This Trio Is Code Dependent

Brittany Wenger

Sophomore, Double-major in computer science and biology

In middle school, Brittany Wenger became fascinated by artificial intelligence.

Using textbooks and online tutorials, she taught herself to code a program that could play soccer. On a gray screen, x’s and y’s symbolizing players scrabbled for an “o” representing the ball. After dozens of games, one of the teams started to win. The program was learning.

Wenger had created an artificial neural network, an algorithm that mimics the human brain and can process vast amounts of information. Her soccer program was just the beginning. A few years later, Wenger created a computerized “brain” that can detect breast cancer.

Detecting cancer early is vital, but it can be tricky: Sometimes healthy cells masquerade as noncancerous cells, and vice versa. Wenger’s app, Cloud4Cancer, can diagnose breast cancer with more than 99 percent accuracy. Doctors can enter information about nine attributes of suspect cells in a breast mass and, with one click, learn whether the mass is malignant or benign. The app is being beta tested in hospitals in the U.S. and abroad; it can test for leukemia as well as breast cancer.

Wenger won the Google Science Fair competition for her app in 2012, thrusting her into the national spotlight. She’s given several TEDx talks; met her idol, Internet pioneer Vint Cerf; presented her ideas at CERN in Geneva, Switzerland; and was named to Time magazine’s “30 Under 30.” This past Christmas, she used code to illuminate the White House Christmas tree.

As the Google Student Ambassador for Duke, Wenger acts as a liaison between Google and Duke’s campus culture. The sophomore wants to keep fusing computer science with medicine and eventually pursue a joint M.D.-Ph.D.

Josh Miller

Junior | Double-major in electrical engineering and computer science

What do code and corn crops have in common?

More than you’d think, at least for junior Josh Miller, who devised a software system to help farmers increase their yields.

Miller runs processing algorithms on satellite images of local farms to find out information, such as the chlorophyll content of plants in a field. He then translates the data into insight farmers can use—for example, they should apply less water here or more fertilizer there. The data also can help farmers pinpoint pests, fungus, or disease. By helping them find these issues on their farm, the data are saving farmers money, he says.

To formalize his idea into a business, Miller recruited four other computer-science majors and formed Farmshots, a company that offers tailored solutions for farmers and agronomic consultants. His partners are Yu Zhou Lee, Sai Cheemalapati, Ouwen Huang, and Stanley Yuan. Farmshots is being tested by farmers in North Carolina. Miller hopes to expand to other states in the future.

Though Miller describes himself as “just a tech guy from Miami” with no education in agriculture, he worries about an impending crisis of food scarcity. “By 2050, the food production in the entire world is going to have to increase by 70 percent,” he says. “We’re sort of reaching this plateau of how much corn and soybeans we can grow on an acre. The only way to keep pushing that limit of yield per acre is through application of technology.”

Ivonna Dumanyan

Junior | Major in mechanical engineering

A few years ago, Ivonna Dumanyan bought a pair of new running shoes and realized she had a pronation problem.

Pronation is the natural inward roll of the foot while running or walking, and too little or too much of it puts an athlete at risk for injury. As a varsity rower, Dumanyan had top-level coaches and fitness specialists who could help her, but she knew that many athletes didn’t.

Last summer, she designed SmartStrides, a small device that attaches to the heel and monitors pronation. Dumanyan learned and wrote the code herself, both for the device hardware and data-processing software. She also printed a plastic model of her own foot using a 3D printer in Duke’s new Innovation Co-Lab, which offers grants and tools to student developers. Using the foot model, she can simulate pronation and study the “signature” of her steps, which looks similar to activity on a heart monitor. The device can tell a runner whether he or she is pronating, and if so, how much.

“Sixty percent of people pronate without knowing it,” says Dumanyan, a junior. “It’s important to be aware of your body and of your form so you can find the least dangerous and most comfortable form.”

At the moment, Dumanyan’s sensor is about the size of a quarter. She’s trying to make it smaller and wireless, and develop a sweat-proof casing. When it’s ready, she plans to have Duke athletes test it.