A real‐time sleep scoring framework for closed‐loop sleep manipulation in mice

Abstract

Subtle changes in sleep architecture can accompany and be symptomatic of many diseases or disorders. In order to probe and understand the complex interactions between sleep and health, the ability to model, track, and modulate sleep in preclinical animal models is vital. While various methods have been described for scoring experimental sleep recordings, few are designed to work in real time – a prerequisite for closed‐loop sleep manipulation. In the present study, we have developed algorithms and software to classify sleep in real time and validated it on C57BL/6 mice (n = 8). Hidden Markov models of baseline sleep dynamics were fitted using an unsupervised algorithm to electroencephalogram (EEG) and electromyogram (EMG) data for each mouse, and were able to classify sleep in a manner highly concordant with manual scoring (Cohen’s Kappa >75%) up to 3 weeks after model construction. This approach produced reasonably accurate estimates of common sleep metrics (proportion, mean duration, and number of bouts). After construction, the models were used to track sleep in real time and accomplish selective rapid eye movement (REM) sleep restriction by triggering non‐invasive somatosensory stimulation. During REM restriction trials, REM bout duration was significantly reduced, and the classifier continued to perform satisfactorily despite the disrupted sleep patterns. The software can easily be tailored for use with other commercial or customised methods of sleep disruption (e.g. stir bar, optogenetic stimulation, etc.) and could serve as a robust platform to facilitate closed‐loop experimentation. The source code and documentation are freely available upon request from the authors.

Probing Causality of the Brainstem-Hypothalamic Murine Models of Sleep-Wake Regulation

“…Furthermore, to detect a statistically significant change in neuronal firing rates during state transitions, with sub-second temporal resolution, many sleep cycles are required. Finally, conventional SOV scoring techniques assign a score to non-overlapping time-windows (Takahashi et al, 2009(Takahashi et al, , 2010Bastianini et al, 2017;Sakai, 2018;Grieger et al, 2021;Huffman et al, 2021) with fixed length and predetermined boundaries. These fixed windows prevent accurate assessment of the SOV dynamics and identification of the true onset of a state.…”

Automatic sleep–wake classification and Parkinson's disease recognition using multifeature fusion with support vector machine

“…These studies used monitoring technologies such as electroencephalogram (EEG), electrocorticography (ECoG), electromyogram (EMG), and electrooculogram (EOG) to analyze the state of sleep and wakefulness (WAKE) 6 . With the development of digital signal processing technology, researchers have explored computer‐aided automatic sleep–wake scoring methods 7–14 . Petrovic et al.…”

“…Additionally, various classification methods, including the hidden Markov model, 12 random forest, 13 and deep learning networks, 14 have been proposed to develop suitable sleep-wake scoring models. Among these, the SVM demonstrates enhanced efficiency in achieving real-time automatic classification owing to its reduced training sample requirement and response time.…”

“…6 With the development of digital signal processing technology, researchers have explored computer-aided automatic sleep-wake scoring methods. [7][8][9][10][11][12][13][14] Petrovic et al utilized EMG power and the δ/θ power ratio as features and employed a two-cluster K-means algorithm for three-state classification. 7 Crisler et al selected various features from the parietal and frontal cortex, along with EMG, and utilized SVM for sleep classification in normal rats.…”

An attention-based temporal convolutional network for rodent sleep stage classification across species, mutants and experimental environments with single-channel electroencephalogram

“…Yaghouby et al (2016) trained hidden Markov models (HMMs) and scored the vigilance state with 90.07% sensitivity. Huffman et al (2021) relied on the mean-squared power of the 80-100 Hz filtered EMG signal to differentiate sleep and wake states; two EEG power ratios to discriminate between NREM and REM within sleep. They reached an overall accuracy of 89.13% with an HMM-based method.…”