The goal of this clinical trial is to determine whether treatment-adjustment decisions based on one night of home monitoring with the Sunrise device are comparable to decisions based on one night of in-laboratory polysomnography (PSG) in adults with obstructive sleep apnea (OSA) who remain insufficiently controlled with continuous positive airway pressure (CPAP) therapy.

The main questions it aims to answer are:

• Do Sunrise-based assessments lead to similar therapeutic decisions as PSG-based assessments?
• Are residual apnea-hypopnea indices measured by Sunrise comparable to those measured by PSG?

Participants will complete both assessment sequences in a randomized cross-over design. They will:

• Use the Sunrise device for several nights with and without CPAP.
• Undergo one night of PSG with and without CPAP.
• Have their CPAP therapy reviewed based on the results of each assessment method.

Obstructive sleep apnea (OSA) affects nearly one billion adults worldwide and is associated with recurrent upper-airway obstruction during sleep, leading to intermittent hypoxemia and multiple cardiometabolic and neurocognitive comorbidities. Continuous positive airway pressure (CPAP) is the standard treatment and is widely prescribed, with more than 1.8 million users in France. Despite its proven efficacy, large-scale analyses show that over 10% of treated patients continue to exhibit elevated residual apnea-hypopnea indices (AHI), a situation associated with reduced adherence and premature discontinuation of therapy.

Evaluation of CPAP effectiveness typically relies on in-laboratory polysomnography (PSG), which remains the reference method for characterizing residual respiratory events and sleep architecture. However, the need for PSG generates substantial healthcare costs, long waiting times, and unequal access to diagnostic resources.

The Sunrise device is a CE-marked medical device that records mandibular movements during sleep and uses artificial intelligence to characterize respiratory events and sleep parameters. Multiple validation studies have demonstrated strong agreement between Sunrise-derived metrics and PSG.

This randomized cross-over trial investigates whether Sunrise can be used to guide therapeutic decisions in patients with insufficiently controlled OSA under CPAP therapy, defined by a residual AHI greater than 10 events per hour. Participants are randomized to one of two sequences:

1. Sunrise assessments first, consisting of several nights with and without CPAP, followed by one night of PSG with and without CPAP; or
2. PSG assessments first, followed by the Sunrise sequence.

This design allows each participant to serve as their own control and enables direct comparison of treatment-adjustment decisions derived from Sunrise versus PSG. The study aims to determine whether a home-based, low-burden assessment could replace or reduce the need for in-laboratory evaluations, thereby improving access to care, reducing costs, and facilitating personalized CPAP management.

Inclusion Criteria:

• Adults aged 18 to 85 years
• Patients treated with Resmed CPAP
• Residual apnea-hypopnea index under CPAP (rAHI(CPAP) > 10 events/hour)
• Access to a smartphone and home internet connection, and ability to use a mobile application
• Signed informed consent
• Affiliated with, or beneficiary of, a national health insurance program

Exclusion Criteria:

• Severe chronic obstructive or restrictive pulmonary disease, with or without supplemental oxygen
• Unstable cardiovascular disease or severe heart failure requiring hospitalization within the past 3 months, or meeting New York Heart Association (NYHA) class III or IV criteria
• Conditions affecting mandibular condyle rotation or temporomandibular joint function
• Refusal to shave excessive facial hair when required for proper placement of the Sunrise device
• Participation in another interventional clinical study or currently within an exclusion period of another study
• Individuals who, in the investigator's judgment, are unlikely to comply with study procedures or requirements
• Individuals who cannot be contacted in case of emergency
• Persons protected under French Public Health Code articles L1121-5, L1121-6, and L1121-8 (including pregnant or breastfeeding women, persons deprived of liberty, individuals under psychiatric care under articles L.3212-1 or L.3213-1, individuals admitted to healthcare or social institutions for reasons other than research, and persons under legal protection or unable to provide informed consent)
• Staff members with a hierarchical relationship to the principal investigator