Fig. 1: Flowchart describing the creation of the main study sample.
From: A deep learning ECG model for identification and localization of occlusion myocardial infarction
Panel A shows inclusion and exclusion criteria for the main study sample (the Swedish Emergency Department database [SwED]). There are three main quantities where N ECGs > N patient visits > N patients. Panel B shows data splits of the derived main study sample. The sets (training, validation, test random, test temporal) are created so that a patient can only belong to one set (to prevent any information leakage between the sets). The validation and test sets are further restricted so they use at maximum one ECG per patient visit. The training, validation and random test set includes patients drawn at random whereas the temporal test set includes patients with their first available ECG recording during the last year (2016).
