Abstract:
|
Measures on heart rate variability, such as the RR-interval, have been used extensively to indirectly assess the autonomic control of the heart. The distributions of these measures are not necessarily normally distributed, and current methodology for modeling these data does not typically incorporate this characteristic. Furthermore, the series of RR-intervals from a single subject is correlated and must be accounted for when modeling these data. In this presentation, an adaptation of the methodologies for fitting random-effects models is proposed with data from a mixture-normal distribution. We assume the random-effects are normally distributed, while the residuals follow a normal-mixture distribution. In our approach, with the use of an EM algorithm in SAS, the estimation can be achieved through existing software for random-effects models. The results from a simulation study are discussed along with examples of applications in different research settings.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.