|
Activity Number:
|
440
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305423 |
|
Title:
|
Application of an Extended Poisson Process to Model Cardiovascular Events
|
|
Author(s):
|
Paul Kolm*+ and Zugui Zhang and Edward Ewen and James Bowen
|
|
Companies:
|
Christiana Care Health System and Christiana Care Health System and Christiana Care Health System and Christiana Care Health System
|
|
Address:
|
131 Continental Drive, Suite 202, Newark, DE, 19713,
|
|
Keywords:
|
Poisson process ; overdispersion ; underdispersion
|
|
Abstract:
|
Count data modeled as a Poisson process are often overdispersed. In this case, the negative binomial distribution is typically used to model the data. In a recent article, Faddy and Smith (Appl. Statist., 2008), developed an extended Poisson process model that includes Poisson and negative binomial distributions as well as distributions less dispersed than the negative binomial, but overdispersed relative to the Poisson. In addition, the extended model can account for underdispersion. The extended process is applied to counts of cardiovascular events in coronary artery diseased patients to assess the appropriate distribution to model the data.
|