Archana Venkataraman, a John C. Malone Assistant Professor in the Department of Electrical and Computer Engineering, has been selected by the National Science Foundation for its prestigious CAREER Award, which recognizes early stage scholars with high levels of promise and excellence.
Using the $500,000, five-year award, Venkataraman and her team are developing new machine learning algorithms that can predict behavioral deficits in neuropsychiatric patients from patterns of brain activity measured through functional magnetic resonance imaging (fMRI). These algorithms have the potential to enhance our understanding of debilitating neurological disorders and to improve patient care in the long term.
“We’re doing this by formulating a new optimization framework that organizes the complex and messy interactions in the brain into interpretable sub-components. These components represent a collection of brain functions, such as vision or movement, which we can visualize based on their locations in the brain. These components are mathematically tied to a clinical variable of interest, like symptom severity, through a potentially nonlinear mapping.” Venkataraman said. “The project also focuses on generalizability because we can use this model to predict the level of symptom severity in new patients, and potentially even track their response to therapy. The latter implication is huge, because we can start to adapt a therapeutic regimen based on personalized responses.”
Though Venkataraman has designed the algorithms to be generally applicable to multiple neurological and neuropsychiatric disorders, her focus with this particular award will be predominately on autism and spinal cord injuries.
“The autism dataset is multisite, so we can work with a larger cohort to really evaluate the generalizability. From a neuroimaging standpoint, autism is a very subtle disorder, meaning that if I showed you a brain scan, you would not be able to tell whether or not the person had autism. Hence, we are really testing the limits of our model here,” Venkataraman said. “The spinal cord injury dataset is much smaller. We don’t have nearly the numbers for typical machine learning algorithms, but we have developed this model essentially to try to combat that problem.”
There is an educational component to Venkataraman’s work, as well. She hopes to educate the next wave of young scientists to not only have a solid foundation in signal processing and machine learning, but to also be engaged in a particular application domain. Example activities include pairing her trainees with a clinical collaborator and organizing networking opportunities for graduate students. She sees the next breakthroughs in medicine as being joint efforts between engineering and biological sciences, thus making it critical that people in both disciplines learn to work together.
Venkataraman’s group is also putting an extra focus on encouraging women to enter the STEM field. They’re planning to do this in a variety of ways, including curriculum development at the high school, undergraduate and graduate levels, and mentoring high school women through Johns Hopkins’ Women in Science and Engineering program.
For Venkataraman, helping the next generation of female engineers is paramount.
“I was very lucky to grow up with strong female role models. My grandmother taught high school physics in India. She was also a single mother to five children and made sure each one of them went on for an advanced college degree. My mother is a professor of electrical engineering; she fostered my love for math and science from the time I could talk,” Venkataraman said. “Because of these role models, it never occurred to me when growing up that there might be barriers for women in the STEM fields, and I want to do my part to make sure the next wave of young women feels exactly the same way.”