Obstructive Sleep Apnea Detection Using Difference in Feature and Modified Minimum Distance Classifier.

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Signal Processing

Background:

• Obstructive Sleep Apnea (OSA) diagnosis relies on Polysomnography (PSG), an uncomfortable overnight hospital procedure.
• Snoring acoustics contain indicators for OSA, but existing algorithms are complex and data-intensive.
• There is a need for simpler, more efficient OSA detection methods.

Purpose of the Study:

• To develop a novel, simplified algorithm for Obstructive Sleep Apnea (OSA) detection using acoustic snoring features.
• To improve the accuracy and efficiency of OSA diagnosis compared to existing methods.
• To offer a more patient-friendly alternative to traditional Polysomnography (PSG).

Main Methods:

• Proposed a new algorithm utilizing a single acoustic feature: the difference between two Mel-frequency cepstral coefficients (MFCCs) from snoring sounds.
• Developed a classifier based on a modified minimum distance criterion.
• Tested the algorithm using patient data.

Main Results:

• The proposed algorithm demonstrated superior performance compared to existing methods.
• Achieved a high detection accuracy of up to 97.1% for Obstructive Sleep Apnea.
• The single-feature, simple-classifier approach proved effective for overnight data analysis.

Conclusions:

• The novel acoustic-based algorithm offers a highly accurate and efficient method for Obstructive Sleep Apnea (OSA) detection.
• This approach provides a more comfortable and less disruptive diagnostic alternative to Polysomnography (PSG).
• The findings suggest potential for widespread clinical application in OSA screening and diagnosis.