In pregnancies where the mother has pre-existing diabetes, female fetuses face higher risks than males—the opposite of what doctors typically expect, according to a new U study that analyzed nearly 10,000 pregnancies using artificial intelligence.
The finding is just one example of how AI could revolutionize prenatal care, says Nathan Blue MS’22, an obstetrician-gynecologist and assistant professor in the Spencer Fox Eccles School of Medicine. The study, published in BMC Pregnancy and Childbirth, found dramatic variations in risk levels among pregnancies that are currently treated identically under clinical guidelines.
Take fetuses in the bottom 10 percent for weight. Current guidelines recommend intensive monitoring for all such pregnancies. But the AI analysis revealed that risk levels in this group varied dramatically—from no riskier than average to nearly 10 times the average risk—based on combinations of factors like fetal sex, pre-existing diabetes, and presence of anomalies like heart defects.
“AI models can essentially estimate a risk that is specific to a given person’s context,” Blue explains, “and they can do it transparently and reproducibly, which is what our brains can’t do.”
The researchers used “explainable AI,” which shows exactly which factors contributed to each assessment. Unlike “closed box” AI systems, this transparency allows doctors to understand the model’s reasoning. While it still needs real-world testing, Blue believes AI-guided risk assessment could transform prenatal care by helping doctors make more personalized treatment decisions.
The research was supported by the U’s One-U Responsible Artificial Intelligence Initiative, a $100 million program launched in 2023 to harness AI for societal benefit.
Comments
Comments are moderated, so there may be a slight delay. Those that are off-topic or deemed inappropriate may not be posted. Your email address will not be published. Required fields are marked with an asterisk (*).