If we can catch early signs of dyslexia and dysgraphia, we can get children the care they need.
Venu Govindaraju, SUNY distinguished professor in the department of computer science and engineering at the University at Buffalo, looks at how to do so.
Venu Govindaraju is vice president for research and economic development at the University at Buffalo, as well as a SUNY Distinguished Professor in the university’s Department of Computer Science and Engineering. A pioneer in artificial intelligence and machine learning, Govindaraju leveraged these technologies for handwriting recognition, transforming the United States Postal Service and the postal industry globally. Currently, Govindaraju is applying his expertise in handwriting recognition to assist children with Dysgraphia and Dyslexia as director and principal investigator of the National Artificial Intelligence Institute for Exceptional Education. This effort is supported by a $20 million grant from the National Science Foundation and the Department of Education’s Institute of Education Sciences. He received his PhD in computer science from UB in 1992
Training AI to Spot Dyslexia
Decades ago, my colleagues and I trained AI technologies to read handwriting. This groundbreaking work enabled the U.S. Postal Service and others to automate the sorting of mail, an improvement in speed and quality that saved these groups billions of dollars.
Now, instead of names and addresses written on envelopes, my research team is training AI models to recognize early signs of dyslexia and dysgraphia in children’s handwriting.
These neurodevelopmental disorders negatively affect reading, writing and even speech. Current screening tools are effective, but can be costly and time-consuming – and, with a nationwide shortage of speech language pathologists and occupational therapists, many children are being diagnosed too late, if at all.
To address this, our team has developed a screening tool that combines several AI models together, summarizes their findings and provides a comprehensive assessment.
The tool uses machine learning and natural language processing to spot misspellings, grammatical errors, limited vocabulary and other indicators of dyslexia and dysgraphia. For writing via a tablet, the models can even analyze writing speed, pen movements and more.
Using writing samples from young students, we’re training the models to complete the Dysgraphia and Dyslexia Behavioral Indicator Checklist. Next, we’ll compare the models’ effectiveness to teachers and speech therapists who administer the test. We’ll also gather insights from these professionals to confirm the tool is viable in the classroom and elsewhere.
Ultimately, the tool will be an aid to speech-language pathologists and occupational therapists, who will continue to play critical roles diagnosing and treating these conditions. It will help identify these learning challenges early, and ensure that children receive the care they need, positively impacting their long-term learning and socio-emotional development.
