Fairfax Co. college students use AI to create an algorithm for classifying mind alerts


Two Fairfax County college students used AI to create an algorithm that can be utilized to categorise EEG alerts, which can be utilized to regulate prosthetics.


Fairfax County college students Gautham Ramachandran and Sriram Nalini used AI to create an algorithm that can be utilized to categorise EEG alerts, which can be utilized to regulate prosthetics. (Courtesy Gautham Ramachandran)

About seven months in the past, Chantilly High School scholar Gautham Ramachandran got here throughout a person in a wheelchair on the sidewalk.

Ramachandran mentioned he was impressed by the person who was navigating life with out his legs. Ramachandran approached him, and so they exchanged pleasantries.

In the times that adopted, Ramachandran mirrored on his want to assist individuals like the person he spoke to.

“I used to be wrestling with a urgent query: How can we leverage expertise to offer autonomy and empower people like him?” he mentioned.

Ramachandran and his classmate, Sriram Nalini, realized the reply to that query concerned synthetic intelligence. They used AI to create an algorithm that can be utilized to categorise electroencephalogram (EEG) alerts, which can be utilized to regulate prosthetics.

“It’s a key to a brand new realm of interplay, the place your ideas seamlessly translate into motion; the place expertise is an invisible, intuitive extension of the human expertise,” Ramachandran advised WTOP.

Ramachandran has been programming since he was in third grade, and used to create video video games utilizing HTML and JavaScript. Instead of enjoying the video video games, although, he mentioned he “obtained hooked on creating them,” and spent hours coding.

When he was in center faculty, Ramachandran mentioned he launched a sport on the platform Steam. But none of that in comparison with his newest efforts.

Ramachandran mentioned he and Nalini thought-about several types of accessible expertise for analyzing mind alerts. In some instances, sufferers get implants, which Ramachandran mentioned are modern however “require a degree of invasiveness that isn’t actually best for everybody.”

A standard EEG headset is one other different, however Ramachandran mentioned these are difficult “on account of needing so many electrodes, and admittedly, not as correct as we’d hope, particularly on the subject of classifying actions.”

So, the 2 college students took the EEG expertise and streamlined it.

“We mainly used this AI algorithm because the embodiment for this imaginative and prescient,” Ramachandran mentioned. “It’s designed to work with only a handful of electrodes.”

The innovation helps with accuracy, he mentioned. Previously, the utmost accuracy he’s seen is 98%, “which implies that there’s going to be two errors out of each 100 actions.”

With the assistance of the algorithm, Ramachandran mentioned there’s “classification accuracy that reaches as much as the zenith, as much as 100% accuracy, which is actually not simply an enchancment, it’s a revolution.”

The expertise, Ramachandran mentioned, “permits individuals to mainly management prosthetics or management units utilizing their mind alerts.”

The pair revealed their findings in a analysis paper.

Their work, he mentioned, helps “present an economical resolution to prosthetics.”

Get breaking information and day by day headlines delivered to your electronic mail inbox by signing up right here.

© 2024 WTOP. All Rights Reserved. This web site isn’t supposed for customers positioned throughout the European Economic Area.



Source hyperlink

Leave a Reply

Your email address will not be published. Required fields are marked *