Predictive Diagnosis of Ulcero-Necrotizing Enterocolitis in Premature Babies Using Artificial Intelligence Analysis of Early Fecal Microbiota

What this Trial Is About

Prematurity affects about 7% of births in France, and necrotizing enterocolitis (NEC) is a serious digestive complication with a mortality rate between 15 and 40%. NEC usually occurs suddenly between 10 and 20 days of life and requires urgent, intensive treatment. Preventing NEC is important, but current measures like breastfeeding and probiotics cannot be personalized because predicting NEC risk remains difficult. This study aims to develop a predictive diagnostic test using artificial intelligence to identify premature babies at higher risk of NEC by analyzing early changes in their digestive microbiota. The study involves collecting stool samples (excluding meconium) from premature infants up to 21 days old. Two stool samples per infant will be analyzed using metagenomic sequencing combined with an AI deep neural network trained on international data. If the first two stool analyses disagree, a third sample will be tested to classify the infant's NEC risk. The diagnosis of NEC will be made by the treating clinician using the Bell classification. This approach is being tested in a larger group of premature infants in the AURA region. Participants will be followed until they return home or transfer to another center. Parents will be contacted by phone when the infant is 3 months old to check for any late NEC cases. Researchers will measure how well the AI test predicts NEC before day 21, characterize the microbiota, and study its relationship with prematurity complications. The study will include about 1000 children, with 200 undergoing the predictive test to assess its accuracy and usefulness in clinical practice.

Predictive Diagnosis of Ulcero-Necrotizing EnteroColitis in Premature Babies Using an Artificial Intelligence Approach Based on Early Analysis of the Fecal Microbiota