The National Science Foundation (NSF) has awarded a research team from Temple University a three-year grant totaling nearly $900,000 to fund a social-science project into the tracking of human behaviors through big data.
This marks the fourth NSF-awarded grant in the last five years that an interdisciplinary team of Temple faculty members has received to study the evolution of digital artifacts using large-scale digital trace data. The collaboration joins researchers from Temple University’s Fox School of Business and College of Science and Technology (CST).
“When humans interact with digital systems, we leave a trace. Every call we make, every website we visit, it’s stamped with time and space information,” said Dr. Youngjin Yoo, the Harry A. Cochran Professor of Management Information Systems at the Fox School, and the research grant’s primary investigator. “What we do is constantly changing, and the trace data can act as DNA. What we focus on through this research is the repeat behaviors in humans that can be captured through digital trace data.
“Using those evolutionary patterns, we believe we can predict future behaviors of individuals and organizations. For example, by detecting the changes of commute patterns of individuals, we can predict overall public-transit systems’ performance in the future. Similarly, we want to be able to predict the changes in individual behaviors based on environmental changes.
Yoo said he and the grant’s co-principal investigators will study digitally enabled processes in complex digital systems, which “are like a living ecosystem, in that they constantly evolve,” he said. If patterns in the trace data represent what they call “behavioral genes,” Yoo said, alterations to those behavioral routines are “gene mutations.” Eventually, he said, the research team envisions developing software that will better predict the changes to those behavioral genes.
The benefits in doing so, according to Yoo, “are endless.” In a healthcare application, trace data could develop a pattern by which a patient sees a doctor or produce an average cost of care per patient. In an industry sense, such “gene mutations” could impact performance and cost.
“On the surface,” Yoo said, “all smart phones, for example, look the same. But everybody’s phone is different because of apps. It used to be that the product’s designer would make the product, and that was the end of the story. Now, it’s only the beginning. Millions of apps are downloaded. They’re changing constantly.
“Our argument is that, particularly in digital space, innovation never remains the same. It constantly changes and takes different forms.”
The research team includes: Yoo; Dr. Sunil Wattal, Associate Professor of Management Information Systems at the Fox School; Dr. Zoran Obradovic, Laura H. Carnell Professor of Data Analytics at CST; and Dr. Rob Kulathinal, Assistant Professor of Biology at the College of Science and Technology.
The NSF-awarded research grant runs through Jan. 31, 2018.
For one researcher at the Fox School of Business, time is literally of the essence.
Dr. Robert T. Krafty, who will supervise research into biomedical time-series data collecting, has received a grant exceeding $843,000, awarded by the National Science Foundation (NSF) and National Institutes of Health (NIH).
“The grant and all of my work looks at how we analyze data that’s collected over time,” said Krafty, Assistant Professor of Statistics. “The specific patterns of the data could tell us important information.”
The grant applications which Krafty will study pay particular attention to body signals, such as heartbeat electrocardiograms (EKGs) and brain-wave electroencephalogram (EEGs), and how these patterns are associated with different things such as measures of the quality of life or how well someone will respond to treatment.
“What I am doing is creating ways in which we can find out how these patterns are associated with certain outcomes,” Krafty said. “The main products are methods and tools that anyone can use to analyze big-time series data. The secondary goal is to apply those methods to our data on electrophysiology to see if we can help find a better way to understand how to treat sleep disorders.”
Krafty, the primary investigator for the grant, has two collaborators with whom he will work – Martica Hall, PhD, and Daniel Buysse, MD who are located at University of Pittsburgh Medical Center. Hall and Buysse are sleep-study researchers responsible for studying older adults who have trouble sleeping. Some of the patients they treat have sleeping issues due to the loss of a spouse, and others are primary caregivers for a spouse who has Parkinson’s disease. The data applied to their current research was collected from a previous study at the University of Pittsburgh, Krafty said, adding that the results produced by this grant will be used to create new statistical methods and programs to analyze collected statistical data more efficiently.
“What we want to know is what sort of patterns of physiology during sleep help indicate a better quality of life or if a patient will respond favorably to a treatment,” said Krafty.
So far in their preliminary research, Krafty and his team have found a connection between patterns of sleep and quality of life that suggests limiting the amount of sleep per night could be helpful in older adults. Krafty explained some global experts advocate that older adults should restrict their sleep. However, there was no actual evidence to back up that assumption until now, he said.
The awarded grant will also fully support one graduate student’s PhD education, Krafty said. Fox doctoral students, Scott Bruce and Zeda Li, are majoring in Statistics and have been selected by Krafty to work on the project.
The extensive time-series research Krafty is conducting will be completed by June 30, 2017.
Krafty has also been invited to speak at the NBER/NSF Time Series Conference , the leading international conference for time series data, which attracts top statisticians from around the world. At the conference, held Sept. 26-27 at the Federal Reserve Bank of St. Louis, in Missouri, Krafty will discuss the discoveries compiled in his research paper entitled, “Penalized Multivariate Whittle Likelihood for Power Spectrum Estimation.”