Multiscale entropy-based methods for heart rate variability complexity analysis

作者：Virgilio Silva Luiz Eduardo; Troca Cabella Brenno Caetano; da Costa Neves Ubiraci Pereira; Murta Junior Luiz Otavio

来源：Physica A: Statistical Mechanics and Its Applications , 2015, 422: 143-152.

DOI：10.1016/j.physa.2014.12.011

• Nonadditive statistics
• Tsallis entropy
• Multiscale entropy
• Complexity
• Heart rate variability

Physiologic complexity is an important concept to characterize time series from biological systems, which associated to multiscale analysis can contribute to comprehension of many complex phenomena. Although multiscale entropy has been applied to physiological time series, it measures irregularity as function of scale. In this study we purpose and evaluate a set of three complexity metrics as function of time scales. Complexity metrics are derived from nonadditive entropy supported by generation of surrogate data, i.e. SDiff(qmax,) q(max) and q(zero). In order to access accuracy of proposed complexity metrics, receiver operating characteristic (ROC) curves were built and area under the curves was computed for three physiological situations. Heart rate variability (HRV) time series in normal sinus rhythm, atrial fibrillation, and congestive heart failure data set were analyzed. Results show that proposed metric for complexity is accurate and robust when compared to classic entropic irregularity metrics. Furthermore, SDiff(qmax) is the most accurate for lower scales, whereas q(max) and q(zero) are the most accurate when higher time scales are considered. Multiscale complexity analysis described here showed potential to assess complex physiological time series and deserves further investigation in wide context.

• 出版日期2015-3-15