Doctoral student receives grant to develop technology for diagnosing dyslexia

UNIVERSITY PARK, Pa. -- Elizabeth Eikey, a doctoral candidate at Penn State’s College of Information Sciences and Technology (IST), is interested in how technology can be used to improve health outcomes. Eikey’s research goal, which is now being supported by the National Science Foundation (NSF), is to advance diagnostic technology that will help struggling children learn to read.

Dyslexia is “a developmental reading disorder that is characterized by difficulty with learning to read fluently and with accurate comprehension despite normal or above-average intelligence,” according to Wikipedia, and “has profound effects that span across a lifetime,” Eikey wrote in her graduate research proposal.

“By bringing dyslexia diagnostics into mainstream education, I can improve accessibility, reduce costs of diagnostics, create a better learning environment for every student (including those without dyslexia) and increase literacy in the U.S., which will foster a better society overall,” said Eikey, who graduated from Penn State with a bachelor’s degree in psychology in fall 2010 and entered the doctoral program at the College of IST in fall 2012.

As a recipient of the NSF Graduate Research Fellowship, Eikey will now have an opportunity to explore learning disabilities technology in depth. The NSF Graduate Research Fellowship Program (GRFP), according to the NSF website, “recognizes and supports outstanding graduate students in NSF-supported science, technology, engineering and mathematics disciplines who are pursuing research-based master's and doctoral degrees at accredited U.S. institutions.”

With its emphasis on support of individuals, according to the NSF website, GRFP offers fellowship awards directly to graduate students selected through a national competition. The GRFP provides three years of financial support within a five-year fellowship period ($32,000 annual stipend and $12,000 cost-of-education allowance to the graduate institution) for graduate study that is in a field within NSF's mission and leads to a research-based master's or doctoral degree.

According to Eikey, “there is an increasing need to find more effective and cost-efficient diagnostics for dyslexia.” In her NSF Graduate Research Proposal, she cited that “dyslexia affects as many as one in five people in the U.S., and has been linked with lower literacy rates among children and higher rates of unemployment among adults.” Furthermore, “students with dyslexia have average or above average IQ, but 20 percent drop out of high school (compared to 8t percent of students in the general population).”

“Although diagnosis prior to the third grade is key to future success,” Eikey said, “many children go undiagnosed.”

Eikey, who is co-advised by Erika Poole and Madhu Reddy, primarily studies issues related to obesity and eating disorders. However, she was motivated to explore the area of learning disabilities by a desire to develop a more effective method of diagnosing dyslexia than the current practices, which are mainly subjective. Shortly after receiving her bachelor’s degree from Penn State, she started working at Best Buy, a consumer electronics chain. A number of the customers that she served, she said, inquired about technology for their children with learning disabilities and other health issues.

“I saw there was a need for (that type of technology) and wanted to pursue it further,” she said.

“While it is recommended that professionals with expertise in psychology, language and education assess children suspected of having dyslexia,” Eikey wrote, “teachers often perform evaluations within schools. Often, teachers use observational techniques, IQ discrepancy models or Response to Intervention (RTI) as diagnostics.” “Unfortunately,” she wrote, “those methods are often not objective or specific enough. Observations are often conducted over many months, and diagnoses focus on learning progress as a measurement.”

“For the most part, teachers watch students for signs of dyslexia,” Eikey said. “You have to wait a long time to get those long-term measures.”

Eikey’s proposal for an alternative method of diagnosing dyslexia involves using machine learning technologies to classify eye movements. “For some time,” she wrote in her proposal, “abnormal eye movements have been linked to poor reading. Studies have shown that people with dyslexia have longer fixation durations, shorter saccade length and greater regression frequency than normal readers.” 

"Those findings,” Eikey said, “can be used in conjunction with eye-tracking and tablet technology to diagnose dyslexia.”

“With advancements in eye-tracking technology, eye gaze data are becoming easier to acquire and more accurate,” she wrote in the proposal. “Eye-tracking software is available for tablets and PCs, and tablets are readily available and relatively inexpensive. Since eye movements of those with dyslexia differ from those without, eye-tracking technology and machine learning methods may hold the key to quickly and accurately differentiating those with dyslexia from those without. A number of previous studies use machine learning algorithms for reading.” Based on this work, she said, “we can develop a tablet technology to diagnose dyslexia in children.”

As a first step of Eikey’s research, she is working with another IST doctoral candidate, Kyle Williams. Together, they are figuring out what the eye movement data of those with dyslexia looks like so they can determine the most effective techniques for classification. The next step will be to test the classification model in an experimental setting.

While Eikey said that she is optimistic about the effectiveness of the machine learning technology in diagnosing dyslexia, she admitted that actual implementation could take years. She must first determine the accuracy of the algorithms used. Afterward, she would need to consult with teachers, parents and other stakeholders to understand their needs when designing the technology.

“Building the technology is only the beginning of actually deploying it within schools; we need to design in an iterative process,” Eikey said. “It may start small and local, but by establishing the viability of this type of technology, we are getting one step closer to helping these children.”

Media Contacts: 
Last Updated May 08, 2014