This study aims to improve treatment strategies for Obstructive Sleep Apnea (OSA), a disorder characterized by recurrent upper airway collapse during sleep, resulting in reduced oxygenation, sleep fragmentation, and excessive daytime sleepiness. The objectives are twofold: to evaluate whether an artificial intelligence (AI)-based model can accurately predict the most effective treatment for individual patients, and to assess whether a mobile health application can enhance adherence to oropharyngeal rehabilitation (OPR) and improve therapeutic outcomes.
The study will be conducted in two phases. In Phase I, a retrospective analysis will be performed using a large dataset of polysomnography (PSG) records obtained from the Sleep Center at National Cheng Kung University Hospital. Machine learning algorithms will be applied to identify predictive features that differentiate responders from non-responders across Continuous Positive Airway Pressure (CPAP), surgical, and OPR interventions. These findings will inform the development of a predictive treatment recommendation model.
In Phase II, a prospective clinical trial will validate the predictive accuracy and clinical utility of the model. Patients newly diagnosed with OSA will be assigned to CPAP, surgery, or OPR interventions according to the model's recommendations, in combination with physician judgment and patient preference. Each intervention will last 12 weeks, followed by repeat PSG and clinical assessments. Within the OPR arm, participants will be further randomized to monitor adherence via an exercise diary or a smartphone application equipped with a pressure sensor and facial motion recognition technology, enabling real-time feedback and remote monitoring.
This trial is expected to determine whether AI can provide clinically reliable treatment recommendations and whether digital telerehabilitation can improve adherence and outcomes, thereby advancing precision medicine in OSA management.