Real-Time Snoring Detection Using Deep Learning: A Home-Based Smartphone Approach for Sleep Monitoring.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Sleep Medicine

Background:

• Sleep-related disorders are common, yet effective home-based snoring detection methods are limited.
• Previous research has not extensively explored deep learning for snoring prediction using smartphone audio.
• This study addresses the need for accessible tools for sleep disorder monitoring.

Purpose of the Study:

• To develop and evaluate a real-time snoring detection method using a Vision Transformer-based deep learning model.
• To assess the feasibility of utilizing smartphone recordings for snoring detection in home environments.
• To provide a scalable and accurate solution for monitoring sleep-related breathing issues.

Main Methods:

• Smartphone audio recordings of sleep-breathing sounds were collected concurrently with polysomnography (PSG).
• A Vision Transformer deep learning model was trained and tested on annotated sleep epochs.
• Model performance was assessed using epoch-by-epoch accuracy and correlation of snoring ratios.

Main Results:

• The model achieved high performance on the test dataset with 89.8% sensitivity and 91.3% specificity.
• A strong correlation (r = 0.97) was observed between predicted and actual snoring ratios.
• The study included 214 participants, analyzing 85,600 sleep epochs.

Conclusions:

• Deep learning models can effectively detect snoring from home-recorded smartphone audio in real-time.
• This approach offers a practical, accurate, and scalable tool for managing sleep-related disorders.
• The findings support the development of home-based sleep health management solutions.