An Automated Disease Prediction System for Sleep Behaviour Analytics Using Deep Learning

Abstract

Sleep is an essential physiological function that plays a critical role in maintaining brain activity, cardiovascular health, and respiratory function. Disturbances in sleep have been associated with a wide range of disorders, including cardiovascular, metabolic, and neurodegenerative diseases. Although polysomnography (PSG) provides rich multimodal physiological data for sleep assessment, existing automated sleep analysis approaches often suffer from limited generalization capability, inadequate multitask learning performance, and poor explainability. To address these challenges, this study proposes **NeuroSleepX**, a deep learning framework that leverages multimodal PSG signals, including EEG, EOG, EMG, ECG, and respiratory data, for simultaneous sleep stage classification and disease risk prediction. NeuroSleepX combines the strengths of 1D Convolutional Neural