The College of Public Health and College of Liberal Arts and Sciences has invited six “junior” scholars to prepare 5-minute presentations that set forth ideas about an area of emerging importance to the general topic of Big Data, Environmental Health, and Geospatial Science. Breakout Groups will follow.

 

When: Friday, December 6 from 1-2:30 PM

Where: IATL 104

 

Presenters include:

Grant Brown (Assistant Professor, CPH) Computational strategies for inference in spatiotemporal infectious disease models

Population level models of infectious diseases are constrained not by our ability to specify biologically and sociologically plausible models, and are often not constrained by the availability of data. Instead, computational intractability tends to limit either the scope/granularity of the population under study, or the features of the models employed. I’m interested in pursuing novel Bayesian computational strategies to improve the scalability of infectious disease models to more, and more granular spatiotemporal data

Ibrahim Demir (Assistant Professor, CoEng) Intelligent Data Analytics and Communication Systems for Disasters

This project involves big data analytics applications for disasters using machine learning, intelligent systems, and virtual/augmented reality technologies. Intelligent Data Analytics and Communication Systems for Disasters

Caglar Koylu (Assistant Professor, CLAS) Big Patient Mobility Data for Identifying Medical Regions, and Spatio-temporal Characteristics and Care Needs of Patients on the Move

Patient mobility is defined as a patient’s utilization of a health care service that is located in a place or region other than the patient’s place of residence. Mobility provides freedom for patients to obtain healthcare from providers across regions (and even countries) to overcome problems such as availability, affordability, and perceived quality of service. It is essential to monitor patient choices in order to maintain the quality standards and responsiveness of the health system, otherwise the health system may suffer from geographic disparities in accessibility to quality and responsive health care.

Techniques used in geographic information science can help to identify medical regions, hospital catchment areas, spatio-temporal and service characteristics of health care utilization and demands for patient mobility (Koylu et al., 2018).

Because the US does not have a universal health care system, it is not possible to obtain data from all health care providers. However, data from Medicaid and Medicare are available. I am interested in collaborating with health service researchers to conduct a pilot study on patient flows and to assess the effectiveness and efficiency of these government supported programs

Kelly Baker (Assistant Professor, CPH) Climate Change, Big Data, and Mobile Microbes

Big Data and Mobile Microbes: Climate change is poised to disrupt ecosystems, drive migration of disease vectors and human populations into shrinking inhabitable zones, and expand habitat that supports survival of pathogens like cholera and Salmonella typhi in the environment. These dynamics could increase the amount of effort required of nations to control disease and expedite the emergence of highly virulent and multi-drug resistant bacteria in areas where pollution control is already poor. Big data could aid in the prediction and surveillance of such outbreaks, but innovative technology systems and validated spatial data on the movement of human, animal, and arthropod vectors is lacking in the global south where such knowledge is most needed.

Sanvesh Srivastava (Assistant Professor, CLAS) Bayesian modeling of massive spatiotemporal data.

With tremendous advancement in spatial referencing technologies, researchers in various disciplines have gathered an unprecedented variety of geocoded temporal data. Consequently, modeling spatiotemporal data with flexible statistical models has become an enormously active area of research over the last decade in many disciplines. Hierarchical Bayesian stochastic process-based models often provide a general framework for realistic modeling of these data, capturing structure at multiple levels. Markov chain Monte Carlo (MCMC) algorithms are used to fit these models, but MCMC computations are inefficient because every iteration of any algorithm passes through the full data. This is a major computational bottleneck and imposes severe restrictions on the applications of Bayesian models to modern large and complex spatiotemporal data. Our focus is on scaling MCMC computations for large- scale Bayesian inference using the divide-and-conquer technique that is free of any restrictive data- or model-specific assumptions.

Xun Zhou (Assistant Professor, TCoB) Spatio-Temporal Big Data Analytics for Urban Sustainability

In this presentation I will introduce various types of spatio-temporal big data generated and collected in modern smart cities, and showcase how spatial computing and analytics techniques make use of these data to improve sustainable urban management practices, such as green transportation.

Breakout Groups

Following these presentations, we will assemble breakout groups that will discuss topics of interest. Each group will have a leader and note-taker and will then report back to the main group on the prominent points of the discussion. These reports will be summarized and used to inform the structure of a subsequent meetings to discuss proposal preparation alternatives in consultation with the Office of the Vice President for Research.

 

Please contact Jill Wiley in the College of Public Health if interested in attending the event.