Discrimination of cheyne-stokes breathing patterns
A method of a processor for detecting a presence of Cheyne-Stokes respiration from a respiration signal includes accessing data representative of a respiration signal. Data is assessed to detect apnea and/or hypopnea events. A cycle length histogram is determined based on the events and an incident of Cheyne-Stokes respiration is detected based on the cycle length histogram.
1. A method implemented by a processor for detecting a presence of Cheyne-Stokes respiration from a respiration signal generated by a respiration sensor, the method comprising:
accessing data representative of the respiration signal, the accessed data comprising respiratory data;
assessing the accessed data to detect apnea and/or hypopnea events;
evaluating the respiratory data directly following the detected events to estimate a shape feature representing a change in the respiratory data;
generating a cycle length histogram based on the detected events, wherein the cycle length histogram includes a plurality of bins associated with a plurality of cycle lengths, each cycle length of the plurality being determined from the detected apnea and/or hypopnea events;
detecting an incident of Cheyne-Stokes respiration based on the cycle length histogram and the estimated shape feature, wherein detecting the incident of Cheyne-Stokes respiration comprises calculating power over a combination of bins covering a select set of cycle lengths; and
responding to the detection of an incident of Cheyne-Stokes respiration, the responding comprising
controlling an adjustment of a therapeutic pressure delivered by a respiratory treatment apparatus.
2. The method of claim 1 , wherein each of the plurality of bins has a midpoint and a bin width.
3. The method of claim 1 , wherein the plurality of bins are evenly spaced.
4. The method of claim 1 wherein assessing the accessed data to detect apnea and/or hypopnea events comprises determining a duration of each event.
5. The method of claim 1 , wherein the shape feature represents at least one of a rise and a fall of a breathing drive of a patient.
6. The method of claim 1 , wherein the shape feature is a jump feature.
7. The method of claim 6 , wherein the jump feature is calculated by selecting a first peak of the respiratory data, selecting a second peak at a predetermined ratio of the first peak and calculating a gradient between the first peak and the second peak.
8. The method of claim 7 , further comprising scaling the gradient between the first peak and the second peak.
9. The method of claim 5 , wherein the shape feature is estimated by fitting an approximating function to the respiratory data.
10. The method of claim 1 , wherein the respiratory data comprises values of respiratory flow peaks or values of tidal volumes.
11. A method implemented by a processor for detecting a presence of Cheyne-Stokes respiration from a respiration signal generated by a respiration sensor, the method comprising:
accessing respiratory flow data representative of the respiration signal;
assessing the accessed respiratory flow data to detect apnea and/or hypopnea events;
determining a cycle length histogram based on the detected events, wherein the cycle length histogram includes a plurality of bins associated with a plurality of cycle lengths, each cycle length of the plurality being determined from the detected apnea and/or hypopnea events;
evaluating peaks in the flow data directly following the detected events to estimate a shape feature representing a change in the peaks of the flow data;
detecting an incident of Cheyne-Stokes respiration based on the cycle length histogram and the shape feature, wherein detecting the incident of Cheyne Stokes respiration comprises calculating power over a combination of bins covering a select set of cycle lengths; and
responding to the detecting of an incident of Cheyne-Stokes respiration, the responding comprising
controlling an adjustment to a therapeutic pressure delivered by a respiratory treatment apparatus.
12. The method of claim 11 , further comprising normalizing the cycle length histogram and the shape feature by converting them into a probability space with a range of values between 0 and 1.
13. The method of claim 11 , wherein assessing the accessed respiratory flow data to detect apnea and/or hypopnea events comprises calculating duration of at least one apneic period and at least one cycle length and further comprising calculating a duty cycle based on the duration of the at least one apneic period and the at least one cycle length.
14. The method of claim 11 , wherein detecting the incident of Cheyne-Stokes respiration comprises determining a Cheyne-Stokes respiration probability using the shape feature, cycle length and the power.
15. The method of claim 14 , further comprising determining an overall Cheyne-Stoke respiration probability over an entire sleep period by combining weighted Cheyne-Stokes respiration probability for multiple selected periods.
16. The method of claim 11 , wherein the shape feature represents at least one of a rise and a fall of a breathing drive of a patient.
17. The method of claim 1 wherein detecting of the incident of Cheyne-Stokes respiration comprises applying the estimated shape feature and the cycle length histogram to a classifier implemented by the processor.
18. The method of claim 1 wherein the controlling an adjustment of a therapeutic pressure delivered by a respiratory treatment apparatus further comprises automatically controlling a blower of a flow generator in response to the detecting of the incident of Cheyne-Stokes respiration that is based on the cycle length histogram and the estimated shape feature.
19. The method of claim 1 wherein calculating power evaluates a histogram power transformation function that determines a first count of a first bin of the plurality of bins as a maximum count and a second count of a second bin of the plurality of bins as a second highest count, and adds the first count and second count when midpoints of first bin and the second bin are each within a predefined range.
20. The method of claim 19 wherein the histogram power transformation function determines a square root of an addition of a square of the first count and a square of the second count when a midpoint of the first bin is within the predefined range and a midpoint of the second bin is outside the predefined range.
21. The method of claim 11 wherein detecting of the incident of Cheyne-Stokes respiration comprises applying the estimated shape feature and the cycle length histogram to a classifier implemented by the processor.
22. The method of claim 11 wherein the controlling the adjustment to a therapeutic pressure delivered by a respiratory treatment apparatus further comprises automatically controlling a blower of a flow generator in response to the detecting of the incident of Cheyne-Stokes respiration that is based on the cycle length histogram and the estimated shape feature.
23. The method of claim 11 wherein calculating power evaluates a histogram power transformation function that determines a first count of a first bin of the cycle length histogram as a maximum count and a second count of a second bin of the cycle length histogram as a second highest count, and adds the first count and second count when midpoints of first bin and the second bin are each within a predefined range.
24. The method of claim 23 wherein the histogram power transformation function determines a square root of an addition of a square of the first count and a square of the second count when a midpoint of the first bin is within the predefined range and a midpoint of the second bin is outside the predefined range.
25. The method of claim 1 wherein the responding further comprises generating and displaying an incident report upon a monitor.
26. The method of claim 11 wherein the response further comprises generating and displaying an incident report upon a monitor.