PSGMiner: A modular software for polysomnographic analysis

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Sleep Medicine

Background:

• Sleep disorders are prevalent, with diagnosis relying on polysomnography (PSG).
• Analyzing PSG data for sleep events requires specialized tools for robust feature extraction and classification.
• Existing commercial software often lacks flexibility and advanced analytical capabilities for novel research.

Purpose of the Study:

• To develop novel software, PSGMiner, for comprehensive analysis and classification of sleep events from polysomnographic data.
• To provide researchers with a flexible platform for hypothesis testing and correlation analysis beyond the scope of commercial software.
• To offer a freely available tool under the GPL3 License for the scientific community.

Main Methods:

• PSGMiner integrates diverse modules including feature extraction, annotation, and machine learning.
• Utilizes digital signal processing and statistical methods for extracting polysomnography features.
• Incorporates five classification algorithms for feature analysis and offers an extensible architecture for new methods.

Main Results:

• The software facilitates visualization, processing, and classification of bioelectrical data from polysomnography.
• Enables extraction and analysis of features from all polysomnographic signals and events, unlike specialized commercial tools.
• Provides a simplified interface for researchers without extensive programming expertise.

Conclusions:

• PSGMiner simplifies polysomnographic signal processing for researchers and clinicians.
• The software can identify correlations between events, potentially aiding in predicting conditions like sleep apnea.
• PSGMiner serves as a valuable tool for both research and educational purposes in sleep science.