You Snooze, You Win: the PhysioNet/Computing in Cardiology Challenge 2018

Area of Science:

• Biomedical Engineering
• Sleep Medicine
• Signal Processing

Background:

• Polysomnography (PSG) is crucial for sleep studies.
• Detecting non-apnea sleep arousals is clinically significant.
• Physiological signals offer insights into sleep disturbances.

Purpose of the Study:

• To develop algorithms for detecting sleep arousals using PSG signals.
• To evaluate various computational methods for arousal detection.
• To benchmark algorithm performance using the area under the precision-recall curve.

Main Methods:

• Utilized a large dataset of 1,983 polysomnographic recordings.
• Employed diverse signal processing and machine learning techniques.
• Included methods ranging from generalized linear models to deep neural networks.

Main Results:

• Twenty-two independent teams developed and submitted algorithms.
• Algorithms were trained on 994 labeled arousal recordings.
• Performance was assessed on a hidden test set of 989 recordings.

Conclusions:

• The challenge highlighted the potential of computational approaches in sleep arousal detection.
• Advanced methods, including deep neural networks, were explored.
• The competition provided a benchmark for future research in sleep analysis.