Lightweight 1-D Convolutional Neural Network Model for Predicting Snoring Duration

This paper introduces an approach to snoring duration prediction utilizing a custom 1-D Convolutional Neural Network (CNN) implemented into TensorFlow Lite. Snoring is a prevalent symptom of sleep disorders which presents challenges in diagnosis and management due to its disruptive effects on sleep quality and potential indication of underlying health issues like obstructive sleep apnea (OSA). Despite the significance of accurate snoring prediction and monitoring, existing research lacks specific exploration of lightweight and accurate CNN models tailored for this purpose. Leveraging the advantages of 1-D CNNs, such as efficient processing of 1D signals without extensive preprocessing, this paper addresses the research gap by proposing a custom TensorFlow Lite-based model for snoring prediction. Through a comprehensive literature review, the efficiency of CNNs in predictive modeling across various domains is highlighted, underscoring the need for specialized models in healthcare applications like snoring prediction. The proposed approach not only supports the gap in existing research but also opens possibilities for lightweight, real-time snoring monitoring, potentially integrated into mobile or wearable devices for continuous assessment of sleep-related disorders.