A development and clinical evaluation of automated diagnostic algorithm for atrial fibrillation using 12-lead EKG

Abstract

The atrial fibrillation (AF or A Fib) has been known as an important risk factor for stroke. The mortality of patients with AF is known to increase two times compared with the patients with no AF. Moreover, the association of heart failure with AF increases the rates of mortality and hospitalization very high. While the accurate detection of atrial fibrillation waves (fwave) is very important in the automated diagnosis of AF, it is difficult to detect and distinguish from noise signal, because they are often too small or very similar in the amplitude and morphology. To solve this problem, we developed the automated AF diagnostic algorithm using f-wave amplitude and RR interval behavior. To improve the performance of the proposed algorithm, we determine the optimal threshold level to detect f-wave and the characteristics of RR interval irregularity using the large size of 12-lead EKG data base. We collected 1,270 consecutive cases including 149 cases (11.7%) with AF from the Chonnam University Hospital using Bionet EKG3000. Diagnosis of the EKGs were confirmed by a EKG specialist. Our diagnostic algorithm showed the sensitivity of 91.9% and specificity of 97.7% in the diagnosis of AF. We also compared the diagnostic values between Bionet EKG3000 and GE Marquette system, MAG5000 in the selected 157 cases comprising of 32 cases with AF. There was no significant difference between the two systems. The diagnostic sensitivity and specificity were 93.7% and 98.4%, respectively in the GE MAC5000, 100% and 98.4%, respectively in the Bionet EKG3000. Our proposed AF diagnostic algorithm showed robustness and high performance in the automated AF detection.