Interactive Sleep Stage Labelling Tool For Diagnosing Sleep Disorder Using Deep Learning.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Artificial Intelligence

Background:

• Manual sleep scoring is time-consuming and requires expert experience.
• Automatic methods using electrooculography (EOG) analysis are gaining attention for cost-effective sleep disorder diagnosis.
• Current computer-aided diagnosis relies on classifying 6 sleep stages (Wake, S1-S4, REM) from 30-second EOG records.

Purpose of the Study:

• To develop an automatic sleep stage scoring method leveraging deep learning advancements.
• To introduce an interactive scoring scheme for practitioner feedback and improved accuracy.
• To enhance the diagnosis of sleep disorders through more efficient and accurate scoring.

Main Methods:

• Utilized deep learning algorithms for automatic analysis of electrooculography (EOG) data.
• Implemented a 30-second epoch-based scoring system for wake, sleep status (S1-S4), and REM.
• Developed an interactive interface for clinicians to provide feedback and correct scoring errors.

Main Results:

• Demonstrated a novel deep learning approach for automatic sleep stage classification.
• Proposed an interactive system to refine scoring accuracy through expert input.
• Aimed to reduce the time and expertise needed for sleep disorder diagnosis.

Conclusions:

• Deep learning offers a promising avenue for automating sleep stage scoring from EOG.
• Interactive feedback mechanisms can significantly improve the accuracy of automated sleep scoring systems.
• This approach has the potential to lower costs and increase accessibility for sleep disorder diagnosis.