At the UF/IFAS AI Summit, Assistant Professor Dr. Xiaoya Zhang presented her innovative, AI-powered research that investigated a support program for high-risk families. Her study examined the effectiveness of the Family Check-Up Online Program, a program designed to reduce emotional problems in youth.
The issue of youth mental health has become a prominent concern in recent years. While many prevention programs are designed to tackle mental health in youth, their effects are often minimal. Dr. Zhang’s goal is to determine why these programs don’t make a bigger impact on the communities they design them for.
After seeing the minimal impact of these programs, Dr. Zhang wondered, “Maybe the issue isn’t that these programs don’t work, but that they don’t work equally well for everyone.”
After posing this question, Zhang set off to investigate how prevention programs may affect certain families more than others.
“My study explores that possibility by using machine learning to move beyond the traditional ‘one-size-fits-all’ model of prevention. Instead of only asking whether an intervention works, I asked: Who does it work best for?” Zhang said.
In her study, she used a machine-learning model to evaluate the program’s effectiveness and identify groups of families that benefited most from it.
“I found that the program did reduce emotional problems overall—but the effects varied widely from family to family. The strongest benefits appeared among youth who started with more emotional difficulties, whose parents were highly motivated to improve, and whose health habits were weaker at baseline,” Zhang said.
Rather than being a universal solution for youth with emotional problems, she found that the program was a formidable tool for high-risk families.
How Dr. Zhang Teaches about AI in FYCS Research
Dr. Zhang joined the FYCS department in 2022, bringing her expertise in artificial intelligence to the Youth Development and Family Science doctoral program. She teaches budding researchers about machine learning models and their applications in FYCS research.
“Students work with real data to see how each method functions in practice: they build models, examine outputs, interpret coefficients or feature importance, and critically assess model performance,” Dr. Zhang said.
When teaching applications of AI in FYCS research, Dr. Zhang prioritizes understanding how to interpret models over strict technical details. Her students learn to identify what the models uncover about the communities they study. They also investigate factors that may impact model performance, such as bias, overfitting, and data limitations.
“By combining conceptual foundations, hands-on application, engagement with the scholarly literature, and critical reflection, I prepare students to use AI as a rigorous, transparent, and ethically responsible research tool in the field,” Dr. Zhang Said.
Dr. Zhang’s integration of AI in the classroom reflects her commitment to equipping students with new skills that can enhance their research. Artificial intelligence can be a valuable tool in FYCS research, according to Zhang.
“Importantly, these advances can support precision-based or personalized intervention and prevention programs—tailoring support to the specific needs, strengths, and circumstances of diverse families and communities,” said Zhang.
Machine learning models can help FYCS researchers discover surprising connections they may have missed using traditional research methods. They may also allow researchers to pose new questions and develop new theories about how families and communities interact with their environment. By embracing these powerful tools, FYCS researchers can extract deeper insights into what truly works for families.
3
