Abstract
A method of predicting an unintentional leak in a respiratory system during a sleep session includes causing, during the sleep session, pressurized air to be delivered from a respiratory device to a user via a conduit coupled to a user interface. The method also includes receiving historical first data associated with pressurized air delivered from the respiratory device during one or more prior sleep sessions; receiving current first data associated with the pressurized air being delivered from the respiratory device during the sleep session; receiving historical second data associated with the user during one or more prior sleep sessions; and receiving, via one or more second sensors, current second data associated with the user during the current sleep session. The method determines, a likelihood that an unintentional leak in the respiratory system will occur within a predetermined amount of time.