The purpose of this study is to compare a modified AutoSet algorithm running on research hardware versus the traditional AutoSet algorithm running on an S7 AutoSet Spirit. The study will determine if the modified AutoSet algorithm reacts appropriately to obstructive apnoeas and hypopnoeas and provides suitable levels of positive airway pressure compared to the S7 AutoSet Spirit.

The hypothesis is that the modified AutoSet algorithm will be more efficacious than the traditional AutoSet algorithm

Subjects in the trial will spend 2 nights in a sleep laboratory being monitored with a full montage polysomnograph (PSG) with each of the algorithms providing positive airway pressure therapy. The PSG records EEG, EMG, EOG, patient airflow, respiratory effort, blood oxygen saturation, and body position.

In a randomised order each subject will spend one night being treated by the modified AutoSet algorithm and a second night being treated by the traditional AutoSet algorithm.

The subjects will be blinded to which therapy they are receiving and the scorer of the PSG will also be blinded to which therapy is being used.

Inclusion Criteria:

• PSG-confirmed diagnosis of OSA, showing an apnoea-hypopnoea index (AHI) of at least 15 events/hr and an obstructive apnoea index (OAI) of at least 5 events/hr.
• On CPAP therapy (fixed pressure or autoCPAP) for at least 1 month.
• Using a ResMed mask (this requirement applies only to phase 2 of the trial) without significant discomfort for at least 2 weeks.
• Age between 18 and 70 years.
• Written informed consent.

Exclusion Criteria:

• Unstable medical condition, including unstable angina, myocardial infarction within previous 6 months, cardiac failure NYHA grades III and IV, epilepsy, psychiatric disturbance.
• Respiratory insufficiency, defined as arterial CO2 greater than 50 mm Hg during the daytime or arterial oxygen saturation less than 90% when asleep and breathing room air on effective CPAP therapy.
• Recent severe epistaxis.
• History of spontaneous pneumothorax.
• History of regurgitation of gastric contents during sleep.
• Unable to give written informed consent.