SURF Project Bridges Data Divides in Health Care

When Ziyong Ma began his Summer Undergraduate Research Fellowship(opens in new window) (SURF) at Carnegie Mellon University, he wasn’t just writing code. He was solving a problem that has long challenged the health care industry: how to make complex medical data more accessible to the people who need it most.

A senior double majoring in mathematics and computer science, Ma worked in the Data Interaction Group(opens in new window) (DIG), a lab within CMU’s Human-Computer Interaction Institute led by Adam Perer(opens in new window) and Dominik Moritz(opens in new window). Under the mentorship of Ph.D. student Venkatesh Sivaraman, Ma spent the summer developing TempoQL, a tool designed to help clinicians query medical databases using natural language, without needing programming expertise.

The idea for TempoQL grew out of earlier research in Ma’s experience, where he helped build interfaces that allowed nonexperts to train machine-learning models. 

“Clinicians often know what they want to predict, such as the likelihood of a disease, but they’re not computer scientists,” Ma explained. “We wanted to create something that bridges that gap.”

But as Ma dug deeper, he discovered that the real challenge wasn’t just building models; it was accessing the data itself. Medical institutions often use different standards and formats, making it difficult to share or combine datasets. 

“We needed a way to unify the data access process,” he said. “That’s where TempoQL came in.”

The tool Ma helped design allows users to type queries in plain English, for example, “show me hourly heart rate data,” and uses artificial intelligence to translate those requests into database-ready code. The system then returns results in both raw and visual formats, helping clinicians interpret the data without needing technical training. 

“It’s about reducing the learning curve,” Ma said. “We want people to focus on their domain expertise, not the code.”

One of the biggest hurdles came during testing at University of Pittsburgh Medical Center (UPMC), where the team discovered their interface wasn’t compatible with the hospital’s environment. Ma quickly pivoted, adapting the tool to run without the user-friendly visual interface by using step-by-step Python script instruction. 

“It wasn’t what we planned, but it worked,” he said. “And it taught me how important flexibility is in research.”

The project has since been submitted to a research conference, and Ma said he is hopeful about its future. But for him, the most rewarding part wasn’t the technical achievement, it was the sense of purpose. 

“I took machine learning as a sophomore and thought, this is cool. But when I saw how it could help real people, that’s when it became meaningful,” he said.

Reflecting on his time at CMU, Ma expressed deep appreciation for the opportunities he’s had to work with brilliant minds and meaningful problems. 

“If I weren’t here, I wouldn’t have had the chance to be part of something like this,” he said.

Asked what advice he’d give to other students considering SURF, Ma didn’t hesitate. 

“Find something you truly care about,” he said. “You’ll be spending your whole summer on it, so make it count.”