Sleep-disordered breathing (SDB) is a condition characterized by respiratory events during sleep that induce cortical arousals and alter autonomic markers. However, limited research has been conducted on detecting respiratory arousal using machine learning (ML) algorithms based on photoplethysmogra-phy (PPG). This paper presents a modern pipeline that explores the potential of PPG signals to determine respiratory arousal in suspected obstructive sleep apnea (OSA) patients. We collected PPG signals from 48 suspected OSA patients and conducted respiratory arousal detection using random forest (RF), extreme gradient boosting (XGBoost), and k-nearest neighbors (k-NN) algorithms with two feature extraction techniques. Specifically, we investigated morphological and physiological features from PPG signals to identify arousal events. Our results indicate that the XGBoost algorithm, utilizing PPG-based morphological features, achieved outstanding accuracy and F1-score (89.83±6.58% and 85.52±8.92%, respectively) for respiratory arousal event classi-fication during sleep. The findings indicate that the proposed method for detecting respiratory arousal based on PPG can be an acceptable alternative to traditional detection systems based on electroencephalogram (EEG).