Office Hours with… Dennis Shung

Dennis Shung came to Yale for residency and stuck around for a Ph.D. Now an assistant professor, he studies how machine learning can enhance medical practice.
Dennis Shung

Dennis Shung (Photo by Andrew Hurley)

After completing a medical residency at Yale, Dennis Shung stuck around for a fellowship. Then he stayed to earn a master’s degree, and then a Ph.D. Last summer he became an assistant professor at Yale School of Medicine.

Shung says it’s the quality of mentorship and sense of community that keeps him at Yale, where he is now exploring how artificial intelligence (AI) is used in medicine and how it might better suit the needs of patients and physicians.

We recently caught up with Shung for the latest edition of Office Hours, a Q&A series that introduces new Yale faculty members to the broader university community.

Title Assistant Professor of Medicine (Digestive Diseases), Director of Digital Health (Digestive Diseases)
Research interest Using machine learning to enhance clinical decision-making in gastrointestinal diseases
Prior institution Yale New Haven Hospital (previously Baylor College of Medicine)
Started at Yale July 1, 2022

How would you describe your research?

Dennis Shung: Initially, I wanted to create better machine learning tools that could help physicians determine patient risk, particularly for acute gastrointestinal bleeding, which is the most common cause of gastrointestinal-related hospitalizations in the United States. I found there were already a few simple tools that were validated and recommended by international guidelines, but that their performance was suboptimal and nobody was using them. So I wanted to make a better tool — and figure out the barriers to adoption.

As I got more into the field, I also realized that these tools learn very differently than humans do, and there was the potential to extend or enhance a clinician’s understanding of risk through human-algorithmic interaction.

One big issue is trust. Another is the way machine learning is viewed as just a tool; patients worry about AI being wrong and providers may be asking different questions than what the AI is designed to answer. Medicine is a team sport, and AI can be thought of as more of a team member rather than a tool. The research I’m pursuing now is about reframing the use of AI in medicine to better match how we practice medicine now and seeing what effect that has on trust.

Why did you decide to pursue a Ph.D.?

Shung: When I finished my research-track fellowship, I had learned how to code, process data, and train and validate a machine learning model. However, I felt like in the world of data science I was still in the shallows, and out there was the deep blue sea. There was so much more that I couldn’t access because I didn’t have enough training. At that point, my mentors encouraged me to consider a Ph.D., which is a well-trodden process that helps you refine your questions and navigate the way to asking deeper questions.

What keeps you here at Yale?

Shung: The generosity of my mentors, who have believed in me, given me time, and opened doors for me. The quality and depth of mentorship is unparalleled. Also, the alignment across the School of Medicine and the health system — particularly in data science informatics — showed me there was a path forward here for my research.

On a personal level, a sense of belonging and community has really blossomed over the past few years. We’re part of a faith community that has become our family, a home away from home.

How does your family like to spend time together?

Shung: The first five years with kids is just survival, so now we’re thinking about what our family culture is and what our traditions will be. We want our kids to recognize needs in their community and how they can contribute, so recently, we made a dinner together for a warming center sponsored by a New Haven nonprofit.

We like to cook a lot (I am a gastroenterologist, after all) and my six-year-old son is the resident food critic; my wife and I have a friendly competition to see what ratings we get.

Share this with Facebook Share this with Twitter Share this with LinkedIn Share this with Email Print this