Closed LegrandNico closed 12 months ago
Thanks for the comment. I have added more information to the figure. I have also added additional figures to the repository. Please check the new figure.
The figures show the true peaks according to the manual annotations of the MIT-BIH dataset.
A false positive means that a sub-segment is predicted to contain an R-peak, but in reality it does not. The figures show that this is usually the case for nearby points close to the true R-peak. False positives are shown in orange color. When false positives are close to the true R-peaks, we can say that the prediction is right. I have calculated the results considering this fact. Results It is possible to improve detection with a post-processing algorithm in these situations. Sometimes, these false positives are caused by a small mistake in the manual annotations.
False negatives are when the model misses to detect a true R-peak.
Although this is just an example, the model predicts with high accuracy. It can be improved to achieve more accurate predictions and detect almost all of the R-peaks.
It is unclear to me what the label represents in the paper's main figure see here. I think this one would benefit from some context about the processing pipeline and analysis goal.
For example, why are some of the R waves not labelled as True peaks when they seem normal?
What do false positive and negative mean exactly?
The prediction appears to be missing some of the R wave, is it expected or a problem?
This issue is related to the review of the package for JOSS (see here).