• Subscribe

Vital software provides critical data to public health policy makers

Influenza simulation in Allegheny County, PA, US. Image courtesy University of Pittsburgh Public Health Dynamics Laboratory.

Researchers at the University of Pittsburgh, together with the Pittsburgh Supercomputing Center (PSC), and the School of Computer Science at Carnegie Mellon University, Pennsylvania, US, have developed a new version of their agent-based modeling system for epidemic dynamics. The large-scale modeling program, known as FRED (Framework for Reconstructing Epidemiological Dynamics), enables researchers to investigate and understand the possible courses of future epidemics - and how they might control and alleviate them.

FRED is able to simulate the fine-grained actions of individuals in a given region, as they go about their daily lives and interact with other individuals in their households, neighborhoods, schools, and workplaces. These simulations combine US census data with other school and workplace databases to ensure a realistic representation of the US population. The numbers of households, people, children in schools, adults at work, and commuting distances are accurate to within a few percent over most US communities

Researchers can introduce a virtual disease into societal models, and see how it transmits from person to person as they evaluate various interventions."What are the possible effective vaccination programs, school closure policies, sick leave policies, or social distancing norms? These are the policy questions we are interested in," says John Grefenstette, director of the Public Health Dynamics Laboratory at the School of Public Health at Pittsburgh.

"FRED currently supports a variety of epidemic research areas from how viruses evolve resistance, to human behaviors like which sub-populations decide to accept a vaccine if it is available or decide to stay home when sick, and how this behavior affects a population level epidemic. The models in FRED account for these variations as well as combinations of variations to predict a more accurately how an epidemic spreads," explains Grefenstette.

Most recently, PSC's Shawn Brown, director of Public Health Applications, and Jay DePasse, senior computational public health specialist, have spearheaded efforts to create a multi-threaded version of FRED. "Since we're running a stochastic model, we have to run several instances of the same model with the same initial conditions to get statistically meaningful results," explains Brown. "Blacklight is a great architecture for us because we can run several instances of a really large shared memory application. The performance is great."

A simulation for a typical US influenza epidemic takes about 3.5 hours running 16 threads on Blacklight. Smaller runs on counties with populations of 1 million, for example, can run on a laptop in 30 seconds - compared to a single threaded run, which takes about two minutes on the same laptop. All told, FRED requires around 750 megabytes of memory per million persons, so not all states in the US could conveniently run on common hardware.

"Simulating the entire state of Pennsylvania would require about 6.5 gigabytes, which could run on a laptop in minutes depending on how powerful it is," adds DePasse. "California or New York would require special computing resources, and if we wanted to run the entire US population we would need roughly 200 gigabytes of memory. And that's just a baseline before layering on additional parameters in the application, such as a more or less sophisticated agent, a more or less severe epidemic, as well as others."

The Public Health Dynamics Laboratory at Pittsburgh is funded by the MIDAS (Models of Infectious Disease Agent Study) National Center of Excellence, which provides policy makers and public health organizations with critical information. Organizations include the US Department of Health and Human Services, Department of Homeland Security, Biodefense Advanced Research Development Authority (BARDA), and the Centers for Disease Control and Prevention.

Last week, the World Health Organization advised healthcare workers around the world of 55 confirmed cases of Middle Eastern Respiratory Syndrome (MERS), mainly in four Middle Eastern countries. They also confirmed that travellers have started to spread the disease. However, MERS is just the latest virus with the potential to morph into an epidemic. HIV, Malaria, SARS, and H1N1 make headlines while scientists struggle to create vaccines for these highly variable and evasive viruses.

Ultimately, systems like FRED are critical to scientists' ability to stay one step ahead of future epidemics, and to intervene and mitigate when outbreaks inevitably happen. In the future, researchers and developers will add more populations and disease models to FRED. Current international populations include Taiwan and Thailand, and work continues on India, China, and Brazil. "The ultimate goal is to provide a multi-disease, multi-population platform for universities and labs studying various models of infectious disease - models that accurately inform policy makers on public health issues," says Grefenstette.

To learn more about the open source modeling system, download the application, or run the simulator, visit the FREDwebsite.

Join the conversation

Do you have story ideas or something to contribute? Let us know!

Copyright © 2019 Science Node ™  |  Privacy Notice  |  Sitemap

Disclaimer: While Science Node ™ does its best to provide complete and up-to-date information, it does not warrant that the information is error-free and disclaims all liability with respect to results from the use of the information.


We encourage you to republish this article online and in print, it’s free under our creative commons attribution license, but please follow some simple guidelines:
  1. You have to credit our authors.
  2. You have to credit ScienceNode.org — where possible include our logo with a link back to the original article.
  3. You can simply run the first few lines of the article and then add: “Read the full article on ScienceNode.org” containing a link back to the original article.
  4. The easiest way to get the article on your site is to embed the code below.