Two-dimensional Poincare plots wereA�also generated by plotting each R-R interval as a function of its.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians.
Correspondence to: Aurelien Pichon, PhD, Laboratoire a�?RSponses cellulaires et fonctionnelles a la��hypoxie, UFR Sante MedecineA�Biologie Humaine, 74 rue Marcel Cachin, 93017 Bobigny,A�France; previous R-R interval obtained at baseline and after MBC.
Table 1a��Anthropometric and Spirometric Characteristics of the Subjects
|Age, yr||26 A� 10||31 A� 12|
|Height, cm||173 A� 10||173 A� 10|
|Weight, kg||69 A� 11||73 A� 12|
|PD20 hg||467 A� 351|
|FVC, L||4.9 A� 1.0||4.9 A� 0.9|
|FVC, %||108 A� 11||108 A� 13|
|FEVb L||3.8 A� 0.7||4.1 A� 0.9|
|FEVj % predicted||101 A� 12||108 A� 15t|
|Physical activity, h/wk||8 A� 7||4 A� 4t|
Data are presented as mean A� SD. PD20 = provocative dose of methacholine causing a 20% fall in FEVPA�tSignificantly different from responder subjects.
A two-dimensional vector analysis was then used to quantify theA�shape of the plots: short-term R-R interval variability (SD1) andA�long-term RR interval variability (SD2) of the plot were separately quantified.
Autoregessive Analysis: Harmonic components of the R-R interval were analyzed by the autoregressive method (HRVA�Analysis Software 1.1 for Windows; Biomedical Signal AnalysisA�Group, Department of Applied Physics, University of Kuopio;A�Kuopio, Finland). Autoregressive coefficients were estimatedA�using the forward-backward linear least-squares algorithm with aA�16th-order autoregressive model. The R-R interval time seriesA�were interpolated at a rate of 2 Hz and detrend prior to theA�analysis. The power density of LF and HF components wasA�calculated and expressed in absolute units (ms) and normalizedA�units (n.u.), which were obtained as follows: HF n.u. = (HFA�ms)/(LF ms + HF ms) X 100). The LF/HF ratio was alsoA�calculated to assess sympathetic/parasympathetic modulation.
Short-Time Fourier Transform: The short-time Fourier transform (STFT) of R-R intervals corresponds to a sliding fast Fourier transform analysis. The STFT processing yields anA�analysis in time-frequency domain that can be exemplified with aA�three-dimensional figure to exhibit the evolution of HRVA�throughout the observed bouts of exercise. The signal is convolved with some constant-duration time window, and the spectral components are calculated for each windowed segment.
The STFT analyses were performed using specific software after Hamming windowing (MATLAB 5.3; The MathWorks; Natick,A�MA). After loading the American Standard Code for InformationA�Interchange file, an R-R periodogram was performed and displayed in order to pick out the more relevant stretch for STFTA�analysis. This stretch needs to be > 320 values to perform aA�STFT on a block of 256 values.