07-25-2019, 03:10 PM
Oscar bug report: Oximeter flagging wrong
If my understanding of the oximeter flags is correct, OSCAR has a bug.
Attached image preferences.png
shows that I have set the thresholds for the SpO2 and pulse change rate flags detection 10 times higher than the standard values, so there should be no detections.
As you can see in SpO2Flag.png and and PulseFlag.png (combined into flags.png, because of forum limits)
I have placed the SpO2 Flag on the SpO2 plot, and the Pulse flag on the Pulse Rate plot.
But plot.png
shows detections of both the SpO2 Flag and the Pulse rate Flag. Notice that the plotted values are within appropriate thresholds, as are the 8 second averaged derivatives.
Perhaps the fact that my pulse oximeter quantizes SpO2 and pulse rate to integral values is the problem?
Perhaps the software uses deviations from a curve fit instead of averaged first differences? But expand.png
and WholeNight.png
show that there is no larger time trend on an expanded time axis, unless the fit doesn't kill outliers that occurred substantially earlier. (I was asleep from about 1:20 AM to about 4:15 AM. The gaps in the flow rate were for using the bathroom, and messing with pillows, a ZEO EEG, and sleep related smartphone software. The gap in the oximetry data occurred because it fell off my finger.
(WholeNight.png, unlike the other images, was done with the default oximeter flag limits, and with Custom CPAP User Event Flagging turned on.)
Attached image preferences.png
shows that I have set the thresholds for the SpO2 and pulse change rate flags detection 10 times higher than the standard values, so there should be no detections.
As you can see in SpO2Flag.png and and PulseFlag.png (combined into flags.png, because of forum limits)
I have placed the SpO2 Flag on the SpO2 plot, and the Pulse flag on the Pulse Rate plot.
But plot.png
shows detections of both the SpO2 Flag and the Pulse rate Flag. Notice that the plotted values are within appropriate thresholds, as are the 8 second averaged derivatives.
Perhaps the fact that my pulse oximeter quantizes SpO2 and pulse rate to integral values is the problem?
Perhaps the software uses deviations from a curve fit instead of averaged first differences? But expand.png
and WholeNight.png
show that there is no larger time trend on an expanded time axis, unless the fit doesn't kill outliers that occurred substantially earlier. (I was asleep from about 1:20 AM to about 4:15 AM. The gaps in the flow rate were for using the bathroom, and messing with pillows, a ZEO EEG, and sleep related smartphone software. The gap in the oximetry data occurred because it fell off my finger.
(WholeNight.png, unlike the other images, was done with the default oximeter flag limits, and with Custom CPAP User Event Flagging turned on.)
