Sleep staging is a critical component in assessing sleep quality and diagnosing sleep-related brain disorders. In alignment with the "China Brain Project" and the strategic goals for brain science in the upcoming national "15th Five-Year Plan," the demand for precise, non-invasive, and home-based brain function assessment has reached a new height at the intersection of neuroscience and artificial intelligence.
The Challenge: Beyond Traditional Polysomnography
Currently, polysomnography (PSG) remains the gold standard for sleep staging. By recording multiple physiological signals—including electroencephalograms (EEG), electrocardiograms (ECG), and electromyograms (EMG)—it provides a comprehensive view of sleep architecture. However, manual interpretation of these signals is often subjective, time-consuming, and labor-intensive.
While deep learning has been applied to automate this process, existing models frequently rely on complex multi-channel data or EEG signals that require intrusive equipment, causing discomfort for patients and limiting their use in home environments.
The Breakthrough: SleepECGFusion
Recently, Professor Yi Zhou's team published a study titled "SleepECGFusion: A Cross-Modal Deep Learning Framework for Automatic Sleep Stage Classification using Single-Lead ECG" in Knowledge-Based Systems (a CAS Q1 Top Journal).
The team proposed SleepECGFusion, an innovative cross-modal framework that integrates information from two complementary domains using only single-lead ECG signals. This method significantly enhances comfort and convenience for long-term monitoring.
Furthermore, the study verifies the hypothesis that increasing the duration of input signals consistently improves classification accuracy across all sleep stages. SleepECGFusion achieved superior performance in ECG-based sleep staging tasks compared to previous studies and demonstrated excellent transferability. Most importantly, the framework provides robust and comparable classification results for both healthy individuals and patients with sleep apnea.
Professor Yi Zhou serves as the corresponding author, with doctoral student Xuanhao Qi (Class of 2024) as the first author. This research marks a significant step forward in making high-precision sleep monitoring accessible outside of clinical settings.
Moving forward, the team aims to integrate additional physiological signals to improve robustness in noisy environments and explore time-contrastive learning to better handle temporal variability in long-term monitoring. The framework's potential application in other neurological disorders remains a key area of future exploration.
Prof. Yi Zhou's research group is deeply committed to health and medical informatics, big data, and medical AI. The team warmly welcomes applications from postdoctoral fellows and graduate students who are passionate about these cutting-edge fields.
Link to the paper: https://www.sciencedirect.com/science/article/pii/S0950705125021677
Source: Zhongshan School of Medicine
