Friday, September 18, 2015

Science Fair Rock Star - Anika Cheerla

This is the sixth in an ongoing series highlighting some of the amazing young women participating in this year's Google Science Fair. (Here are the first, second, third, fourth and fifth posts.) Each has survived the rigorous regional rounds, and are awaiting the results of the final testing to see who will win this year's prizes. The results will be announced on September 21, 2015, and you can be sure I'll be watching!

As always, I am impressed with all the participants and their projects, but the nature of this blog is to highlight the stories of kickass women and girls, so I'm focusing on them for this series. Over the next week I will be sharing stories of many of these remarkable young women. Stay tuned!

Anika Cheerla (second from left in the photo above) is one of those amazingly well-rounded young adults you know you'll see more of as she grows up and into whatever amazing career she pursues. She's only 13 years old, but already she's an athlete, a musician, a coder, a philanthropist, a teacher, and a scientist and engineer. When her great-aunt was diagnosed with Alzheimer's she was inspired to set her sights on finding a way to improve existing diagnostic tools with the hope that earlier treatment could help delay onset and deterioration.

She set out to create an automated tool that could accurately diagnose Alzheimer's, using neural networks to train itself to diagnose Alzheimer's. Existing diagnostics include outdated mental tests and often results in doctors taking a "wait and see" approach, wasting valuable time when a patient could use beneficial medicine and other techniques to slow down the disease's progression.

Anika used her coding skills to create different neural net structures -- a single stage neural network, chained neural networks, and hierarchical neural networks -- to test each one on the ability to detect brain damage when presented with MRI scans and clinical features. She included fractal dimension, entropy and other information to boost the effectiveness, and proved that while MRI scans are important, they are not conclusive without mental examinations as well.

Amazingly, her classifier has a 95% accuracy!  She then used her coding skills to create a simplified user interface to make it easier for doctors to input the information.

Her takeaway on what she learned from this project? "What I took from this project, and what people should always remember is that you can stand on the shoulders of giants when you want to touch the stars."

For more reading, please check out the Google for Education blog post about Anika Cheerla.

Photo source: California State Science Fair


Post a Comment