Department of Statistics, The Ohio State University
Statistics and Biostatistics Colloquium Series

We consider different aspects of the study of heart period data, namely, data acquisition (artifact identification), data processing (non-stationarity and stability of spectral measures), and data analysis (modeling and quantification of HRV measures). We examine the distribution of successive heart period differences using five different data sets (three from humans and two from animals) and conclude that in the absence of artifacts in the study interval, the successive differences exhibit symmetry about zero and the absolute differences follow a Weibull distribution. Using this model, we explore the statistical properties of some time domain measures of HRV/HPV, and propose a semi-parametric artifact detection criterion. We establish an association between the Weibull scale parameter estimate and respiratory sinus arrhythmia (RSA), which is a measure of change in heart period associated with respiration. In addition, we also model the heart period data as a mixture distribution. Finally, we provide an assessment of the influence of the duration of heart period data recording on the spectral measures of HRV. This is done for a group of heart-failure patients at baseline and post-treatment and for dogs at baseline and peak-failure.

This is a joint work with my adviser, Prof. H. N. Nagaraja and Professors Gary Berntson (Psychology) and Philip Binkley (Internal Medicine).