249 Linear and Nonlinear Properties of Heart Rate Variability: Influence of Obesity A. GASTALDELLI, R. MAMMOLITI, E. MUSCELLI, S. CAMASTRA, L. LANDINI, E. FERRANNINI, AND M. EMDIN C.N.R. Institute of Clinical Physiology, Department of Internal Medicine and Department of Informatic Engineering, University of Pisa, 56126 Pisa, Italy INTRODUCTION Physiological systems are best characterized as time-varying processes exhibit- ing rhythmic and complex behavior. The interaction among system variables, exter- nal noise, and state changes modulates the overall variability of physiological signals such as heart rate, arterial pressure, and respiration, which may therefore present both linear and nonlinear patterns. To describe the complex and periodic dy- namics of living systems, various analytical tools have been employed, especially in the cardiovascular field. 1 Among them, power spectral analysis (PSA) 2 and recur- rence quantification analysis (RQA) 3,4 have been used to describe, respectively, lin- ear and nonlinear dynamics of heart rate variability (HRV). PSA is a validated method that quantifies autonomic nervous modulation of cardiac activity by describ- ing the fluctuations of HR linked to vasomotion and respiration. RQA evaluates complexity and determinism in time series by detecting state changes in drifting or exciting dynamical systems. RQA can be easily applied to cardiovascular signals be- cause it does not require any a priori mathematical assumption, such as stationarity or linearity; parameters introduced by RQA, based on distance, recurrence, and en- tropy of recurrence plots (RP), 5 may be related to different physiological states. Nev- ertheless, no correlation has been shown between RQA parameters and autonomic nervous activity. It has recently been shown that obesity is a state of reduced sensitivity of the si- noatrial node to both sympathetic and vagal influences. 6 Data from obese and lean subjects were therefore analyzed by PSA and RQA, and parameters derived by the two methods were compared for the two groups of subjects. METHODS PSA and RQA were applied to the R-wave peak interval (RR interval) time series as derived by continuous electrocardiographic (ECG) monitoring (250-Hz frequency sampling). We analyzed 21 ECG tracings recorded during 60 min of quiet, supine rest. Subjects were divided into two groups, 13 obese and 8 lean, on the basis of their body mass index (BMI > 28 kg·m -2 ). The characteristics of the subjects are shown in TABLE 1.