|
Activity Number:
|
505
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309756 |
|
Title:
|
Parameter Calibration in an Agent-Based Model of Leishmania Major Infection
|
|
Author(s):
|
Garrett Dancik*+ and Karin Dorman
|
|
Companies:
|
Iowa State University and Iowa State University
|
|
Address:
|
819 24th St, Ames, IA, 50010,
|
|
Keywords:
|
computer experiments ; Gaussian process ; parameter calibration
|
|
Abstract:
|
Computer models of disease provide in silico environments for understanding complex interactions between pathogen and host. Often, modelers take ad-hoc approaches to parameter estimation when models are high dimensional and resource intensive, so in silico results have limited real-world applicability. However, statistical methods for calibration of complex, resource-hungry models to data are available. We use a Gaussian process approximation of the computer code and estimate five immunological and pathogen-related parameters in an agent-based model of Leishmania major infection. We verify the method using simulated field data and then calibrate using published biological observations. Our results suggest that L. major has a slow growth rate and replicates for an extended time before damaging the host cell. We also discuss sensitivity analysis as a strategy for future data collection.
|
