The UC San Francisco and Wayne State University research team was able to examine complicated pregnancy data in a matter of minutes using generative AI. In a health research test, AI tools processed large medical datasets to predict preterm births, sometimes beating human teams.
The study shows how generative AI could transform biomedical data science.
Preterm birth remains the leading cause of newborn death, with around 1,000 premature births each day in the United States. To better understand risk factors, researchers compiled microbiome data from about 1,200 pregnant women across nine studies.
University of California San Francisco Professor of Paediatrics Marina Sirota said AI could ease one of the biggest bottlenecks in health research. “These AI tools could relieve one of the biggest bottlenecks in data science: building our analysis pipelines,” she said.
In the experiment, eight AI chatbots were asked to generate analytical code using the same datasets previously analysed during a global DREAM challenge. Only four produced usable models, but some matched or exceeded the performance of human teams. The entire AI-driven project took six months, compared with nearly two years to consolidate earlier results.
With AI support, even a small team made up of a master’s student and a high school student built working prediction models in minutes. Tasks that normally require experienced programmers for days were completed rapidly through detailed prompts.
Wayne State University professor Adi L. Tarca said generative AI allows researchers to focus more on scientific questions rather than coding.