|Title||:||Modeling in the Time of Ebola - Using HPC Simulations to Understand Infectious Disease Dynamics|
|Speaker||:||Srini Venkat (Virginia Tech, USA)|
|Details||:||Thu, 11 Feb, 2016 12:00 PM @ BSB 361|
|Abstract:||:||Advances in disease surveillance, access to social datasets and availability of computing resources have heralded the transformation of epidemiology from a specialized domain, to a highly multi-disciplinary area. Further, as recent epidemic outbreaks (Ebola, Swine Flu) have shown, a highly interconnected world is primed for rapid disease spread. While traditional ODE models help provide initial estimates to the extent of epidemic spread, they are incapable of addressing more pertinent questions: Which regions or communities are at immediate risk? Which public health interventions should be implemented?
In this talk, we provide a brief overview of NDSSL’s simulation-based approach built atop synthetic populations, to address this issue. This ability to quickly study virtual outbreaks on realistic populations help public health experts to conduct various what-if scenarios, and produce more plausible forecasts. Using the West African Ebola outbreak of 2014 as an example, we show how our pipeline works, starting from models and data, to producing forecasts and recommendations. Specifically, we will focus on how we use a combination of optimization and Bayesian approaches for model calibration. Finally, we will describe our efforts in the recently concluded RAPIDD Ebola challenge, organized by National Institute of Health (NIH) to compare and contrast existing approaches, and train the modeling community in ‘peace time’ to improve predictive capabilities.
Speaker Bio: Dr. Srini Venkat is currently a postdoctoral researcher at the Network Dynamics and Simulation Science Laboratory (NDSSL), which is part of the Biocomplexity Institute of Virginia Tech. He obtained his PhD from the Dept. of ECE, Indian Institute of Science, for his study on influence dynamics on social networks. At NDSSL, he is responsible for designing, analyzing and implementing provable algorithms for analyzing complex networks arising in the study of large socio-technical, biological and information systems. His areas of interest include computational epidemiology, academic social networks and mobile opportunistic systems.