Around the world, approximately 50 million people suffer from epilepsy, a disorder of the central nervous system that causes sudden, unpredictable seizures. But what if you could train an AI algorithm to detect a seizure before it happens? For Jeremy Pemberton, a Maricopa Community College student and a scholar in the AI for Workforce Program, this question wasn’t just academic. He has a personal stake in the issue. “My daughter suffers from seizures, so this project is near and dear to my heart,” he said.
That project was conceived for Intel’s AI Global Impact Festival, which Pemberton participated in with classmates, Wyatt Whitney and Arnav Bawa. After debating what idea to pursue for the festival, the team decided to develop this AI algorithm due in part to Jeremy’s personal connection to the issue.
When the team began working on the project, they consulted a study by researchers from the Massachusetts Institute of Technology (MIT). The study monitored the brainwaves of 23 participants. It identified the data points needed to detect seizures. What researchers found was that too much data, such as electroencephalogram (EEG) monitoring or data from multiple patients, makes the detection process less accurate because brainwave activity varies too much from person to person.
With this knowledge, Pemberton, Bawa, and Whitney decided to develop an AI model to focus on managing one patient at a time. In developing and testing this AI-powered algorithm, they found that focusing on one patient increased the accuracy of the model to 83%. Furthermore, as the AI monitored the patient over time, it could learn and become more accurate in predictions. This could vastly improve care for people who suffer from epilepsy. It can enable healthcare professionals and family members to prepare the individual for the seizure. As well as, move the individual to a safer location or position.
Pemberton plans to continue his research, understanding the impact that it could have on the lives of individuals like his daughter. He said that the program has been critical to this early research. Jeremy is encouraged by Intel’s support. “I’m getting so much value from the AI for Workforce Program, and I hope to be able to pay that forward and give back in the future.”