Agent-based Model of Patient Flow within a Hospital Network: Understanding How Changes in One Hospital Can Affect Other Hospitals in a Region
SESSION: Research Poster Reception
EVENT TYPE: Poster
TIME: 5:15PM - 7:00PM
AUTHOR(S):Kim F. Wong, S. Levent Yilmaz, Richard A. Christie, Bruce Y. Lee, Susan S. Huang
ABSTRACT: Since different hospitals share patients, changes in patient demographics/characteristics or epidemiology in one hospital may affect those of other hospitals within a region. We developed a computer model of patient flow among the adult acute care hospitals and communities in Orange County, CA and demonstrated how increases (e.g., outbreaks or changes in populations and risk factors) and decreases (e.g., improvements in disease prevention or control measures) in disease prevalence at different hospitals disseminated throughout the network. This agent-based model represents the hospital network at the individual patient level and describes patient flow based on a Monte Carlo algorithm. Time propagation of this synthetic population, thus, provides a stochastic realization of the hospital transfer dynamics. This very stochasticity, however, imposes the requirement for averaging numerous realizations in order to obtain statistical convergence of emergent phenomena. We utilize MPI to parallelize the ensemble of realizations across processors to achieve linear scaling.