Abstract

Info

Title:: Evaluation of the precision of probability density function method using R-R intervals for detecting atrial fibrillation from the whole MIT-BIH arrhythmia database

Author(s):: LU Hongwei; ZHONG Gaoyan; LONG Yufan; LIU Zhangcai; School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Keywords:: ; atrial fibrillation; R-R intervals; probability density function method; premature heart beats; arrhythmia

Abstract:: The precision of probability density function method for detecting atrial fibrillation (AF) by R-R intervals is evaluated with the use of the whole MIT-BIH arrhythmia database. With the normal sinus rhythm with frequently occurring premature heart beats (PHB), the number of PHB/the total number of heart beats larger than 9.3%, AF detection precision is reduced to about 70%; with the left bundle branch block with frequently occurring PHB, the precision is decreased to 80%; while AF with frequently occurring PHB has little effects on the AF detection precision, and the detection precision remains 91%. Therefore, the proposed algorithm is suitable to differentiate sequences with high correlation from sequences with weak correlation. The neighboring R-R intervals of normal sinus rhythm and left bundle branch block are highly correlated. Frequently occurring PHB weakens the correlation among neighboring R-R intervals, which degrades the AF detection precision. However, there are no correlations among neighboring R-R intervals of AF, which indicated that frequently occurring PHB has no effects on the AF detection precision. The algorithm is evaluated with the whole MIT-BIH arrhythmia database which contains about 110,000 heart beats. Though many kinds of arrhythmia accompanied with frequently occurring PHB are included in the database, the AF detection precision of the whole database remains 82%-86%.