Computer Modeling To Help Cities Prepare for Worst-Case Scenarios

A new and novel computer modeling platform can help hospitals and cities to be more prepared for catastrophic public health scenarios,

by / June 11, 2009
Previous Next

Screen shot: A snapshot of the Manhattan model during the simulation of a Sarin gas attack.

A new and novel computer modeling platform developed through intensive, multidisciplinary collaboration at New York University (NYU) can help hospitals and cities to be more prepared for catastrophic public health scenarios, according to an article published in the American Medical Association's Disaster Medicine and Public Health Preparedness journal.

The article, "A Novel Approach to Multihazard Modeling and Simulation," is based on the authors' test of the NYU computerized disaster simulation framework known as "Plan C" with a hypothetical malicious sarin release in several Manhattan locations. With the input of city demographic information, hospital resource and public transit data, the results showed that under certain circumstances, up to 22,000 individuals might become exposed, leading to 178 intensive care unit admissions.

According to the NYU web site, PLAN C (which is an acronym Planning with Large Agent-Networks against Catastrophes), has been able to simulate the complex dynamics of emergency responses in different urban catastrophic scenarios such as a chemical agent, a bomb explosion, food poisoning and small pox. It uses a powerful, large-scale computational, multi-agent based disaster simulation framework involving as many as thousands of variables or agents - from existing hospital beds and emergency department services to hospital surge capacity and behavioral and psychosocial characteristics to anticipate public response to an attack.

It can also devise plans that optimize multiple objective functions such as number of casualties, economic impact, time to recovery, etc. It is designed to be easy to use by relatively unsophisticated users. Additionally, the technology can be easily transferred to any urban setting, to multiple computer platforms, and to different modes (offline or online) of planning.

Screen shot: The main PLAN C interface.

In the article, the authors note that implementing disaster plans within 30 minutes compared to two hours of an incident diminished mortality and waiting times and reduced the number of patients who were severely affected. And GIS portability to other urban locations was demonstrated.

"An agent-based modeling approach," the authors write, "provides a mechanism to assess complex individual and system wide effects in rare events."

To accomplish this, the system offers the following features:

* A large number of computational actors/agents, comprising five different classes: Person, Hospital, On-Site Responder, Ambulance and Catastrophe;
* A flexible number of parameters for describing the computational agents' behavior and interaction, the time course of the disease, environmental factors, etc., which enable the user, e.g., an emergency manager, to modify adaptively to diverse urban scenarios;
* Several communication channels for information (health / resource levels, hospital operation mode, etc.) exchange among similar and differing computational agents;
* Modeling people as selfish and bounded rational beings, with stochastic personality traits emulating panic or contrarian behavior;
* Realistic models of medical / responder units and catastrophe chemical agent effects (disease prognosis and dosage response), validated by medical, sociological and legal experts from the NYU CCPR;
* Integration of real urban infrastructure constraints (streets, subways, hospitals, etc), via publicly available Geographic Information System data of a city such as New York City;
* Computer software for parallel and distributed concurrent computing on large-scale clusters of workstations, using the integration between ProActive and RePast.
* Integration of medical, legal, public safety, and social professionals' perspectives in the model design and development.

Currently, the model is being adapted to model pandemic influenza and the authors aim to continue expanding the model in their efforts to further preparedness efforts.

More information is available at


Blake Harris Editor