AI Bots Predict Preterm Birth from Big Data Analysis

UCSF and Wayne State researchers find AI can rapidly analyze pregnancy data to identify patterns predictive of preterm birth

Published on Feb. 24, 2026

Got story updates? Submit your updates here. ›

Researchers at UC San Francisco and Wayne State University found that generative AI tools could perform orders of magnitude faster than computer science teams in analyzing large amounts of pregnancy data to predict preterm birth. Even a junior research duo of a master's student and a high school student were able to generate working computer code in minutes using AI assistance, a task that would have taken experienced programmers hours or days. The AI's strength came from its ability to write code to analyze health data based on a short but highly specialized prompt. This enabled the junior scientists to quickly run experiments, verify results, and submit their findings to a journal in just a few months.

Why it matters

A faster path from data to discovery could lead to more reliable diagnostic testing for preterm birth - the leading cause of newborn death and a leading cause of long-term motor and cognitive impairment in children. About 1,000 babies are born too soon in the U.S. every day, and scientists still don't know much about what leads to preterm birth. Using AI to quickly analyze large datasets could help unlock new insights and enable researchers to focus more on answering the right biomedical questions.

The details

The researchers instructed eight AI tools to build algorithms to make pregnancy assessments using data from three previous DREAM challenges, which were crowdsourced competitions to find patterns in pregnancy data that could indicate preterm birth and better estimate the stage of pregnancy. The AI chatbots were given natural language prompts to accomplish this, similar to how ChatGPT works. Only 4 of the 8 AI tools produced prediction models that performed as well as the DREAM teams, but sometimes the AI outperformed the human teams. And the entire generative AI project - from inception to submission of a paper - took just six months, much faster than the nearly two years it took to compile and publish the DREAM challenge results.

  • The DREAM challenges took nearly two years to compile and publish the results.
  • The generative AI project took just six months from inception to paper submission.

The players

Marina Sirota

A professor of Pediatrics who is the interim director of the Bakar Computational Health Sciences Institute (BCHSI) at UCSF and the principal investigator of the March of Dimes Prematurity Research Center at UCSF.

Tomiko T. Oskotsky

The co-director of the March of Dimes Preterm Birth Data Repository, associate professor in UCSF BCHSI, and co-senior author of the paper.

Reuben Sarwal

A master's student at UCSF who, with the help of AI, was able to generate working computer code in minutes to predict preterm birth.

Victor Tarca

A high school student who, with the help of AI, was able to generate working computer code in minutes to predict preterm birth.

Adi L. Tarca

The co-senior author and a professor in the Center for Molecular Medicine and Genetics at Wayne State University in Detroit, MI, who had led the two other DREAM challenges.

Got photos? Submit your photos here. ›

What they’re saying

“These AI tools could relieve one of the biggest bottlenecks in data science: building our analysis pipelines. The speed-up couldn't come sooner for patients who need help now.”

— Marina Sirota, Professor of Pediatrics, interim director of the Bakar Computational Health Sciences Institute (BCHSI) at UCSF, and principal investigator of the March of Dimes Prematurity Research Center at UCSF (Cell Reports Medicine)

“This kind of work is only possible with open data sharing, pooling the experiences of many women and the expertise of many researchers.”

— Tomiko T. Oskotsky, Co-director of the March of Dimes Preterm Birth Data Repository, associate professor in UCSF BCHSI, and co-senior author of the paper (Cell Reports Medicine)

“Thanks to generative AI, researchers with a limited background in data science won't always need to form wide collaborations or spend hours debugging code. They can focus on answering the right biomedical questions.”

— Adi L. Tarca, Co-senior author and professor in the Center for Molecular Medicine and Genetics at Wayne State University in Detroit, MI (Cell Reports Medicine)

What’s next

The scientists still need to be on guard for misleading results and step in when the AI fails. The technology is no replacement for human expertise, but its power could enable scientists to quickly parse massive amounts of data - freeing them up to think more deeply about the right biomedical questions.

The takeaway

This study demonstrates how AI can dramatically accelerate the analysis of large health datasets, enabling even junior researchers to quickly generate insights that could lead to better diagnostic testing and care for preterm births - a major public health challenge. However, human oversight and expertise remain crucial to ensure the reliability of the AI-generated findings.