Automatic sleep-stage classification of heart rate and actigraphy data using deep and transfer learning approaches

Rest-Activity Rhythm Phenotypes in Adults with Epilepsy and Intellectual Disability

Objective

Sleep and rest-activity rhythms (RARs) are perturbed in many forms of neuropsychiatric illness. In this study, we applied wrist actigraphy to describe the extent of RAR perturbations in adults with epilepsy and intellectual disability ("E+ID"), using a cross-sectional case-control design. We examined whether RAR phenotypes correlated with epilepsy severity, deficits in adaptive function and/or comorbid psychopathology.

Methods

Primary caregivers of E+ID adults provided informed consent during routine ambulatory clinic visits and were asked to complete standardized surveys of overall epilepsy severity (GASE, Global Assessment of Severity of Epilepsy), adaptive function (ABAS-3, Adaptive Behavior Assessment System-3) and psychopathology (ABCL, Adult Behavior Checklist). Caregivers were also asked to ensure that subjects wore an Actiwatch-2 device continuously on their nondominant wrist for at least ten days. From recorded actograms, we calculated RAR amplitude, acrophase, robustness, intradaily variability (IV), interdaily stability (IS) and estimates of sleep quantity and timing. We compared these RAR metrics against those from (i) a previously published cohort of adults with epilepsy without ID (E-ID), and (ii) a cohort of age- and sex-matched intellectually able subjects measured within the Study of Latinos (SOL) Ancillary actigraphy study (SOL). Within E+ID subjects, we applied k-means analysis to divide subjects into three actigraphically distinct clusters.

Results

46 E+ID subjects (median age 26 [20-68], 47% female) provided a median recording duration of 11 days [range 6-27]. Surveys reflected low to extremely low levels of adaptive function (ABAS3 General Adaptive Composite score: median 50 [49-75]), and low/subclinical levels of psychopathology (ABCL total score: median 54.5 [25-67]). Compared with E-ID (n=57) and SOL (n=156) cohorts, E+ID subjects displayed significantly lower RAR amplitude, robustness and IS, with significantly higher IV and total daily sleep. K-means clustering of E+ID subjects recognized an intermediate cluster "B", with RAR values indistinguishable to E-ID. Cluster "A" subjects displayed pronounced hypoactivity and hypersomnia with high rates of rhythm fragmentation, while cluster "C" subjects featured hyper-robust and high amplitude RARs. All three clusters were similar in age, body mass index, antiseizure medication (ASM) polytherapy, ABAS3 and ABCL scores. We qualitatively describe RAR examples from all three clusters.

Interpretation

We show that adults with epilepsy and intellectual disability display a wide spectrum of RAR phenotypes that do not neatly correlate with measures of adaptive function or epilepsy severity. Prospective studies are necessary to determine whether continuous actigraphic monitoring can sensitively capture changes in chronobiological health that may arise with disease progression, iatrogenesis (e.g., ASM toxicity) or acute health deteriorations (e.g., seizure exacerbation, pneumonia). Similar long-term data is necessary to recognize whether behavioral interventions targeted to 'normalize' RARs may promote improvements in adaptive function and therapy engagement.

A novel deep learning model based on transformer and cross modality attention for classification of sleep stages

The classification of sleep stages is crucial for gaining insights into an individual's sleep patterns and identifying potential health issues. Employing several important physiological channels in different views, each providing a distinct perspective on sleep patterns, can have a great impact on the efficiency of the classification models. In the context of neural networks and deep learning models, transformers are very effective, especially when dealing with time series data, and have shown remarkable compatibility with sequential data analysis as physiological channels. On the other hand, cross-modality attention by integrating information from multiple views of the data enables to capture relationships among different modalities, allowing models to selectively focus on relevant information from each modality. In this paper, we introduce a novel deep-learning model based on transformer encoder-decoder and cross-modal attention for sleep stage classification. The proposed model processes information from various physiological channels with different modalities using the Sleep Heart Health Study Dataset (SHHS) data and leverages transformer encoders for feature extraction and cross-modal attention for effective integration to feed into the transformer decoder. The combination of these elements increased the accuracy of the model up to 91.33% in classifying five classes of sleep stages. Empirical evaluations demonstrated the model's superior performance compared to standalone approaches and other state-of-the-art techniques, showcasing the potential of combining transformer and cross-modal attention for improved sleep stage classification.

Sleep-stage dependence and co-existence of cardio-respiratory coordination and phase synchronization

Interactions between the cardiac and respiratory systems play a pivotal role in physiological functioning. Nonetheless, the intricacies of cardio-respiratory couplings, such as cardio-respiratory phase synchronization (CRPS) and cardio-respiratory coordination (CRC), remain elusive, and an automated algorithm for CRC detection is lacking. This paper introduces an automated CRC detection algorithm, which allowed us to conduct a comprehensive comparison of CRPS and CRC during sleep for the first time using an extensive database. We found that CRPS is more sensitive to sleep-stage transitions, and intriguingly, there is a negative correlation between the degree of CRPS and CRC when fluctuations in breathing frequency are high. This comparative analysis holds promise in assisting researchers in gaining deeper insights into the mechanics of and distinctions between these two physiological phenomena. Additionally, the automated algorithms we devised have the potential to offer valuable insights into the clinical applications of CRC and CRPS.