Although Dean of Natural Sciences Lisa Gabel began her neuroscience studies on dyslexia when she was in graduate school, the diagnosis of dyslexia in her three children continues to motivate her research.
“They are my main driving force behind my efforts to help children and families,” she said.
Gabel’s research focuses on the identification of dyslexia in children between kindergarten and second grade.
She described her research as aiming “to identify children with reading impairments before they can read.”
Children in Gabel’s studies are presented with a virtual maze to complete, the data of which is then analyzed by machine learning.
“The path that the kid takes through the maze is used to predict how likely it is that they’re dyslexic,” said Leah Boyle ’26, a neuroscience and computer science major. Boyle is the research assistant who maintains the code on the software used. Different measurements, such as maze completion time and deviation from the intended path, are used as predictors, according to Gabel.
Another one of Gabel’s research assistants, Amanda Belej ‘27, had undiagnosed dyslexia until her freshman year of high school and cited Gabel’s research as a reason she attended Lafayette.
“I thought that would be so amazing if, in the future, a kid like me could be diagnosed young and get intervention and help early on,” said Belej, who is studying neuroscience and data science.
Katey Peretz ‘26, a biology major, is one of the researchers who work with the children on whom the maze is tested. She administers both the maze test and a standard dyslexia test, such as the Woodcock-Johnson, and then analyzes the results to see which is more effective.
“My favorite part, and why I’ve been so excited about this, is because we’re actually working with human interaction,” Peretz said. “Anytime we go into a classroom to take a kid to test, they’re so excited because they know they get to have fun.”
Student research assistant Padmanabh Kaushik ‘25, an electrical and computer engineering major, works on developing the machine learning algorithm.
“The initial results are really promising, and we have around 95% accuracy in classifying dyslexia,” Kaushik said. “But again, that’s not representative of the whole population.”
“We’re trying to take it to a level where you can just play a game online, and then we can give you a diagnosis of whether you might have dyslexia or not,” he said.
Gabel emphasized the importance of students in her research.
“It’s not just about the research that we’re doing and what’s ultimately come out of it,” she said. “It’s the training of students along the way, their participation in the study. They collect data, they analyze data, they publish, they present at professional conferences. That, to me, is just as important as what the study ultimately reveals.”