Zadie Moon is working with professor Akram Bayat on research into using AI to train therapists working with patients who have chronic diseases.
OAKLAND, Calif. — Zadie Moon didn’t use a computer, or any kind of digital device, until she was 16. The oldest of eight children, Moon grew up in a family where resources were limited and technology wasn’t a priority.
Fast forward seven years, Moon is working on her master’s degree in computer science at Northeastern University in Oakland and is a product manager at a company focused on enhancing digital accessibility.
Moon, who earned a bachelor’s degree in public health in 2022, is pursuing her master’s through Northeastern’s Align program — designed for students without a computing background.
Align starts with two semesters of foundational coursework and includes four to eight months of full-time work experience, preparing graduates for careers in STEM.
Moon saw the program as a way to merge her two passions: technology and public health.
“In every hackathon I’ve done, I’ve been the idea creator and the delegator and the person that has the overall vision for the design,” she says. “And when people are confused and they’re kind of frustrated and they don’t know how to talk it out, I’m able to get down on their level and understand how to communicate.”
On track to complete the Align program this year, Moon is working with Akram Bayat, an assistant professor at Northeastern’s Khoury College of Computer Sciences, on research into using AI to train therapists working with patients who have chronic diseases.
There is a shortage of therapists to work with patients going through treatment for chronic diseases like diabetes, Bayat says. Patients with chronic illnesses often require emotional support, as the ongoing challenges of managing their condition and coping with its impact on daily life can frequently lead to feelings of anxiety and depression.
One of the project’s goals, she says, is to develop an app that can be used to help therapists understand how to best support these patients.
“We can digitize this training process so we can help more therapists to be trained in this process and help more communities,” Bayat says.
The project also aims to evaluate training tools that already exist, including virtual patient personas. These AI-generated patients that therapists interact with before they meet real patients might not accurately represent real people, Bayat says.
Working with Moon and Boston-based undergraduate student researcher Vatsal Mehta, Bayat hopes to develop a robust method of evaluating artificial intelligence personas for bias or inaccuracy and create scenarios to simulate the therapy process.
The project overlaps with research that Bayat and Moon worked on in the fall. Last semester they reviewed current literature to understand ways to mitigate for bias in machine learning models used to diagnose and manage chronic diseases.
When models are trained using data sets that don’t accurately represent overall racial, socioeconomic and geographic characteristics, they may perpetuate bias and lead to inaccurate diagnoses, Bayat says. Their research investigated the use of open source frameworks like IBM’s AI Fairness 360 and Microsoft’s Fairlearn, which allow engineers to better evaluate the fairness of the AI systems they are building.
As Bayat’s research apprentice, Moon is helping to explore how large language models can be used for therapist training and how to evaluate these tools for both accuracy and bias.“By addressing bias in machine learning algorithms for chronic disease detection, our research contributes to the larger conversation on ensuring AI serves all populations fairly, particularly those most at risk of being left behind,” Bayat says.
Moon chose to study computer science specifically so she could work on these kinds of issues. After receiving her bachelor’s degree in public health, Moon worked as a user experience researcher for One Degree, a web-based nonprofit in San Francisco that links low-income people to services they need. That experience helped make it clear that she wanted to merge her two interests.
“I wanted to explore computer science because I realized there was a gap in understanding, from the engineering perspective, of how to provide information in a simple and accessible format,” she said. Most computer science programs, she says, don’t address this gap or attempt to identify populations that might find technology inaccessible.
For non-digital natives who need to make an appointment, check their medical records or research a health issue, but who get overwhelmed by web-based health portals, the stakes are high. Just closing out a page because it’s poorly designed could mean not getting critical information.
“The majority of engineering education does not involve a foundational understanding of feeling overwhelmed by information,” she says, “and then how that actually diverts people from receiving information or continuing to be curious.”
While working on her master’s, Moon started a Google Developer Group on the Oakland campus where undergraduates and graduate students can host events on campus and attend events at Google’s headquarters in Mountain View to meet industry professionals. More than 100 students participated last semester.
“This experience hopefully helps students step outside the box and identify potential opportunities that pique their interest that they wouldn’t otherwise know about,” she says.