A noisy label correction framework for apnea-hypopnea index estimation from sleep breathing sounds

Key Points

• Estimation accuracy of the apnea-hypopnea index is significantly improved with a noise reduction approach.
• The developed framework utilizes a novel label correction method to enhance estimation from sleep breathing sounds.
• Application of machine learning algorithms helps achieve robust detection of sleep-related breathing events.
• Implications of this framework suggest it could lead to better diagnosis and treatment strategies in sleep medicine.