Avyuk Dixit’s trip to Pondicherry, India, was a learning experience in more ways than one.
The Mount Vernon resident spent four weeks last summer working with medical professionals on the other side of the world to develop a machine learning model that can diagnose glaucoma.
Now a junior at Thomas Jefferson High School for Science and Technology in Alexandria, Dixit says his time in south India not only gave him the opportunity to conduct research in a real-world environment, but it also opened his eyes to a different culture as well as the challenges that many people face in terms of accessing adequate healthcare services.
“It was really just a humbling experience all around to go somewhere and know that I could be helping so many people,” Dixit said. “It made me realize how privileged I was to be in a position where I have access to resources. I have access to a great healthcare infrastructure and a great education, whereas so many people in India don’t have access to those same resources.”
Dixit became interested in computer science at a young age thanks to his father, a data scientist who introduced him to the subject and encouraged him to learn new programming languages and skills.
When he began attending Thomas Jefferson, Dixit enrolled in courses on more advanced topics like artificial intelligence that fueled his fascination specifically with computer vision, a field that uses artificial systems to obtain information from images and videos, and machine learning, a type of data analysis where computer systems are trained to identify patterns and make decisions from a set of data without needing explicit instructions.
Dixit started to apply what he learned in the classroom to a project of his own when he learned in his sophomore year of high school that his uncle had been diagnosed with glaucoma.
Glaucoma is a group of eye diseases that damage the optic nerve, potentially resulting in vision loss and blindness, according to the National Eye Institute.
Dixit noticed how the disease affected his uncle, a teacher who lives in New Delhi, India, during a visit.
“I saw how his declining vision hurt his ability to teach, read fine-printed textbooks, and detracted generally from his livelihood,” Dixit said.
In response to his uncle’s dilemma, Dixit decided to develop a machine learning model capable of diagnosing glaucoma as effectively as any human screening processes.
According to the National Eye Institute, glaucoma can be detected using a comprehensive eye exam that tests the individual’s ability to see at various distances and their peripheral vision. The exam can also include a pupil dilation to examine the retina and optic nerve and measurements of pressure inside the eye and the cornea’s thickness.
There is currently no cure for glaucoma, but an early diagnosis is critical for getting treatment that may delay the disease’s progression, the National Eye Institute says.
However, getting a diagnosis can be difficult for people who do not have convenient access to ophthalmologists, doctors who specialize in the detection and treatment of eye disorders.
After initially developing his machine learning model with the help of his teachers at Thomas Jefferson, Dixit presented his model to an ophthalmology professor at Johns Hopkins University.
The professor suggested that Dixit travel to India to work with the Aravind Eye Care System, which could eventually integrate his model into its existing electronic medical records system.
“They felt it would be helpful for screening populations in underdeveloped areas,” Dixit said.
That is how Dixit wound up at the Aravind Eye Hospital in Pondicherry last summer.
In addition to working on his machine learning model at the hospital, Dixit joined Aravind doctors on weekend trips to rural villages where they set up eye camps to screen for diseases.
During these trips, Dixit got a firsthand look at the challenges that medical professionals face in delivering healthcare to rural populations, since Aravind Hospital’s limited resources and staffing meant that the doctors essentially had to take weekends off to conduct the eye camps.
“Only using humans…means we get to screen smaller amounts of populations and are less efficient in reaching as many people as possible,” Dixit said. “I think that having a kind of model for screening that can do it automatically through computers allows hospitals to open up their resources and be more efficient and reach the people that need this the most.”
Dixit also became immersed in the Tamil language, customs, and food of Puducherry, the southeastern state that boasts Pondicherry as its capital and most populous city.
“It was pretty accommodating to go to Aravind, and they were really helpful in getting me all set up and making me feel at home,” the Thomas Jefferson student said.
While Aravind Eye Hospital is not using Dixit’s work yet, he has now mostly completed the machine learning model, which uses an algorithm to analyze images of eyes to detect glaucoma.
The model has an accuracy rate of 87 percent, according to Dixit.
Since returning from India, Dixit also started a small business called Linkpedia that he sees as a way to apply machine learning to local community projects.
“I really hope that I can just continue to use machine learning to help people both on a larger and smaller scale,” Dixit said.
Projects that the company is currently tackling include the use of surveillance videos to detect potentially violent behavior in a crowd and helping the Mount Vernon Estate understand how weather affects tourism.
Dixit’s work has already drawn the attention of Fairfax County leaders like Mount Vernon District Supervisor Dan Storck, whose office asked the student to share his experience in India on Feb. 14.
“I am continually awed and inspired by the talented youth in Fairfax County, like Avyuk, who push boundaries and strive to make our world a better place for all,” Storck said. “It is achievements like these that will help solve the challenges we face as a society.”