Open-source logic-based automated sleep scoring software using electrophysiological recordings in rats

Area of Science:

• Neuroscience
• Computational Biology
• Sleep Science

Background:

• Manual scoring of physiological sleep recordings is labor-intensive, causing delays and increased costs.
• Automating sleep state scoring is crucial for efficient sleep research.

Purpose of the Study:

• To develop and validate MATLAB-based software for automated scoring of sleep/waking states in rats.
• To reduce the time and personnel costs associated with sleep study data analysis.

Main Methods:

• Developed two programs: Sleep Scorer (manual) and Auto-Scorer (automated).
• Auto-Scorer analyzes electroencephalographic (EEG) and electromyographic (EMG) signals using defined thresholds and logic.
• Evaluated Auto-Scorer performance against manual scoring across 6 rats and 2 conditions by 3 users.

Main Results:

• Auto-Scorer reduced scoring time by 75% compared to manual scoring.
• Mean agreement between Auto-Scorer and manual scoring was 80.24% ± 7.87%.
• Agreement between Auto-Scorer and manual scoring was comparable to inter-rater agreement among manual scorers (83.03% ± 4.00%).

Conclusions:

• The open-source Auto-Scorer provides a rapid and accurate method for scoring sleep/waking states in rat recordings.
• The software is potentially extendable to other animal models and recording systems.
• Automated scoring, with user review, can significantly accelerate sleep research.