Improving explainability of deep neural network-based electrocardiogram interpretation using variational auto-encoders

Rutger R van de Leur; Max N Bos; Karim Taha; Arjan Sammani; Ming Wai Yeung; Stefan van Duijvenboden; Pier D Lambiase; Rutger J Hassink; Pim van der Harst; Pieter A Doevendans; Deepak K Gupta; René van Es

doi:10.1093/ehjdh/ztac038

Figure 3

An external file that holds a picture, illustration, etc.
Object name is ztac038f3.jpg

Relationship of the electrocardiogram factors with conventional electrocardiogram measurements. (A) Hexagon plots where datapoints of electrocardiogram factor–electrocardiogram measurements pairs over all samples in the variational auto-encoder variational auto-encoder dataset are binned into hexagon grids to relate values of Factors 8, 25, 30, and 10 to the PR interval, QRS duration, QT interval, and ventricular rate, respectively. (B) Pearson correlation coefficients between electrocardiogram measures of ventricular rate, PR interval, QRS duration, QT interval, Bazett corrected QT interval, R-axis, and T-axis, and electrocardiogram factor values over all samples in the variational auto-encoder dataset.