A pediatric ECG database with disease diagnosis covering 11643 children.

Electrocardiogram (ECG) is a common non-invasive diagnostic tool for cardiovascular diseases. Adequate data is crucial in utilizing deep learning to achieve intelligent diagnosis of ECG. The existing ECG datasets almost only focus on adults and most of them do not provide cardiovascular disease diagnosis. In this study, we propose an ECG database with cardiovascular disease diagnosis for children aged 0-14 years old. This dataset is acquired from 11643 hospitalized children at the First Affiliated Hospital of Zhengzhou University from 2018 to 2024, including 14190 pediatric ECG records, of which 12334 were 12 lead and 1856 were 9 lead. The sampling rate is 500 Hz and the record length is 5-120 seconds. We followed the recommendations of AHA/ACC/HRS and the diagnostic statements in the consensus of Chinese ECG experts to encode and convert all ECG records. In this dataset, 3516 ECG records were diagnosed with cardiovascular diseases, and these labels were derived from 19 common diseases in the pediatric cardiovascular field, including myocarditis, cardiomyopathy, congenital heart disease, and Kawasaki disease.