Flow-based sleep stage determination

Abstract

A method and system of determining a sleep stage of a user involves receiving a respiratory flow signal of a user, obtaining at least one respiratory feature from at least part of the respiratory flow signal, and determining a sleep stage from the at least one respiratory feature.

20 . A method of automatically determining sleep quality and effectiveness of a user using a respiratory system configured to provide respiratory therapy to the user, the method comprising:
 using a controller of the respiratory system and one or more sensors of the respiratory system:
 receiving a respiratory flow signal of the user from the one or more sensors; 
 obtaining one or more measurements from the respiratory flow signal; 
 automatically determining a sleep stage of the user from the one or more measurements; and 
 calculating a sleep score based at least in part on the determined sleep stage of the user. 
   
     
     
         21 . The method of  claim 20 , wherein the sleep score is calculated further based on one or more metrics: sleep efficiency, deep sleep time, rapid eye movement (REM) time, or sleep fragmentation. 
     
     
         22 . The method of  claim 21 , wherein the sleep efficiency is defined as total sleep time over total time in bed. 
     
     
         23 . The method of  claim 21 , wherein the sleep fragmentation is defined as a number of times arousals or awakes of the user occur during the night. 
     
     
         24 . The method of  claim 20 , wherein the sleep score is calculated further based on one or more indices: sleep time (ST), total time at deep sleep (DST), total time at REM (REMT), number of sleep disruptions (NSD), or number of sleep disordered breathing (SDB) events (NSDB). 
     
     
         25 . The method of  claim 24 , wherein the sleep score is calculated using an equation X1*ST+X2*DST+X3*REMT+X4*NSD+X5*NSDB, where X1-X5 are coefficients. 
     
     
         26 . The method of  claim 25 , where X1-X5 are each set to 0.2. 
     
     
         27 . The method of  claim 25 , where the sleep score is between 0 to 1. 
     
     
         28 . The method of  claim 20 , wherein the one or more measurements comprise at least one centre of mass measurement, at least one duration measurement, at least one amplitude measurement, or at least one volume related measurement. 
     
     
         29 . The method of  claim 20 , further comprising:
 identifying, within the respiratory flow signal, at least one breath signal representing a breath of the user; and   obtaining at least one breath measurement from a portion of the respiratory flow signal within which the at least one breath signal is identified.   
     
     
         30 . The method of  claim 20 , further comprising:
 identifying, within the respiratory flow signal, a window containing a plurality of breath signals; and   obtaining respective breath measurements of the plurality of breath signals within the window.   
     
     
         31 . The method of  claim 30 , wherein automatically determining the sleep stage of the user further comprises automatically determining the sleep stage of the user from at least one of a mean or a standard deviation of the breath measurements within at least part of the window. 
     
     
         32 . The method of  claim 20 , further comprising determining the sleep stage at least partly by applying at least one of a supervised learning algorithm, an unsupervised learning algorithm, or a semi-supervised learning algorithm. 
     
     
         33 . The method of  claim 20 , wherein the sleep stage is one of a plurality of sleep stages configured to be determined by the method, the plurality of sleep stages comprising:
 awake, N1, N2, N3, and REM;   awake, light sleep, deep sleep, and REM;   awake, non-REM, and REM; or   awake and asleep.   
     
     
         34 . The method of  claim 20 , further comprising implementing a context module to apply contextual clues to improve accuracy of determining the sleep quality and effectiveness of the user. 
     
     
         35 . The method of  claim 34 , wherein the contextual clues comprise prolonged periods without breathing, the method further comprising determining the sleep stage is awake in response to detecting a prolonged period without breathing. 
     
     
         36 . The method of  claim 34 , wherein the contextual clues comprise SDB events, the method further comprising assuming the user is asleep prior to a detected SDB event and determining the sleep stage is light sleep, deep sleep, or REM in response to the detected SDB event. 
     
     
         37 . The method of  claim 34 , wherein the contextual clues comprise neighboring consensus, the method further comprising ignoring individual short periods of the respiratory flow signal that differ from neighboring portions of the respiratory flow signal. 
     
     
         38 . The method of  claim 20 , further comprising using one or more machine learning methods to identify markers in the respiratory flow signal for determining the sleep stage of the user. 
     
     
         39 . The method of  claim 38 , wherein the machine learning methods are configured to adapt determining of the sleep stage to specific user profiles.