AP NEWS

DANBURY Student uses A.I to identify Lyme

September 17, 2018

DANBURY — An artificial intelligence program that can diagnose Lyme disease with 90 percent accuracy based solely on a photo of a rash is being developed here in Danbury.

The program is already winning national recognition, not just for its success but for its designer — Khushi Parikh, who is only 14 years old and just started her freshman year at Danbury High School.

Parikh’s project was named a national semi-finalist in the 2018 Broadcom MASTERS, making her one of the top 300 students nationwide competing in the program by the Society for Science & the Public.

She will find out Tuesday if she makes it to the round of 30 national finalists who will compete fin Washington D.C. in October.

“Artificial intelligence is all around us,” Parikh said. “Machines were originally created to make human life easier, so that’s what I wanted to do with my project — something for social benefit. Specifically, I wanted to help with Lyme disease because it’s such a big deal in Connecticut.”

The project began a year ago when Parikh was an eighth grade student at Westside Middle School Academy and preparing for the school’s intensive physical and life sciences project.

She had taken an online course in the Python computer programming language over the summer between her seventh and eighth grade years and saw an opportunity to use that to develop an artificial intelligence program and apply it to a public health program.

With Lyme’s origins in Connecticut and 95 percent of cases coming from the Northeast, she decided to teach a “neural network” to spot Lyme disease based on the skin rash it often leads to, but that can be difficult to diagnose.

“Even though only 70 percent of people get the rash, there is a clear connection between misinterpreting the rash and misdiagnosing it as something else and then being on the wrong cycle of medication,” she said. “The point is to make the doctor’s life easier so they can focus on the blood tests and making those more accurate.”

Parikh’s network needed images of Lyme rashes to learn, so she turned to Dr. Neeta Connally, a Western Connecticut State University biology professor and director of the Tickborne Disease Prevention Laboratory.

They were able to feed the program 100 images and it can now distinguish between Lyme, eczema, minor rashes and clear skin but it still needs more photos become even more accurate, Parikh said.

Westside Principal Frank Labanca said the project was one of, if not the best, projects the school has produced and “we don’t do volcanoes at Westside; we don’t do plants under colored lights,” he said.

Parikh and nine other classmates competed at the state level in the spring. Her project won the state’s physical sciences category — but it wasn’t the only project with an amazing level of sophistication for a middle school student.

Classmates Timothy Chen won the engineering category for his project studying the rotation speeds and efficiencies of wind turbine blades, and Paloma Lenz won in biotechnology for her study on how certain minerals could absorb pollutants in storm-water runoff if installed in street drains to filter the water before it goes to a watershed.

Superintendent Sal Pascarella and members of the school board honored Parikh at their meeting last week. Now she’s on to juggling both the DHS swim and robotics teams.

“I think she really embraces working independently on something that’s really interesting to her,” Labanca said. “She’s really developed the skills necessary to manage the project and work with adult mentors, meet deadlines and she exemplifies what we want in a student.”

zach.murdock@hearstmediact.com

AP RADIO
Update hourly