An evaluation is carried out to compare the performance of interpolation of the proposed algorithm with a third party software i.e., Kubios HRV which is commonly used for HVR analysis.
Kubios HRV uses cubic spline interpolation . Five levels of interpolations i.e., very low, low, medium, strong and very strong are available in the software.
For evaluation we have only considered time
domain features since time domain features are more
sensitive to artifacts as we have seen from the previous
discussion. Table 4 shows the mean absolute differences
when performing interpolation using the proposed
algorithm (considering k=2) and Kubios HRV software
(considering Interpolation level medium). The algorithms
are fist tested with 30 ‘With artifacts’ measurements where
we can see that the proposed algorithm performs better.
However, the error is exceptionally higher for the software
because of there is 3 measurements for which it somehow
fails to manipulate. So, a calculation is also performed
without considering these 3 measurements in which the
software performs better than the proposed algorithm.