CERL at UNT

[+] | [-]
Modeling the Demographic Disparity in the Prevalence of HIV
DynSNIC : Dynamic, Evolving Intimate Contact Social Networks
Modeling an Influenza Outbreak
Modeling of the Human Papilloma Virus
Hidden Markov Models and Bayesian Disease Modeling
Mobile Vaccination Clinics


Modeling the Demographic Disparity in the Prevalence of HIV

Our research addresses health disparities in the domain of infectious diseases in general and HIV/AIDS in particular, through the development of computational models and tools that facilitate the predictive analysis of disease manifestation in a given population. While health disparities are presumed to be the result of differentiated access to health care, there are other factors that are causing differences in disease manifestation across a given population. We conjecture that specific behavioral patterns play a significant role in the spatio-temporal distribution of infectious diseases and drastically impact their spread dynamics.

A non-temporal demographic, discrete time mathematical model has been developed. Our model facilitates quantification of HIV prevalence disparity among demographic groups. Specifically, the model is able to represent demographic stratifications that determine classes of risk behavior. As part of the HIV model, an online tool was created that interfaces with the non-temporal discrete time model; the web-based user interface overlays the underlying mathematical model. This overlay simplifies input of the different parameters controlling the model and returns experimental results in a clear, concise, graphical format. This tool, available at HIV Model Interface, accepts the necessary disease and demographic parameters of HIV to predict the endemic prevalence in a demographically stratified population. Experimental analysis of the disparity model and online interface has commenced. Population risk behaviors are being compiled from national and state-wide behavioral risk surveys. These surveys include the Youth Risk Behavior Surveillance Survey (YRBSS), Behavioral Risk Factor Surveillance Survey (BRFSS), and the National Health and Social Life Survey (NHSLS). The garnered properties will facilitate experiment design and further model development.



DynSNIC : Dynamic, Evolving Intimate Contact Social Netowrks

Recent growth in the prevalence of sexually transmitted diseases and infections in developing and developed countries general population has prompted a great deal of inter-disciplinary research to curb the population wide effect of these diseases. Public health professionals often have limited budgets and resources must be specifically tailored to achieve maximum results. The utilization of computational social networking tools would allow for those within the public health industry to anticipate the impact of demographic specific predictions, and tailor awareness, educational, vaccination, and prophylactic programs for the greatest impact within their population. With limited funding and resources available to help prevent infectious disease, public health professionals need tools to help them to make decisions regarding where the most effective measures would be taken.

Sexually transmitted diseases and infections are, by definition, transferred among intimate social settings. Although the circumstances under which these social settings are established and maintained may vary, the common prerequisite remains an intimate level of social atmosphere. For this reason, the development of sexually transmitted disease mathematical and computational models must utilize a precise and efficient social networking tool. CERL presents DynSNIC (Dynamic Social Network of Intimate Contacts), a computational simulator created to embody the intimate dynamic and evolving social networks related to the transmission of sexually transmitted diseases and infections. DynSNIC's utilization by health professionals will facilitate evaluation of targeted intervention strategies and public health policies.



Modeling an Influenza Outbreak

In an effort to prevent an influenza pandemic as the one witnessed in 1918, which killed as many as 100 million people world wide, disease monitoring and syndromic surveillance methods have been deployed. The methods are designed to identify early cases of influenza and guide the allocation of public health resources to control and contain an outbreak. Nevertheless, the dynamics and progression of influenza in a given population remains elusive and cannot be easily derived. At CERL, faculty and students are currently developing computational models that attempt to reverse engineer influenza outbreaks, thereby extracting the geographic and demographic characteristics that might affect influenza outbreak patterns. The development of a framework for the simulation of influenza outbreaks in multiple regions with different geography, infrastructure, and populations with diverse demographics forms the basis for this effort. Local health officials can generally only observe the combined demand for treatment in the event of an infectious disease outbreak. The tools developed at CERL will facilitate the analysis of an outbreak as the superposition of multiple smaller outbreaks in distinct regions or demographic subgroups. With these tools, epidemiologists and public health officials can engage in a detailed what-if-analysis, thereby experimenting with different vaccination or prevention strategies and optimizing the allocation of public health resources across the region.


Modeling of the Human Papilloma Virus

In the past several years there have been significant improvements in our understanding of cervical cancer.  In 2001, the United States health care system spent over $1.5 billion on treatment for cervical dysplasia and an additional $2 billion on screening tests such as pap smears.  Human Papilloma Virus (HPV) DNA is found in 99.7% of all cervical cancers.  An effective HPV vaccine would have significant impact on HPV infections and cervical disease.  Candidate vaccines finished phase 2 testing in the United States and phase 3 trials have begun.  Because of the health care costs associated with this virus, it is important to have an effective vaccination strategy in place when this vaccine becomes available in the near future.

Computational models are important tools in determining the transmission dynamics of disease and an efficient and effective vaccination strategy.  Our effort to create these models at the CERL is to aid in our understanding of disease patterns and the probable impact of an intervention or vaccine.  Our HPV model stratifies a population into different subgroups based on sexual mixing patterns.  We analyze population demographics and census data to extract demographic parameters for our model and mine risk behavior studies in youths to determine the sexual partner exchange rates for a population.  Using the HPV model, the CERL is developing, we can offer an effective vaccination solution.


Hidden Markov Models and Bayesian Disease Modeling

Bayesian models are designed to portray the dynamics of diseases in epidemiological sciences. The incidence and prevalence of diseases in a given population, with varied geographic and demographic settings, are analyzed over the temporal domain to build dynamic Bayesian networks. The network illustrates the stochastic dependencies of the demographics on the prevalence of symptoms and their related diseases. An underlying hidden Markov model is designed to inter-link the hidden disease characteristics to the observed prevalence of diseases. A disease outbreak simulator generates synthetic data of epidemic outbreaks in specific populations. While the study involves mining synthetic data generated from the disease outbreak simulator for useful information, it shall also be applied to real data from disease studies to uncover previously unknown inferences. High performance computing is a requisite to port the model onto larger domains with finer granularity of results. The Bayesian model shall aid as a predictive framework for analyzing prevalence of diseases in varied geographic and demographic settings.


Mobile Vaccination Clinics

The recent demand for influenza vaccination has resulted in ad-hoc mass-vaccination clinics held by public health departments throughout the nation. Two such clinics were held in Denton County, Texas on October 15th and 22nd, 2004. These events marked an opportunity to test strategies for clinic setup, crowd control, flow control, and logistics, which had been developed in the context of bioterrorism response. In previous years, most of the influenza vaccinations were conducted at the public health department during regular office hours, throughout the flu season. The lack of adequate supplies of vaccine nationwide, together with the specific timeframe during which vaccination could be obtained has resulted in a scenario that resembles that of a mass vaccination during a disaster situation (e.g. smallpox outbreak). Officials from the Denton County Health Department (DCHD), together with students and faculty from the Computational Epidemiology Research Laboratory at the University of North Texas, used this opportunity to observe and record problems, delays, and other adverse effects within the clinics. Faculty and students at the Computational Epidemiology Research Laboratory have been collaborating with DCHD for one year. The primary purpose of this collaboration is the design of outbreak models that facilitate the allocation of public health resources during a public health emergency. In response to what was learnt during the recent vaccination event, we believe it is imperative to investigate additional strategies for expedient mass vaccination. At the clinic we found that the elderly or mobility impaired individuals could not stand in line for hours, which required the use of ?mobile nurses? to attend to individuals who were unable to tolerate the wait. Considering that the recently conducted clinics were restricted to young children, elderly, and immune deficient individuals, waiting lines of several hundred are expected to increase to several thousands in the event of a disaster that triggers mass vaccination. In order to address some of the problems observed, we would like to model and analyze the use of mobile distributed vaccination clinics as an alternative vaccination strategy in Denton County.