This post presents an analysis of sleep stage classification validation using polysomnography (PSG) as the ground truth and a headband-based device as the test instrument. The workflow includes generating synthetic sleep data, loading real-world datasets, and comparing sleep stage distributions between the two devices. The validation process includes Bland-Altman analysis, Pearson correlation, Cohen’s Kappa, and statistical measures of agreement. The results provide insight into the accuracy and bias of wearable sleep-tracking devices in sleep stage classification.
