Extracting Fuzzy Rules From Polysomnographic Recordings for Infant Sleep Classification

An electroencephalography-based sleep index and supervised machine learning as a suitable tool for automated sleep classification in children

STUDY OBJECTIVES

Although sleep is frequently disrupted in the pediatric intensive care unit, it is currently not possible to perform real-time sleep monitoring at the bedside. In this study, spectral band powers of electroencephalography data are used to derive a simple index for sleep classification.

METHODS

Retrospective study at Erasmus MC Sophia Children's Hospital, using hospital-based polysomnography recordings obtained in non-critically ill children between 2017 and 2021. Six age categories were defined: 6-12 months, 1-3 years, 3-5 years, 5-9 years, 9-13 years, and 13-18 years. Candidate index measures were derived by calculating spectral band powers in different frequent frequency bands of smoothed electroencephalography. With the best performing index, sleep classification models were developed for two, three, and four states via decision tree and five-fold nested cross-validation. Model performance was assessed across age categories and electroencephalography channels.

RESULTS

In total 90 patients with polysomnography were included, with a mean (standard deviation) recording length of 10.3 (1.1) hours. The best performance was obtained with the gamma to delta spectral power ratio of the F4-A1 and F3-A1 channels with smoothing. Balanced accuracy was 0.88, 0.74, and 0.57 for two-, three-, and four-state classification. Across age categories, balanced accuracy ranged between 0.83 and 0.92 and 0.72 and 0.77 for two- and three-state classification, respectively.

CONCLUSIONS

We propose an interpretable and generalizable sleep index derived from single-channel electroencephalography for automated sleep monitoring at the bedside in non-critically ill children ages 6 months to 18 years, with good performance for two- and three-state classification.

CITATION

van Twist E, Hiemstra FW, Cramer ABG, et al. An electroencephalography-based sleep index and supervised machine learning as a suitable tool for automated sleep classification in children. J Clin Sleep Med. 2024;20(3):389-397.

A robust two-stage sleep spindle detection approach using single-channel EEG

Objective.Sleep spindles in the electroencephalogram (EEG) are significant in sleep analysis related to cognitive functions and neurological diseases, and thus are of great clinical interests. An automatic sleep spindle detection algorithm could help decrease the workload of visual inspection by sleep clinicians.Approach.We propose a robust two-stage approach for sleep spindle detection using single-channel EEG. In the pre-detection stage, a stable number of sleep spindle candidates are discovered using the Teager energy operator with adaptive parameters, where the number of true sleep spindles are ensured as many as possible to maximize the detection sensitivity. In the refinement stage, representative features are designed and a bagging classifier is exploited to further recognize the true spindles from all candidates, in order to remove the false detection in the first stage.Main results.Using the union of all experts' annotations as the ground truth, its performance outperforms state-of-the-art works in terms of F1-score (F1) on two public databases (F1: 0.814 for Montreal archive of sleep studies dataset and 0.690 for DREAMS dataset). The annotation consistency between the proposed method and certain selected expert as the trainer could exceed the consistency between two human experts.Significance.The proposed sleep spindle detection method is based on single-channel EEG thus introduces as less interference to the subjects as possible. It is robust to subject variations between databases and is capable of learning certain annotation rules, which is expected to help facilitate the manual labeling of certain experts. In addition, this method is fast enough for real-time applications.

An Open Source Classifier for Bed Mattress Signal in Infant Sleep Monitoring

Objective

To develop a non-invasive and clinically practical method for a long-term monitoring of infant sleep cycling in the intensive care unit.

Methods

Forty three infant polysomnography recordings were performed at 1–18 weeks of age, including a piezo element bed mattress sensor to record respiratory and gross-body movements. The hypnogram scored from polysomnography signals was used as the ground truth in training sleep classifiers based on 20,022 epochs of movement and/or electrocardiography signals. Three classifier designs were evaluated in the detection of deep sleep (N3 state): support vector machine (SVM), Long Short-Term Memory neural network, and convolutional neural network (CNN).

Results

Deep sleep was accurately identified from other states with all classifier variants. The SVM classifier based on a combination of movement and electrocardiography features had the highest performance (AUC 97.6%). A SVM classifier based on only movement features had comparable accuracy (AUC 95.0%). The feature-independent CNN resulted in roughly comparable accuracy (AUC 93.3%).

Conclusion

Automated non-invasive tracking of sleep state cycling is technically feasible using measurements from a piezo element situated under a bed mattress.

Significance

An open source infant deep sleep detector of this kind allows quantitative, continuous bedside assessment of infant’s sleep cycling.

Time-frequency analysis and fuzzy-based detection of heat-stressed sleep EEG spectra

Nowadays, sleep disorders are contemplated as the major issue in the human lives. The current work aims at extraction of time-frequency information from recorded dataset and provides an efficient sleep stage detection method. Recordings of brain signal namely electroencephalogram (EEG), electrooculogram (EOG), and electromyogram (EMG) were carried out under defined clinical condition for the classification of sleep EEG. Subsequent upon the extraction of various features from the raw EEG data, neuro-fuzzy system is trained to classify the sleep stages into three major classes namely awake, slow wave sleep (SWS), and rapid eye movement sleep (REM). This classification would enable medical professionals to diagnose sleep related disorders accurately. The results obtained clearly indicate that the mean performance for SWS stage is profound as compared to REM and awake stage. Specificity and sensitivity of the proposed method are obtained as 95.4% and 80%, respectively. The average accuracy of the system employing neuro-fuzzy approach is found to be 90.6% in which SWS stage was best detected among the other stages of sleep EEG.Graphical abstract.

Automated sleep staging of OSAs based on ICA preprocessing and consolidation of temporal correlations

An automated sleep staging based on analyzing long-range time correlations in EEG is proposed. These correlations, indicating time-scale invariant property or self-similarity at different time scales, are known to be salient dynamical characteristics of stage succession for a sleeping brain even when the subject suffers a destructive disorder such as Obstructive Sleep Apnea (OSA). The goal is to extract a set of complementary features from cerebral sources mapped onto the scalp electrodes or from a number of denoised EEG channels. For this purpose, source localization/extraction and noise reduction approaches based on Independent Component Analysis were used prior to correlation analysis. Feature extracted segments were then classified in one of the five classes including WAKE, STAGE1, STAGE2, SWS and REM via an ensemble neuro-fuzzy classifier. Some techniques were employed to improve the classifier's performance including Scaled Conjugate Gradient Method to speed up learning the ANFIS classifiers, a pruning algorithm to eliminate irrelevant fuzzy rules and the 10-fold cross-validation technique to train and test the system more efficiently. The performance of classification for two strategies including (1) feature extraction from effective cerebral sources and (2) feature extraction from selected channels of denoised EEG signals was compared and contrasted in terms of training errors and test accuracies. The first and second strategies achieved 92.23 and 88.74% agreement with human expert respectively which indicates the effectiveness of the staging system based on cerebral sources of activity. Our results further indicate that the misclassification rates were almost below 10%. The proposed automated sleep staging system is reliable due to the fact that it is based on the underlying dynamics of sleep staging which is less likely to be affected by sleep fragmentations occurred repeatedly in OSA.

Electronic Sleep Stage Classifiers: A Survey and VLSI Design Methodology

First, existing sleep stage classifier sensors and algorithms are reviewed and compared in terms of classification accuracy, level of automation, implementation complexity, invasiveness, and targeted application. Next, the implementation of a miniature microsystem for low-latency automatic sleep stage classification in rodents is presented. The classification algorithm uses one EMG (electromyogram) and two EEG (electroencephalogram) signals as inputs in order to detect REM (rapid eye movement) sleep, and is optimized for low complexity and low power consumption. It is implemented in an on-board low-power FPGA connected to a multi-channel neural recording IC, to achieve low-latency (order of 1 ms or less) classification. Off-line experimental results using pre-recorded signals from nine mice show REM detection sensitivity and specificity of 81.69% and 93.86%, respectively, with the maximum latency of 39 [Formula: see text]. The device is designed to be used in a non-disruptive closed-loop REM sleep suppression microsystem, for future studies of the effects of REM sleep deprivation on memory consolidation.

A low computational cost algorithm for REM sleep detection using single channel EEG

The push towards low-power and wearable sleep systems requires using minimum number of recording channels to enhance battery life, keep processing load small and be more comfortable for the user. Since most sleep stages can be identified using EEG traces, enormous power savings could be achieved by using a single channel of EEG. However, detection of REM sleep from one channel EEG is challenging due to its electroencephalographic similarities with N1 and Wake stages. In this paper we investigate a novel feature in sleep EEG that demonstrates high discriminatory ability for detecting REM phases. We then use this feature, that is based on spectral edge frequency (SEF) in the 8–16 Hz frequency band, together with the absolute power and the relative power of the signal, to develop a simple REM detection algorithm. We evaluate the performance of this proposed algorithm with overnight single channel EEG recordings of 5 training and 15 independent test subjects. Our algorithm achieved sensitivity of 83%, specificity of 89% and selectivity of 61% on a test database consisting of 2221 REM epochs. It also achieved sensitivity and selectivity of 81 and 75% on PhysioNet Sleep-EDF database consisting of 8 subjects. These results demonstrate that SEF can be a useful feature for automatic detection of REM stages of sleep from a single channel EEG.

RECOGNITION OF SLEEP STAGES BASED ON A COMBINED NEURAL NETWORK AND FUZZY SYSTEM USING WAVELET TRANSFORM FEATURES

Recognition of sleep stages is an important task in the assessment of the quality of sleep. Several biomedical signals, such as EEG, ECG, EMG and EOG are used extensively to classify the stages of sleep, which is very important for the diagnosis of sleep disorders. Many sleep studies have been conducted that focused on the automatic classification of sleep stages. In this research, a new classification method is presented that uses an Elman neural network combined with fuzzy rules to extract sleep features based on wavelet decompositions. The nine subjects who participated in this study were recruited from Cheng-Ching General Hospital in Taichung, Taiwan. The sampling frequency was 250 Hz, and a single-channel (C3-A1) EEG signal was acquired for each subject. The system consisted of a combined neural network and fuzzy system that was used to recognize sleep stages based on epochs (10-second segments of data). The classification results relied on the strong points of combined neural network and fuzzy system, which achieved an average specificity of approximately 96% and an average accuracy of approximately 94%.

Automated detection of apnea/hypopnea events in healthy children polysomnograms: preliminary results

A methodology to detect sleep apnea/hypopnea events in the respiratory signals of polysomnographic recordings is presented. It applies empirical mode decomposition (EMD), Hilbert-Huang transform (HHT), fuzzy logic and signal preprocessing techniques for feature extraction, expert criteria and context analysis. EMD, HHT and fuzzy logic are used for artifact detection and preliminary detection of respiration signal zones with significant variations in the amplitude of the signal; feature extraction, expert criteria and context analysis are used to characterize and validate the respiratory events. An annotated database of 30 all-night polysomnographic recordings, acquired from 30 healthy ten-year-old children, was divided in a training set of 15 recordings (485 sleep apnea/hypopnea events), a validation set of five recordings (109 sleep apnea/hypopnea events), and a testing set of ten recordings (281 sleep apnea/hypopnea events). The overall detection performance on the testing data set was 89.7% sensitivity and 16.3% false-positive rate. The next step is to include discrimination among apneas, hypopneas and respiratory pauses.

NON-INVASIVE NOCTURNAL HYPOGLYCEMIA DETECTION FOR INSULIN-DEPENDENT DIABETES MELLITUS USING GENETIC FUZZY LOGIC METHOD

Hypoglycemia, or low blood glucose, is the most common complication experienced by Type 1 diabetes mellitus (T1DM) patients. It is dangerous and can result in unconsciousness, seizures and even death. The most common physiological parameter to be effected from hypoglycemic reaction are heart rate (HR) and correct QT interval (QTc) of the electrocardiogram (ECG) signal. Based on physiological parameters, a genetic algorithm based fuzzy reasoning model is developed to recognize the presence of hypoglycemia. To optimize the parameters of the fuzzy model in the membership functions and fuzzy rules, a genetic algorithm is used. A validation strategy based adjustable fitness is introduced in order to prevent the phenomenon of overtraining (overfitting). For this study, 15 children with 569 sampling data points with Type 1 diabetes volunteered for an overnight study. The effectiveness of the proposed algorithm is found to be satisfactory by giving better sensitivity and specificity compared with other existing methods for hypoglycemia detection.

Analyzing the dynamics of emotional scene sequence using recurrent neuro-fuzzy network

In this paper, we propose a new framework to analyze the temporal dynamics of the emotional stimuli. For this framework, both electroencephalography signal and visual information are of great importance. The fusion of visual information with brain signals allows us to capture the users' emotional state. Thus we adopt previously proposed fuzzy-GIST as emotional feature to summarize the emotional feedback. In order to model the dynamics of the emotional stimuli sequence, we develop a recurrent neuro-fuzzy network for modeling the dynamic events of emotional dimensions including valence and arousal. It can incorporate human expertise by IF-THEN fuzzy rule while recurrent connections allow the fuzzy rules of network to see its own previous output. The results show that such a framework can interact with human subjects and generate arbitrary emotional sequences after learning the dynamics of an emotional sequence with enough number of samples.

Application of recurrence quantification analysis to automatically estimate infant sleep states using a single channel of respiratory data

Previous work has identified that non-linear variables calculated from respiratory data vary between sleep states, and that variables derived from the non-linear analytical tool recurrence quantification analysis (RQA) are accurate infant sleep state discriminators. This study aims to apply these discriminators to automatically classify 30 s epochs of infant sleep as REM, non-REM and wake. Polysomnograms were obtained from 25 healthy infants at 2 weeks, 3, 6 and 12 months of age, and manually sleep staged as wake, REM and non-REM. Inter-breath interval data were extracted from the respiratory inductive plethysmograph, and RQA applied to calculate radius, determinism and laminarity. Time-series statistic and spectral analysis variables were also calculated. A nested cross-validation method was used to identify the optimal feature subset, and to train and evaluate a linear discriminant analysis-based classifier. The RQA features radius and laminarity and were reliably selected. Mean agreement was 79.7, 84.9, 84.0 and 79.2 % at 2 weeks, 3, 6 and 12 months, and the classifier performed better than a comparison classifier not including RQA variables. The performance of this sleep-staging tool compares favourably with inter-human agreement rates, and improves upon previous systems using only respiratory data. Applications include diagnostic screening and population-based sleep research.

Automated detection of rapid eye movements in children

We present an automated multiple-step tool to identify Rapid Eye Movements (REMs) in the polysomnogram, based on modeling expert criteria. It begins by identifying the polysomnogram segments compatible with REMs presence. On these segments, high-energy REMs are identified. Then, vicinity zones around those REMs are defined, and lesser-energy REMs are sought in these vicinities. This strategy has the advantage that it can detect lesser-energy REMs without increasing much the false positive detections. Signal processing, feature extraction, and fuzzy logic tools are used to achieve the goal. The tool was trained and validated on a database consisting of 20 all-night polysomnogram recordings (160 hr) of healthy ten-year-old children. Preliminary results on the validation set show 85.5% sensitivity and a false positive rate of 16.2%. Our tool works on complete polysomnogram recordings, without the need of preprocessing, prior knowledge of the hypnogram, or noise-free segments selection.

Analysis and automatic identification of sleep stages using higher order spectra

Electroencephalogram (EEG) signals are widely used to study the activity of the brain, such as to determine sleep stages. These EEG signals are nonlinear and non-stationary in nature. It is difficult to perform sleep staging by visual interpretation and linear techniques. Thus, we use a nonlinear technique, higher order spectra (HOS), to extract hidden information in the sleep EEG signal. In this study, unique bispectrum and bicoherence plots for various sleep stages were proposed. These can be used as visual aid for various diagnostics application. A number of HOS based features were extracted from these plots during the various sleep stages (Wakefulness, Rapid Eye Movement (REM), Stage 1-4 Non-REM) and they were found to be statistically significant with p-value lower than 0.001 using ANOVA test. These features were fed to a Gaussian mixture model (GMM) classifier for automatic identification. Our results indicate that the proposed system is able to identify sleep stages with an accuracy of 88.7%.