Figure 1.

Sample Visualization of the input and output of the ML tools. The ML network predicts the ischemic tissue volumes (top) from the torso surface ECG recordings (bottom, ST40% values interpolated). The intracardiac needle electrodes recording the electrograms were visualized within the heart geometry (gray spheres). Based on the error of the specific prediction, the minimum (left) and maximum (right) error cases were found and the input-output pairs were visualized, respectively. The ST40% values of the best and worst performing heartbeat were used to visualize across the torso geometry. For demonstration purposes, the best-performing network #3 was used for the error evaluation here.