Fig. 5: Feature distribution maps (using \(s{h}_{\alpha }^{{\prime\prime} }\) vs. \(s{h}_{{\rm{HR}}}^{{\prime\prime} 2}\)), histograms, and overall information of 4 Databases.
From: mRMEBP: a unified framework for online detection of atrial fibrillation utilizing deep learning
a Feature distribution maps of AF and nAF beats from the LTAFDB, with accompanying histogram sub-figures. b Feature distribution maps of AF and nAF beats from the AFDB, with accompanying histogram sub-figures. c Feature distribution maps of AF and nAF beats from the MITDB, with accompanying histogram sub-figures. d Feature distribution maps of nAF beats from the NSRDB, with accompanying histogram sub-figures. e Overall information on training, validation, and testing for the 4 databases. The distribution maps are created using 100,000 uniformly randomly sampled data points from each database. See text for abbreviations.
