Evaluating the Harmonization of Native Digital and Digitized ECGs for ECG-AI Research

Large epidemiological studies often have only paper or scanned ECG records instead of native digital files, which limits artificial intelligence applications in ECG analysis. Researchers compared 15 ECGs available as native digital XML files and as PDF tracings digitized using ECGScan software. Both versions were processed using a set of five convolutional neural networks that predicted 10-year heart failure risk. The predictions were strongly correlated with a Pearson correlation coefficient of 0.804 and Spearman coefficient of 0.893. The mean predicted probability was 0.195 (SD 0.048) for native digital ECGs and 0.171 (SD 0.042) for digitized ECGs. These preliminary findings support further evaluation of digitized ECGs for ECG-AI applications.