How has Ani Eloyan, only five years past her Ph.D., already conducted published research on attention deficit and hyperactivity disorder, autism, multiple sclerosis, epilepsy, and Alzheimer’s disease? Because she is a biostatistician. She can make fundamental contributions to a wide variety of brain science investigations.
Eloyan, assistant professor of biostatistics, works with clinical specialists in each of those conditions and others, but her focus is on the statistical techniques for interpreting the voluminous data of brain scans. In particular she has developed methods to mine insights from magnetic resonance images of the brain’s structure and function.
Often Eloyan’s goal is to discern meaningful connections and relationships among different regions of the brain. Are areas of interest lighting up in the scans in a significant, synchronized pattern that suggests they are part of the same network? Are the patterns of activation different in people with a disorder than in people without that disorder?
For Eloyan, the interest in math came first. At Yerevan State University in her native Armenia, Eloyan earned her bachelor’s degree in applied mathematics and computer science in 2004. She then came to graduate school at North Carolina State University to earn a master’s degree and a Ph.D. in statistics. It was during an exam at the end of her second year of the doctoral program that she acquired the interest in brain science.
The test included writing a review paper about an assigned topic in a month. Her topic was independent component analysis (ICA), a particular way of discerning interpretable “low-dimensional” features from massive datasets. There are many applications, but one in particular appealed to her.
“I found that one of the applications was brain imaging,” Eloyan said. “When I started working on my Ph.D. thesis, I was interested in the statistical problem mainly because of this one application.”
After graduation in 2010, she seized the opportunity to apply her learning to brain science by taking a postdoctoral scholar position at Johns Hopkins University. Within a year she had a prize to show for it. She and her group won the ADHD200 Global Competition, beating 20 other teams by analyzing MRI, demographic, and cognitive data from children with and without ADHD to predict disease diagnosis.
In that contest Eloyan applied techniques such as machine learning, which are automated methods that learn from data. In other papers she has applied many other methods, including ICA. For example she was the lead author of a study in Biostatistics in 2013 applying ICA to explore the function of the brain of a large population of healthy adults, and is the co-author of an upcoming study finding unusual links between the visual and motor regions of the brain in children with autism.
At Brown and its affiliated hospitals, where brain science is an enterprise involving more than 100 faculty members, Eloyan will have abundant opportunities to participate in more discoveries.
“I love working with people,” she said. “My collaborators either already have their data or are thinking about collecting data and need help with their statistical analysis or designing their experiments.”
And when she shares her expertise, she is gratified to learn in return.
“For me it’s about the statistical problem and learning as much as I can about the data,” she said. “I also love learning about diseases.”