new self-recorded test data set to analyse the models’ generalisability to a completely new cohort with different characteristics with regard to age and health status. Our results show that the two deep learning approaches performed better than the random forest. Moreover, all models are generalisable and therefore suitable for sleep stage classification on a new three-stage classification data set. However, generalisability for the four-stage classification task shows poorer performance, and therefore requires new approaches such as transfer learning or a larger data set to train the models.