Diagnosis, monitoring, and treatment of respiratory disorders

Abstract

Automation for a system and/or method detects and/or controls treatment of inspiratory flow limitation. The system may include a flow rate sensor configured to generate a signal representing a respiratory flow rate of a patient. It may include a recording device configured to record the generated respiratory flow rate signal during a diagnosis session. It may include a computing device 7040 configured to detect a degree of inspiratory flow limitation of the patient on the recorded respiratory flow rate signal. The method may include extracting an inspiratory portion of each breath during a detection and/or monitoring session from a respiratory flow rate signal of the patient, calculating a feature vector from each inspiratory flow portion, labelling each feature vector as flow limited or not flow limited, and/or computing a metric based on the labels, the metric indicating the degree of inspiratory flow limitation of the patient during the session.

1 . A method comprising:
 extracting inspiratory portions of breaths of a patient from respiratory flow rate signals generated by a flow rate sensor;   calculating feature vectors representative of the inspiratory portions;   applying labels to the inspiratory portions based on the feature vectors, wherein the labels indicate whether the inspiratory portions are flow limited; and   adjusting a treatment pressure of a respiratory pressure therapy being delivered to the patient by a respiratory pressure therapy device based on the labels applied to the inspiratory portions.   
     
     
         2 . The method of  claim 1 , wherein the feature vectors are non-negative matrix factorisation (NMF) feature vectors. 
     
     
         3 . The method of  claim 2 , wherein calculating the NMF feature vectors comprises multiplying the inspiratory portions by pseudoinverses of factor matrices. 
     
     
         4 . The method of  claim 3 , wherein each factor matrix is a left factor matrix of a non-negative matrix factorisation of a database matrix of inspiratory portions. 
     
     
         5 . The method of  claim 4 , wherein the non-negative matrix factorisation is based on sparsity constraints of the left factor matrix and a right factor matrix. 
     
     
         6 . The method of  claim 3 , wherein the pseudoinverses are left inverses of the factor matrices. 
     
     
         7 . The method of  claim 1 , wherein the feature vectors are difference filter (DF) feature vectors. 
     
     
         8 . The method of  claim 7 , wherein features in the DF feature vectors are calculated by summing samples of the inspiratory portions with a weighting of either +1 or −1, and wherein the weightings are coefficients of binary-valued difference filters. 
     
     
         9 . The method of  claim 1 , wherein applying the labels to the inspiratory portions comprises applying a binary classifier to the feature vectors. 
     
     
         10 . The method of  claim 9 , wherein the binary classifier is a linear classifier. 
     
     
         11 . The method of  claim 9 , wherein the binary classifier has been trained using a Support Vector Machine and a plurality of labelled inspiratory portions. 
     
     
         12 . The method of  claim 9 , wherein the binary classifier has been trained by optimising an objective function over a parameter space of the classifier. 
     
     
         13 . The method of  claim 1  further comprising:
 computing a metric indicating a degree of inspiratory flow limitation of the patient based on the labels applied to the inspiratory portions. 
 
     
     
         14 . The method of  claim 13 , wherein the metric is a fraction of a total number of the breaths having an inspiratory portion to which a label has been applied that indicates the inspiratory portion is flow limited. 
     
     
         15 . The method of  claim 13 , wherein the metric is an average rate of the breaths having an inspiratory portion to which a label has been applied that indicates the inspiratory portion is flow limited. 
     
     
         16 . The method of  claim 13 , wherein the metric is a maximum rate of the breaths having an inspiratory portion to which a label has been applied that indicates the inspiratory portion is flow limited. 
     
     
         17 . The method of  claim 13 , further comprising:
 displaying, with an output device, the metric.   
     
     
         18 . The method of  claim 1 , wherein the adjusting comprises:
 increasing the treatment pressure in response to a label being applied to an inspiratory portion that indicates the inspiratory portion is flow limited; or   decreasing the treatment pressure in response in response to a label being applied to an inspiratory portion that indicates the inspiratory portion is not flow limited.   
     
     
         19 . A system comprising:
 one or more processors configured to:   extract inspiratory portions of breaths of a patient from respiratory flow rate signals generated by a flow rate sensor;   calculate feature vectors representative of the inspiratory portions;   apply labels to the inspiratory portions based on the feature vectors, wherein the labels indicate whether the inspiratory portions are flow limited; and   adjust a treatment pressure of a respiratory pressure therapy being delivered to the patient by a respiratory pressure therapy device based on the labels applied to the inspiratory portions.   
     
     
         20 . The system of  claim 19  further comprising:
 the flow rate sensor configured to generate the respiratory flow rate signals. 
 
     
     
         21 . The system of  claim 20 , wherein the flow rate sensor comprises a nasal cannula in communication with a pressure transducer, and the respiratory flow rate signals are nasal pressure signals generated by the pressure transducer. 
     
     
         22 . The system of  claim 19  further comprising:
 the respiratory pressure therapy device configured to deliver respiratory pressure therapy to the patient. 
 
     
     
         23 . The system of  claim 22 , wherein the respiratory pressure therapy device configured to deliver respiratory pressure therapy to the patient comprises the one or more processors. 
     
     
         24 . An apparatus comprising:
 means for extracting inspiratory portions of breaths of a patient from respiratory flow rate signals generated by a flow rate sensor;   means for calculating feature vectors representative of the inspiratory portions;   means for applying labels to the inspiratory portions based on the feature vectors, wherein the labels indicate whether the inspiratory portions are flow limited; and   means for adjusting a treatment pressure of a respiratory pressure therapy being delivered to the patient by a respiratory pressure therapy device based on the labels applied to the inspiratory portions.