Revealing Sleep Dynamics with PCT-CRV: A Novel Approach for Automatic Sleep Staging and Tracking Transitions using...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Sleep Medicine

Background:

• Accurate sleep staging via polysomnography (PSG) is crucial for diagnosing sleep disorders.
• Existing automatic sleep staging methods often overlook time-frequency synchronization, local/global features, and inter-stage transition patterns.

Purpose of the Study:

• To propose a novel framework, the polynomial chirplet transform-derived characteristic response vector (PCT-CRV), for enhanced sleep stage assessment.
• To address limitations in current methods by incorporating time-frequency domain synchronization, comprehensive feature extraction, and transition tracking.

Main Methods:

• Utilized time-domain PCT (TPCT) and frequency-domain PCT (FPCT) to improve time-frequency representations of PSG signals.
• Constructed characteristic response vectors (CRVs) from PCT representations by analyzing correlation matrices and eigenvectors.
• Extracted local and global features from PCT-CRVs and applied machine learning classifiers for sleep stage classification.

Main Results:

• The proposed PCT-CRV method demonstrated superior performance across three independent datasets compared to existing techniques.
• PCT-CRV outperformed traditional wavelet-based and synchrosqueezed-based CRV methods.
• Sub-band PCT-CRVs effectively tracked transitions between sleep stages, highlighting the method's dynamic assessment capabilities.

Conclusions:

• The PCT-CRV framework offers a significant advancement in automatic sleep staging accuracy and reliability.
• The method's ability to capture complex signal patterns and track sleep stage transitions provides a more comprehensive sleep analysis.
• This approach holds promise for improved diagnosis and management of sleep-related disorders.