AI Accurately Predicts Prematurity Complications in Newborns from Blood Samples

Premature birth is a leading cause of infant illness and long-term disability. However, doctors still struggle to predict which newborns will develop serious complications affecting the brain, lungs, eyes, or digestive system. Babies born at the same gestational age and weight can follow very different medical paths, making early clinical decision-making difficult. Researchers have now shown that patterns hidden in routine blood samples collected shortly after birth can be used to forecast these medical trajectories, offering a way to anticipate and potentially prevent complications of prematurity.

Researchers from Stanford Medicine (Stanford, CA, USA) applied artificial intelligence (AI) to analyze biochemical data already collected through routine newborn screening programs. Using small blood spots obtained at birth, the AI model examined metabolic markers such as amino acids and fat breakdown products. These measurements were combined with basic clinical information to create a metabolic health index capable of distinguishing different biological forms of prematurity rather than treating all early births as a single condition.

Image: The AI algorithm finds patterns in premature infants’ blood samples that correlated with their health later in infancy (Photo courtesy of 123RF)

The researchers analyzed data from 13,536 premature infants born in California and validated the results using an independent cohort of 3,299 preterm infants from Ontario, Canada. The AI system identified molecular patterns linked to four major prematurity complications, including lung, intestinal, eye, and brain disorders. The findings, published in Science Translational Medicine, showed that a set of six blood-based measurements, combined with clinical factors, predicted individual complications with more than 85 percent accuracy.

The findings suggest that prematurity is not a single condition but a collection of distinct biological pathways that lead to different outcomes. Early risk prediction could help determine which infants need specialized neonatal intensive care and allow physicians to tailor monitoring and treatment strategies from birth. Researchers plan to expand the model by integrating maternal health data, electronic medical records, and additional biological measurements, with the long-term goal of preventing complications before they fully develop.

“The goal is to have a whole new taxonomy of prematurity, so you can see where a child is headed and understand what is causing differences in their health trajectories,” said study co-author David Stevenson, MD, “That will allow us to intervene, prevent, and treat complications.”

Related Links:
Stanford Medicine