Exploring the Link Between Brain Waves and Sleep Patterns with Deep Learning Manifold Alignment.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Data Science

Background:

• Medical data frequently combines diverse formats like text, images, and audio, posing challenges in analyzing inter-modal relationships.
• Polysomnography (PSG) datasets, comprising hypnograms, electrocardiograms (ECG), and electroencephalograms (EEG), are crucial for sleep disorder research but exhibit complex inter-modal associations.

Purpose of the Study:

• To explore the intricate relationship between sleep architecture and electroencephalogram (EEG) features using a novel deep learning manifold alignment technique.
• To assess the efficacy of deep learning manifold alignment in identifying associations between sleep architecture and EEG data, compared to conventional statistical methods.

Main Methods:

• Leveraged a deep learning manifold alignment method to analyze multi-modal Polysomnography (PSG) datasets.
• Investigated the associations between sleep architecture and electroencephalogram (EEG) features within PSG data.

Main Results:

• The deep learning manifold alignment method successfully identified significant associations between sleep architecture and EEG datasets.
• These findings align with previous research utilizing PSG data for diagnosing sleep disorders and monitoring sleep quality.
• The Spearman correlation method failed to detect the correlations identified by the deep learning approach.

Conclusions:

• Deep learning manifold alignment offers a powerful tool for uncovering hidden relationships within multi-modal medical data, specifically in sleep research.
• This method enhances our understanding of sleep stages and brain activity dynamics, surpassing the capabilities of traditional statistical techniques like Spearman correlation.