Figure 1.

Reported Baseline Characteristics, Internal and External Performance for Included Models

(A) These heatmaps display baseline characteristics reported in derivation (top) and external validation (bottom) studies. Each row represents a study, and each column indicates a specific characteristic (eg, age, sex, comorbidities). Green shows that the characteristic was reported, while red indicates it was not. (B) The difference between internal validation area under the receiver-operating characteristic curve and external validation area under the receiver-operating characteristic curve across models. The purple diamonds represent external validation reported in the original paper. The red diamonds show external validation conducted in separate papers. All values are referenced to internal area under the receiver-operating characteristic curve; therefore, a value higher than zero indicates a higher internal area under the receiver-operating characteristic curve compared to the area under the receiver-operating characteristic curve during external validation. AUROC = area under the receiver-operating curve; BWH = Brigham and Woman's Hospital; CNN = convolutional neural network; ECG = electrocardiogram; MGH = Massachusetts General Hospital; UCSF = University of California San Francisco; VAE = variational auto encoder.