Log in
Pitt Logo

Font Size

Cpanel

2011 MIDAS Pilot Grant Awardees

Seven researchers were awarded MIDAS Pilot Grants in 2011. Five faculty members were awarded "Developmental Awards," which carry a maximum award of $20,000 annually and provide researchers the opportunity to develop techniques or novel approaches to compuational modeling and simluation. Developmental grants may be awarded to investigators from various backgrounds and training who may be new to the field of infectious disease modeling. Two post-doctoral fellows were granted "Seed Awards," which provides them with up to $5,000 annually to conduct research on computational modeling and simulation of infectious diseases. Seed awardees must specify a senior investigator from the University of Pittsburgh's MIDAS National Center of Excellence, who serves as a mentor and co-investigator on the project.

Please read below to learn more about the awardees and their projects. 

Developmental Awards 


Hasan Guclu, PhD guclu

Department of Biostatistics
University of Pittsburgh

Project Title: "A Unified Framework for Modeling Infectious Diseases in Aging Populations with Decaying Immunity"

Abstract: This project aims to provide a convenient and effective modeling framework for risk analysis and policy making for long-term control of infectious diseases such as influenza, measles, and dengue fever. This framework combines the longitudinal demographic data such as the rate of births and deaths and aging in the population, and the social changes such as relocation and migration with the lack or decay of immunity in individuals. This project will fill a gap between these different modeling efforts and serve as a practical tool to test the effectiveness of various countermeasures. The specific tasks of the project will rely on accurate demographic data, theoretical foundations for the immunity modeling, high-performance agent-based simulations and social network analysis.


Louis Luangkesorn, PhD 

louis

Department of Industrial Engineering
University of Pittsburgh

Project Title: "Development of Statistical Methods to More Effectively Inform Policy Analysis Using Agent-Based Simulation Models of Infectious Diseases"

Abstract: Compared to other modeling methods, agent based simulation is able to model a system in greater detail including relationships between individual agents who are able to gather information about their environment and make independent decisions. However, this makes agent based models computationally intensive, making modelers slower to be able to respond to policy maker queries. Other modeling communities have employed methods based on statistical design of experiments to efficiently reach conclusions on the efficacy of interventions into the system. Currently, these questions are addressed in the practice of agent based models through the application of sensitivity analysis techniques. This project proposes to identify and implement design of experiment/optimization via simulation methods to MIDAS infectious disease models for use in policy analysis and evaluation. This identification will be made based on characteristics of the model/system under study as well as the question that the model is intended to address. Specifically, this project will be in conjunction with ongoing work within MIDAS by the PHICOR group on inter-hospital spread of MRSA. Successful completion of the pilot project will improve the support MIDAS provides to public health policy makers improving the ability of MIDAS modelers to respond to policy maker queries and requests.


Eunha Shim, PhD e_shim

Department of Epidemiology
University of Pittsburgh

Project Title: "Influence of Risk Perception on Vaccine Behavior: Rotavirus Vaccination as an Example" 

Abstract: Coming Soon.


 

 


Aarti Singh, PhDAartiSingh

Department of Machine Learning
Carnegie Mellon University 

Project Title: "Using Non-Local Connectivity Information to Identify Nascent Disease Outbreaks"

Abstract: This proposal seeks to develop computationally and statistically efficient procedures for detecting epidemics, such as influenza and MRSA (methicillin-resistant Staphylococcus aureus), in their nascent stages by leveraging non-local connectivity between geographically disparate locations. Recent studies suggest that disease spreading pathways are highly non-local in today's globalized world, e.g. due to ever-increasing volume of air traffic, patient transfers across different hospitals, etc. Most state-of-the-art disease outbreak detection methods use the gravity model to fuse data from local nearby regions. Such methods are plagued with high false alarm rates and poor detection capabilities, particularly when trying to identify an outbreak in its formative stages from crude and unreliable indicators that may not be statistically significant in any spatial region. The proposed research aims to achieve the following two goals: 1) Identify the occurrence of an influenza or MRSA outbreak in its nascent stages by projecting data onto an appropriate basis that is adapted to the structure of the non-local connectivity graph between geographically disparate regions/hospitals. 2) Learn the non-local connectivity graph which influences spread of the influenza virus or MRSA between different locations.

Larry Wasserman, PhD larry

Department of Statistics
Carnegie Mellon University 

Project Title: "Solving the NowCasting Problem" 

Abstract: In many epidemiological settings, knowing the true daily incidence of an infectious disease can be tremendously useful in predicting its future course and in optimizing public health intervention. It is especially helpful to be able to estimate daily incidence in real time. This has been dubbed the "NowCasting" problem. While direct measurement of disease incidence is practically impossible, we do have access to many data streams that are correlated with Influenza incidence. These data streams differ in their time granularity (from daily to weekly), their geographic granularity (from zipcode-level to U.S. Regional) and in their sampling procedure and completeness. However, they all contain some information about daily Influenza incidence. We propose to develop a principled solution to the Influenza NowCasting problem, based on a combination of statistical methodology and machine learning techniques. We seek a solution that can combine available data streams of any time and geographic granularity in a unified modeling framework.


Seed Awards

Supriya Kumar, PhDSupriya

Department of Epidemiology
University of Pittsburgh

Project Title: " Explaining the Mechanism Behind Persistent Racial Disparities in Influenza Vaccine Uptake." 

Abstract: From 2000 to 2010, consistently fewer Blacks over the age of 65 years got the seasonal influenza vaccine compared to Whites aged greater than 65 years in the US. Economic, geographic, and attitudinal barriers to vaccines among minorities have been proposed as reasons for disparities, but the mechanisms by which these factors could lead to the rates of uptake by race seen in the population have not been elucidated. Because surveys are rarely able to simultaneously capture the impacts of individual, geographic, and social network factors, and unable to capture feedbacks from multiple levels on individual vaccination behavior over time, we propose to build an agent-based model of Allegheny County in which individual agents make vaccine uptake decisions based on inputs from multiple sources. Agents with individual-level attributes, including demographic factors, employment status, and health insurance will update their attitude toward the vaccine daily as a function of feedbacks from pharmaceutical advertisements of vaccine and the attitudes of others in their social network. The resulting rates of vaccine uptake by race will be compared to rates seen in the Allegheny County Behavioral Risk Factor Surveillance System; such a model will allow us to not only elucidate a plausible mechanism that could cause the observed disparities, but also lead to future studies on the structure of influential social networks in such decision-making and reciprocal determinism between population-level vaccine uptake and the location of pharmacies in neighborhoods.


Olabisi Ojo, PhDbisi

Department of Microbiology and Molecular Genetics
University of Pittsburgh

Project Title: Agent Based Modeling of Tuberculosis Transmission in Allegheny County: a Pilot Evaluation" 

Abstract: Tuberculosis (TB) is an infectious disease found in humans and animals. It is mainly aerogenically transmitted between individuals as they intermingle in different communities. It has more recently become a worldwide research priority due to re-emergence in the developed world and co-infection with Human Immunodeficiency Virus-Acquired Immune Deficiency Syndrome especially in TB endemic countries. Modeling of infectious agents incorporates computational simulation of infectious diseases events to decipher different aspects of disease dynamics. The aims of this proposal are to carry out primary exploration of literature, data and tools learning for modeling TB transmission via the agency of the aerosol droplet by constructing an Agent Based Model (ABM), in the Allegheny County, Pennsylvania. Current data will be collected from health units from Allegheny County, to drive the proposed model. Each computer agent will represent a virtual person, discrete entities, with a potential for TB infection status, capable of moving among the general community for each of the simulated features. The ABM will be parameterized with probabilities of features of an artificial population. The ABM will be implemented in NetLogo® and the modeling proprietary software developed by the University of Pittsburgh Supercomputing core. It is expected that the model will be calibrated and the impact of this could therefore be evaluated by the introduction of preventive strategies, results of which could be potentially utilized in public health decision making. It is expected that the primary data generated from this project would be developed into a future grant application.