TDA awards seed grants to four interdisciplinary teams

Translational Data Analytics has awarded seed grants to four faculty teams tackling challenges ranging from detecting unexploded land mines to predicting dementia. Selected by a group of 18 faculty reviewers from throughout the university, the groups receiving the awards represent nine departments in four colleges. The funding, which totals nearly $120,000, will be available to the awardees Jan. 1 – Dec. 31, 2017.

Now in its second year, TDA’s seed grant program is designed to foster a dynamic, interactive community of data analytics scholars dedicated to identifying, defining, and providing translational solutions to real-world problems. It encourages the formation of new, interdisciplinary teams to pursue new opportunities or novel extensions of existing activities and generate preliminary study concepts, technologies, data, and results that will foster subsequent applications for extramural funding.

Congratulations to this year’s grant recipients!

The awardees:

Project Title: “Building articulatory models for large child-caretaker vocal interaction corpora”

Team

  • Derek Houston, Principal Investigator, Associate Professor, Otolaryngology-Head and Neck Surgery, College of Medicine; Director, Buckeye Hearing and Development Center at Wexner Medical Center
  • Amanda Miller, Co-Principal Investigator; Senior Researcher, Department of Linguistics, College of Arts and Sciences
  • Eric Fosler-Lussier, Faculty Participant; Professor, Department of Computer Science and Engineering, College of Engineering
  • Andrew Plummer, Post-Doctoral Researcher, Department of Linguistics, College of Arts and Sciences
  • Jongmin Jung, Post-Doctoral Researcher, Department of Otolaryngology-Head and Neck Surgery, College of Medicine

Overview
Speech sound production is fundamental to oral communication. Studies in the mid-1990s showed dire real-world consequences of speech differences found between toddlers who could hear their own early vocalizations and those who could not. Speech sound disorders are observed in 2 – 25% (median = 8-9%) of children, with the higher proportion found in younger children. Better understanding of children’s articulatory mechanisms by building early development models will directly aid Speech Language Pathologists teaching young children to become more competent communicators. We propose to develop novel methods for estimating articulatory configurations and motor control strategies used by infants and young children to produce vocalizations, and build models of speech production that are appropriate for infants and toddlers. Throughout the duration of the project, speech-language pathologists from Nationwide Children’s Hospital who are working in the Buckeye Center for Hearing and Development will be actively involved in providing input about their needs and the challenges they face in working to improve the articulation of children experiencing speech sound disorders.

Award Amount: $26,300.75

 

Project title: “Optimization of agricultural conservation practices under a changing climate”

Overview
Agricultural lands in watersheds draining to western Lake Erie are under increasing pressure to maintain high crop yields while reducing the environmental, and particularly water quality, impacts of intensive agriculture. In August 2014, a harmful algal bloom in western Lake Erie compromised the water supply of the city of Toledo and led to a three-day “do not drink” advisory for the city and surrounding areas. This was due to a combination of agricultural nutrient runoff and climatic conditions; Lake Erie’s western basin has had several larger algal blooms in other recent years (2011 and 2015). In February 2015, the U.S. states and Canadian province bordering Lake Erie agreed to a phosphorus load reduction to the lake that scientists expect will reduce the size of algal blooms (and other nutrient enrichment problems in the lake) to a more manageable level.

The primary source of Lake Erie’s nutrient enrichment is fertilizers and manures applied to agricultural lands that run off or leach into streams, rivers, and ultimately the lake. Decision-makers acknowledge that reaching water quality goals will be challenging and may reduce crop yields in the region. At the same time, agricultural stakeholders are already adapting their growing techniques to maintain high yields in a climate that has hotter and drier summers and wetter springtime conditions, when getting on the field is critical. The current political pressure to improve water quality, when added to the extreme events farmers are already experiencing due to climate variability and change, creates an opportunity for farmers and other stakeholders in the region to adopt more resilient farming practices.

Despite significant advances in quantifying how climate change will impact agricultural production, significant knowledge gaps remain, particularly related to future climate extremes (i.e., floods, droughts). These gaps include: (1) spatial and temporal scale mismatches, (2) lack of agreement between climate models and scenarios, and (3) impacts that are a result of complex coupled human-natural systems are difficult to accurately define and simulate. The goal of this project is to support decision-making under deep uncertainty by identifying an optimal set of conservation practices for agricultural production given future climate changes.

Team

  • TDA affiliate Margaret Kalcic, PI; Assistant Professor, Department of Food, Agricultural and Biological Engineering, College of Food, Agricultural, and Environmental Sciences
  • TDA affiliate Steven Quiring, Co-PI; Professor, Atmospheric Sciences Program, Department of Geography, College of Arts and Sciences

Award amount: $30,000

 

Project title: “Quantifying the structure of human experience with data streams from smartphones”

Team

  • Jihun Hamm, Research Scientist, Department Computer Science and Engineering, College of Engineering
  • Per Sederberg, Associate Professor, Department of Psychology, College of Arts and Sciences
  • TDA affiliate Mikhail Belkin, Associate Professor, Department of Computer Science and Engineering, College of Engineering

Overview
By understanding how people go through each day along with what they perceive and learn during real-life events, we can have a tremendous impact on many fields. For example, a population-level model of what people do when and where can improve urban planning and public resource management. An individualized model of daily experience can be used for computerized personal assistants or detecting changes in one’s behavioral pattern that might be linked to an illness. Statistical patterns of daily events have been mined from databases such as American Time Use Survey (ATUS), which consists of timelines of daily activity and places collected from a large number of (n~50,000) participants. Yet, with a phone interview-based survey like ATUS, it is practically impossible to collect data in a time scale beyond a few days. Nor can we obtain fine-grained measurements of individual experiences from surveys with enough detail to differentiate patterns of, e.g., people with mild cognitive impairment and healthy groups.

However, the situation has changed with the proliferation of smartphones, with an estimated 200 million devices in the U.S. alone. Smartphones are an enabling technology for collecting behavioral data unobtrusively from people in their everyday environments, through continuous sensing of location, motion, audio, image, and calling and messaging activity. The PIs and their collaborators have worked on building such a life logging system, which we have validated and is currently ready to deploy for data collection initiatives.

Using rich sensory and meta-data acquired from the system, we intend to develop machine-learning algorithms to quantify daily human experience by event segmentation, classification, and prediction. The specific aims:

  • Capture daily experience of individuals using continuous sensory recordings from smartphones
  • Build accurate machine learning algorithms for segmenting, classifying, and predicting events from sensory data
  • Conduct a pilot study with 20 participants from healthy and mildly cognitively-impaired group to: a) quantify changes in the structure of experience that predict disease, and b) lay the foundation for a memory intervention

Award amount: $30,000

 

Project title: “Unexploded ordnance and agricultural productivity: Using remote sensing to detect the long-term impact of U.S. bombing of Cambodia”

Team

  • Erin Lin, PI; Assistant Professor, Department of Political Science, College of Arts and Science
  • TDA affiliate Rongjun Qin, Co-PI; Assistant Professor, Department of Civil, Environmental and Geodetic Engineering and Department of Electrical and Computer Engineering, College of Engineering

Overview
The weapons left behind from war—carpet bombs, cluster munitions, landmines—pose a significant risk to the communities left to live on that land. However, unexploded ordnance (UXO) are difficult to find, oftentimes hidden underneath several inches of ground. It remains a challenge for farmers, clearance teams, and aid organizations to identify contaminated areas.

Instead of traditional, high-risk in-person surveying, we will develop a remote-sensing method to identify the location of unexploded ordnance. This method combines declassified U.S. Air Force data on payload drop sites from the secret bombing of Cambodia with high-resolution maps of terrain features and contemporary vegetation. By developing machine-learning methods on the images, we can identify an important element of post-bombardment geography: bomb craters, which provide a physical evidence of the number of bombs that have detonated within each drop zone. The U.S. Air Force data provide the number of bombs that were dropped in each payload, with the coordinates of each drop site. Since, for each payload, we know the total number of bombs dropped and the number of bombs detonated, we can estimate the number and density of unexploded ordnance left within each drop zone.

As part of the process, we will use moderate resolution multispectral imagery to identify crop activity and yield, a new method that employs both temporal and reflective spectral statistics to generate granular data on cultivation and crop productivity. These remotely sensed images have been captured, rendered, and disseminated every eight days since 1998 through a U.S. government initiative, so we can develop a rich sense of land use patterns for the past two decades.

The impact of UXO on the quantity of arable land can be estimated from these data using maximum likelihood techniques or a threshold regression model. By the end of data collection, we will be able to estimate the degree to which UXO change the total area of active agricultural land, and how persistent this effect is, from 1998 to present day. The impact of UXO on arable land will be reported to raise the awareness of the importance of de-mining.

Award amount: $29,960

Share this page
Suggested Articles
Visualization challenge teaches the art of telling data’s stories

The winning team’s presentation At first blush, Translational Data Analytics and creativity might seem worlds apart. When it comes to data collection, after all, “creative” is often a euphemism for...

A science first: Measuring glaciers from the sky

Researchers from Ohio State recently returned from a journey measuring the internal temperatures of the Greenland ice sheet. What makes this feat so impressive is for the first time they...

Ohio State hosts popular network science conference

With support from the Translational Data Analytics Institute, TDAI affiliates Skyler Cranmer and Janet Box-Steffensmeier from the Department of Political Science, along with Alison Craig, hosted colleagues from nearly 70...

Ümit Çatalyürek named an IEEE Fellow

Ümit Çatalyürek, TDA affiliate and professor and vice chair of academic affairs in the Department of Biomedical Informatics, was recently named the newest IEEE Fellow at Ohio State in recognition of...

Big Data for Good: Eye in the sky

In  a new  Big Data for Good feature, TDA affiliate Rongjun Qin, professor of computer science and geodetic engineering, discusses his work developing algorithms that interpret data from satellites, drones, and...