Projects
Evolution of learned behaviors in genetic and geographic contexts
Director: Nicole Creanza, Arts & Sciences
Collaborator: TS Harvey, Arts & Sciences
About
Dr. Creanza’s lab focuses on the evolution of learned behaviors, integrating analysis of genomic and behavioral data to understand geographic patterns in genetic and cultural evolution. In both empirical systems that we study—the learned vocalizations of songbirds and the languages of human populations—we have found that learned behaviors can retain evolutionary information across great distances and over long timescales. In this line of research, we analyze linguistic data to bring new perspectives to geographic questions in population genetics.
Anthropogenic linguistic patterns
Research Goal: Speculation remains about the demographic history of Native Americans, which is complicated by multiple waves of migration across the Bering Strait, admixture between groups, and numerous potential migration routes.
Methods: We are merging genotyped populations to assemble the largest Native American genome dataset and then test whether signals of population history in combined genomic and linguistic data can shed light on hypotheses for multiple independent migrations into the Americas.
Ornithological song patterns
Research Goal: As modern cities continue to grow and sprawl, there is increasing concern about how wildlife communities are affected. We continue our focus on the relationship between genes and language by focusing on the effect urbanization has on birdsong characteristics.
Methods: We will analyze many species across the U.S., merging citizen-science bird recordings with nighttime-lights satellite images (a proxy for urbanization). To analyze thousands of songs with consistency, we developed song analysis software. Our software enables the involvement of Vanderbilt biology students in this large-scale study (BSCI 1512L). Combining these techniques for song analysis and geospatial analysis of satellite images, we will explore whether urbanization affects birdsong characteristics.
Using isoscapes to identify origins of victims from the Shining Path War, Peru
Director: Tiffany Tung, Arts & Sciences
Collaborator: Michael Newton, Law School
About
The Shining Path war in Peru of the 1980s-1990s—a conflict between the Peruvian State and insurrectionist guerillas—led to the genocide of 69,000 mostly indigenous peoples. Forensic teams have recently exhumed thousands of bodies, and Dr. Tung has provided pro bono assistance to them by analyzing skeletons for violence-related trauma.
Helping human rights teams by providing information on victim origins
Research Goal: To identify the geographical origins of massacre victims, information that enables human rights teams to search those areas for surviving family members and provide that information to human rights teams, enabling them to search those areas for surviving family members.
Methods: We achieve this by tracing isotopes from the victims’ bones and teeth to specific locales in Peru. Oxygen, strontium, and lead isotopes naturally occur in water and soil, and they are incorporated into the skeleton through the food chain. Thus, isotope ratios from victims can be tied back to particular geographic areas.
Dr. Tung has processed hundreds of isotope samples from water, soil, plants, and skeletons from her bioarchaeological research, and those data can now be applied to this humanitarian cause. Doing so will require generation of predictive isoscape maps to narrow the potential geographic origins of victims. Such isoscape modeling can only be achieved using advanced geospatial tools. Professors Tung and Newton (Law School) have complementary areas of expertise to help ensure that the forensic and spatial data can be appropriately applied in legal contexts. Newton and his students have provided legal advice to the Peruvian government, is a member of the International Institute of Humanitarian Law, and was an appointed expert to the Task Force on Genocide Prevention.
Understanding spatial variation in healthcare delivery and outcomes
Director: Stephen Deppen, School of Medicine
About
Local differences in population, economics and built environment drive variation in healthcare delivery and outcomes, and are often found to vary geographically. Simply measuring individual risk factors or market forces are not sufficient to understand resulting population outcomes. Defining and measuring the interaction of geographically varying disease and the local mosaic of causes is necessary to better tailor interventions and policy.
Differentiating lung cancer from benign histoplasmosis
Research Goal: Benign infections from mycotic fungi like histoplasmosis capsulatum, generate benign lung injuries that mimic lung cancer and defy our best imaging technologies’ ability to discriminate benign from malignant disease. Spatial analysis can help differentiate cancer from infection, reducing the number of unnecessary lung surgeries.
Methods: These questions require spatial analytical knowledge to answer. The complexity of our proposed multivariable spatial models requires not only dedicated understanding of spatial data manipulation, but complex modeling and supercomputer programming skills. Such skills transcend those commonly found at the medical center or biostatistics. Partnering with VIIGR and SARL enable us to further these questions and define and implement locally specific solutions to the problems arising from variation in healthcare delivery and outcomes.
Population mobility and engagement in HIV care
Director: Kate Clouse, School of Medicine
About
South Africa is home to more people living with HIV than any other country and to the world’s largest antiretroviral therapy (ART) program. The national treatment program is under tremendous pressure to ensure continuous, lifelong care to the four million adults and children on treatment. One group at high risk of dropping out of HIV care is postpartum women. South Africa has a highly mobile population, with a pattern of frequent cycling between urban and rural areas due to employment and family obligations.
Cell phone usage and mobility
Research Goal: Dr. Clouse’s recent work has shown that recently postpartum women are highly mobile, often returning to rural homes for months at a time, which may interrupt HIV care for the mother and infant. However, characteristics of this mobility still are poorly understand.
Methods: Cell phone usage is ubiquitous and growing in South Africa. Through a Development Award from the TN CFAR, Dr. Clouse developed CareConekta, a smartphone app that uses the phone’s GPS to record patient locations to prospectively characterize mobility and allow real-time intervention. CareConekta is the first smartphone app to track mobility for improving health outcomes. This innovative project will generate large volumes of spatial data and will require the advanced GIS analytic capabilities. The concepts demonstrated through this work can apply to mobile populations in any geographic area and across disciplines.
Geospatial analysis for optimizing delivery of HIV care
Director: Carolyn Audet, School of Medicine
About
In rural Mozambique persons living with HIV (PLHIV) are trying and failing to manage their health. Only 60% of patients are retained in care one year after initiation of treatment. To stem high levels of loss to follow up, Ministry of Health policy mandates the use of “treatment partners” to support adherence for PLHIV on antiretroviral therapy. Many patients are hesitant to recruit someone in their “circle of trust” for fear of disclosure. Hence, they forego care.
Traditional Healers as Caregivers
Research Goal: While patients are often uncomfortable disclosing their status to friends and family, they do trust traditional healers.
Methods: Cell phone usage is ubiquitous and growing in South Africa. Through a Development Award from the TN CFAR, Dr. Clouse developed CareConekta, a smartphone app that uses the phone’s GPS to record patient locations to prospectively characterize mobility and allow real-time intervention. CareConekta is the first smartphone app to track mobility for improving health outcomes. This innovative project will generate large volumes of spatial data and will require the advanced GIS analytic capabilities. The concepts demonstrated through this work can apply to mobile populations in any geographic area and across disciplines.
Massive scale archaeological survey using convolutional neural networks
Director: Steve Wernke, A&S
Collaborator: Ralph Bennartz, A&S
Collaborator: Andreas Berlind, A&S
About
Archaeology contributes unique insights into the material and spatial dimensions of the human story, but big interregional networks in antiquity are notoriously difficult to document. Attempts to scale up traditional survey through multispectral Remote Sensing have met with limited success, because archaeological sites seldom produce spectral “signatures.” Advances in Convolutional Neural Networks (CNNs) in fields as diverse as medical imaging and astrophysics provide promising pathways to automated site detection, as CNNs can learn to identify the distinctive shapes of objects in images. However, CNNs require thousands of training images labeled by people to learn the shape properties of archaeological features.
Neural networks as pattern finders
Research Goal: By creating tools to build these training datasets, we hope to execute CNN-based automated archaeological survey on a massive scale in Andean South America.
Methods: With support from an NEH Digital Humanities Startup Grant, Dr. Wernke with co-PI Parker van Valkenburgh (Brown University) have built GeoPACHA (Geospatial Platform for Andean Culture, History and Archaeology), an online, edited archaeological discovery tool. GeoPACHA enables “virtual archaeological survey” by visually scanning high resolution satellite imagery. GeoPACHA immerses students in archaeological discovery as they learn to identify archaeological sites over tens of thousands of square kilometers. The thousands of sites identified with GeoPACHA will be used to train a CNN-based system. Our implementation-scale grant proposals are in review at ACLS and NEH.