Estimation of local scale exponents for heartbeat time series based on DFA

• Published in: Nonlinear dynamics Vol. 74; no. 4; pp. 1183 - 1190
• Main Authors: Xia, Jianan, Shang, Pengjian, Wang, Jing
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.12.2013; Springer Nature B.V
• Subjects: Automotive Engineering; Classical Mechanics; Control; Data analysis; Dynamical Systems; Engineering; Exponents; Fibrillation; Heart; Mechanical Engineering; Original Paper; Outliers (statistics); Time series; Variations; Vibration; Volatility; Windows (intervals); Local scale exponent; DFA; Heartbeat time series
• ISSN: 0924-090X; 1573-269X
• DOI: 10.1007/s11071-013-1033-2

The paper mainly introduces a method combining detrended fluctuation analysis (DFA) with moving fitting window to analyze the heartbeat time series in different pathologic states. Compared to traditional approaches, the estimation of the local scale exponent method shows more details of scale properties and provides a reliable analysis. We also quantify the effects of outliers, subseries length and gender on scale spectrum α ( s ). Comparing the results of the healthy subjects, patients with congestive heart failure (CHF), and patients with atrial fibrillation (AF) indicate clearly that at small scales, the exponents show great volatility and they all have their own scale pattern. In addition, the length of the series should be relatively longer to avoid properties losing. For the effect of gender, contrary to healthy results, the scale exponents of CHF male subjects are continuously greater than female. The spectrum of DFA scale exponents provides a new way to measure the heartbeat series and distinguishes healthy and pathologic groups.