Abstract:Objective To evaluate the effectiveness of artificial intelligence-based ECG algorithm for detection of clinical arrhythmia. MethodsA total of 13 949 clinical data of arrhythmia were collected from 13 145 patients who were treated with artificial intelligence ECG network in our hospital. The data was analyzed using Lepu medical artificial intelligence ECG network system. The diagnosis results of the doctor team were defined as the gold standard, the sensitivity, specificity, positive predictive rate, negative predictive rate and correct rate of the screening test were used to evaluate the effectiveness of the artificial intelligence-based ECG algorithm compared with the gold standard. ResultsCompared with the gold standard, 17 types of arrhythmia were detected by the artificial intelligence-based ECG algorithm including sinus arrhythmia, atrial fibrillation, etc. The overall sensitivity, specificity and accuracy were 98.08%, 99.84% and 99.84%, respectively. The consistency Kappa coefficients of 6 types of arrhythmias(pairing of supraventricular contractions, atrial escape and ventricular escape, etc.) were greater than 0.4 but less than 0.75, which meet the consistency requirements but without strong consistency. Conclusion The test results of artificial intelligencebased ECG algorithm for arrhythmia is highly consistent with the clinical ECG test results. The artificial intelligence-based ECG algorithm has good clinical practice prospects.