AI Helps Students Build Preterm Birth Models in Minutes

A high school student just helped build medical prediction models that usually take expert programmers days to complete, and he did it in minutes with AI assistance.

Victor Tarca, a high school student, teamed up with Reuben Sarwal, a master's student at UC San Francisco, to tackle one of medicine's toughest challenges: predicting preterm birth. Using generative AI chatbots, they analyzed data from over 1,000 pregnant women and created functional computer code almost instantly.

The results shocked researchers. Four of the eight AI tools tested matched or even beat the work of expert teams who had spent months on the same challenge.

This matters because every day in the United States, about 1,000 babies are born prematurely. Preterm birth is the leading cause of newborn death and a major contributor to lifelong disabilities. Yet scientists still don't fully understand what triggers it.

The research team at UCSF and Wayne State University wanted to know if AI could crack complex medical datasets faster than traditional methods. They ran a direct comparison: human experts versus AI-assisted teams, all working on the same pregnancy data.

The junior team with AI help went from starting their research to submitting findings to a scientific journal in just a few months. Previous similar projects took nearly two years to compile and publish.

Marina Sirota, a professor at UCSF who led the study, calls this a game changer. "These AI tools could relieve one of the biggest bottlenecks in data science," she said. "The speed-up couldn't come sooner for patients who need help now."

The AI systems generated analytical code from simple text instructions, similar to how ChatGPT works. Researchers typed in what they needed in plain language, and the AI translated that into working computer programs.

The Ripple Effect

This breakthrough extends far beyond pregnancy research. Medical scientists spend countless hours writing and debugging code instead of actually interpreting their findings. With AI handling the technical heavy lifting, researchers can focus on asking better questions and finding answers faster.

The technology could democratize medical research too. Scientists with limited coding experience won't always need large expert teams or expensive consultants. A clear question and the right AI tool might be enough to launch meaningful studies.

Researchers emphasize that human oversight remains essential. AI can still produce errors, and experts must verify results. But the combination of human creativity and AI speed could accelerate discoveries across all medical fields.

The study appears in Cell Reports Medicine, proving that what once seemed like science fiction is now helping real students make real contributions to science.

For families waiting for answers about preterm birth and thousands of other medical mysteries, help may arrive sooner than anyone expected.