Figure 1.
Scheme of the LMS adaptive filtering to separate respiratory influences from the tachogram.
Figure 2.
Block diagram of the simulation study.
Figure 3.
Example of the simulation study of a typical subject using the ARMAX model in step 1 and 2.
Separation of the tachogram is performed using OSP. The signals are shown in both time (only 120 s are shown here) and frequency domain. From top to bottom: ,
,
,
. In dashed black, the estimated components using OSP are displayed.
Figure 4.
Boxplots of the normalized root-mean-squared errors (NRMSE) between and
(left), and between
and
(right) obtained using the simulation study.
Outliers are not shown.
Figure 5.
Boxplots of the squared errors (SE) in LF power (top) and HF power (bottom) between and
(left), and between
and
(right) obtained using the simulation study.
Outliers are not shown.
Figure 6.
Block diagram of the stability study.
Figure 7.
Boxplots of the normalized root mean-squared errors (NRMSE) (left), and the squared errors (SE) (right) in LF power (top) and HF power (bottom) between and
obtained using the stability study.
Outliers are not shown.
Figure 8.
Example of stability study using orthogonal subspace projection.
The residual signal of the full 6-minute period (, thick grey) and the three 2-minute segments (
, dashed black) are shown.
Figure 9.
Mean ROC of all classifiers.