Patch-Type Heart Rate Variability Analysis with Artificial Intelligence for Detection of Obstructive Sleep Apnea.

Area of Science:

• Cardiology
• Sleep Medicine
• Biomedical Engineering

Background:

• Obstructive sleep apnea (OSA) is a prevalent condition in Taiwan, impacting millions.
• Current screening tools like oximeters and ApneaLink® may compromise sleep quality and lack precision.
• There is a need for more accurate and user-friendly OSA screening methods.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for accurate obstructive sleep apnea (OSA) screening.
• To evaluate the efficacy of a patch-type heart rate variability (HRV) analyzer combined with AI for OSA detection.
• To compare the AI model's performance against traditional screening methods.

Main Methods:

• Enrolled 277 adults with observed snoring, performing home sleep apnea testing (HSAT) with ApneaLink® and simultaneous overnight HRV monitoring.
• Processed HRV indices from ECG signals using time-, frequency-, and nonlinear-domain analyses on 86 subjects after data quality control.
• Developed an AI model incorporating a novel Cardiovascular Hypopnea Index (CVHI) using leave-one-out validation.

Main Results:

• The AI model achieved 81.4% accuracy in screening for obstructive sleep apnea (OSA).
• This performance surpassed demographic-based (73%) and previous ECG-based (70.6%) screening methods.
• The model demonstrated strong classification for moderate-to-severe OSA (AUC >0.8) at an apnea-hypopnea index (AHI) cutoff of 15.

Conclusions:

• A patch-type HRV analyzer coupled with AI analysis offers accurate, low-interference screening for OSA.
• This approach is well-suited for large-scale clinical implementation and home-based monitoring.
• The AI-driven HRV analysis presents a promising advancement in OSA detection.