How Undergraduate Research Helped One Student Turn Curiosity into a Career in Healthcare Technology

Modern biomedical research is producing data at an unprecedented scale, but turning that data into meaningful insight remains a challenge - especially when researchers are working across thousands of samples. At Stony Brook University, Jachao Lee has been working on ways to speed up that process, particularly for studying Alzheimer's disease, by using artificial intelligence to analyze complex brain tissue images in seconds instead of hours.

Lee, a student in the Department of Computer Science's accelerated BS/MS program, didn't start with a clear goal. "I didn't have a particular interest in AI for healthcare, or even a clear direction in computer science more broadly," he said. "I think that's a fairly common situation."

Like many students, Lee knew what he didn't want-front-end programming-but not much beyond that. He turned to research to figure it out.

"I think the only way you can really learn is by actually experiencing the specific fields," Lee said. "Research is more accessible than internships, provides working experience, and is necessary for a PhD, which I was considering at the time."

That decision led him to the lab of Chao Chen, an associate professor of biomedical informatics at Stony Brook University and affiliated faculty in computer science, where he began working at the intersection of artificial intelligence and medicine.

With guidance from doctoral students and faculty, Lee quickly picked up deep learning methods and began applying them to digital pathology. Chen pointed to Lee's approach to the work as a defining factor in his project.

"Jachao has been impressive since the start. He picked up state-of-the-art deep learning methods remarkably fast under the guidance of our senior PhD students, and it's been impressive to see how easily he applies these tools to our data. What I value most, however, is Jachao's independence. He's naturally curious and often comes to meetings having already explored new ideas or gone a step beyond what we discussed," Chen said.

Lee's work focuses on a central challenge in Alzheimer's research: understanding how the disease affects different regions of the brain. That requires analyzing high-resolution images of brain tissue-a slow, manual process that becomes difficult to scale.

He developed a deep learning pipeline that can automatically identify and segment 12 distinct anatomical regions within the hippocampus, a part of the brain closely associated with memory and one of the first areas affected by Alzheimer's disease. The model makes it possible to map those regions quickly and compare healthy and diseased tissue across large numbers of samples.

The work was published in the SPIE Medical Imaging conference and is now supporting a larger effort to analyze thousands of brain samples.

"By letting the AI handle the heavy lifting, we can compare healthy and diseased brains at an unprecedented scale, measuring exactly how the disease invades different areas of the brain," Chen said.

Lee is now building on that work with a second project. In addition to distinguishing between healthy and diseased tissue, Lee is investigating which brain regions are most important to help the model in making those distinctions. The goal is to identify subtle patterns picked up by deep learning, which might not be visible through traditional analysis and could point to earlier detection or new treatment approaches.

Chen also highlighted Lee's persistence as he took on more complex work.

"His first venture into digital pathology didn't go as planned; despite his hard work, the results weren't what we hoped for. But instead of being discouraged, Jachao treated it as a learning period, quietly building the skills and knowledge he would eventually need. When we launched this new brain histology project, he was ready," Chen said.

Over time, Lee also developed skills beyond the technical side of the work, including how to communicate complex ideas and organize research into publishable results.

"Research is particularly valuable because you have to do most of the work independently. It forces independence and learning how to solve open-ended problems," Lee said. "In meetings, you get regular feedback, and at conferences, you have to present your work to people who know little about your project. Being able to talk about your work is a skill that is beneficial regardless of what you want to do."

This spring, Lee will graduate with his master's degree and begin his career as a software developer at Epic Systems, one of the nation's leading healthcare technology companies. While his role may span a range of software development areas, the experience he gained in research-working with complex data, collaborating across disciplines, and tackling real-world problems-will continue to shape how he approaches his work.

"Jachao's path offers a useful perspective on how computer science and AI education should evolve in the era of powerful generative AI. As these tools increasingly automate aspects of problem-solving, the more critical skill for professionals is the ability to formulate meaningful problems. Accordingly, we should train students to be deeply embedded in specific domains and to understand domain-specific challenges, rather than preparing them as generic programmers or data scientists." Chen said.