AI-Driven Sleep Apnea Detection And Prediction

Sleep apnea, a common yet underdiagnosed sleep disorder, causes a person’s breathing repeatedly stops and starts, which poses serious health risks if left untreated. Existing diagnostic methods are often inconvenient, expensive, and inaccessible to many individuals. To address these challenges, this paper introduces an AI-powered system DreamGuardian for the early detection and prediction of sleep apnea. This system utilizes a combination of machine learning algorithms and advanced signal processing techniques to analyze data from non-invasive sensors integrated into a wearable device. These sensors monitor various physiological signals during sleep, including respiratory patterns, heart rate variability, and movement, enabling comprehensive sleep analysis in real-time. The methodology that we have followed for sleep apnea detection and prediction involves acquiring a diverse dataset from sleep clinics and repositories, formatting it, and selecting key features based on existing literature. Non-invasive sensors are integrated into a wearable device prototype to monitor physiological parameters. Standardized protocols are developed for data collection, and ethical considerations are being followed. The data collected from the users will be analysed using a machine learning algorithm. The system accurately predicts the presence or absence of sleep apnea based on variables such as heart rate, BMI, age, gender, respiratory rate, ESS score, AHI, and oxygen saturation.