<p>Sleep apnea, a serious sleep disorder, causes abnormal breathing pauses and affects neural activity that is also affected by the associated sleep stage. Electroencephalography (EEG), widely used for neural activity analysis, can be more effective for apnea detection when combined with sleep stage information. In this paper, a machine learning-based apnea detection method is proposed using EEG signal and sleep stage information, considering both subjects-combined and subject-independent classification scenarios. First, a sub-frame-based multi-band feature extraction scheme is designed to extract the temporal feature variation pattern within an EEG frame. The feature variation patterns are further processed using statistical analysis. The resulting features are used in the K-nearest neighbor (KNN) classifier for sleep stage classification. The detected sleep stage information is utilized as a potential feature by combining it with some previously extracted features to form the final feature vector for sleep apnea detection through the KNN classifier. Two publicly available databases are used for experimentation and performance analysis. Using stage information along with other statistical feature variation pattern analyses, the proposed method gives 91.45% sensitivity, 85.62% specificity, and 88.53% accuracy under 10-fold cross-validation considering all subjects, which are better than those of the existing methods.</p>