AI Model Shows High Accuracy in Detecting Life-Threatening Pregnancy Condition

A new artificial intelligence (AI) model demonstrated accurate detection of placenta accreta spectrum (PAS), a pregnancy condition, according to research presented at the Society for Maternal-Fetal Medicine (SMFM) 2026 Pregnancy Meeting. PAS, which involves the abnormal attachment of the placenta to the uterine wall, is a leading cause of maternal mortality and morbidity. Current screening methods diagnose approximately half of all cases during pregnancy.

Understanding Placenta Accreta Spectrum (PAS)

Placenta accreta spectrum is a life-threatening pregnancy complication characterized by the placenta abnormally attaching to the uterine wall. The condition is frequently associated with previous uterine surgical procedures, such as cesarean delivery. The incidence of PAS is reported to be increasing in the U.S.

Undiagnosed prior to delivery, PAS can lead to massive maternal hemorrhage, multisystem organ failure, and maternal death.

High-risk pregnancies are typically screened using a combination of risk factors and ultrasounds. However, these methods can often result in inconclusive findings or misdiagnosis, highlighting a critical need for more reliable diagnostic tools.

Innovative AI Research Methodology

Researchers from Baylor College of Medicine developed an AI program to directly address the challenges in diagnosing PAS. This program retrospectively reviewed 2D obstetric ultrasound images from a cohort of 113 patients.

All patients involved in the study were identified as being at risk for PAS and delivered at Texas Children's Hospital between 2018 and 2025. The mean gestational age at the time of the maternal ultrasound examinations was 30.89 weeks.

Promising Findings Revealed

Based on the retrospective analysis of the 2D ultrasound images, the AI model demonstrated remarkable diagnostic capability. The AI model detected the presence of all cases of PAS within the study group.

The analysis identified two false positives and reported no false negative findings for placenta accreta.

This outcome suggests a 100% sensitivity for the AI model in this study, representing a significant advancement in diagnostic accuracy.

Future Implications and Publication

Alexandra L. Hammerquist, MD, a maternal-fetal medicine fellow at Baylor College of Medicine, commented on the substantial potential clinical implications of this model for timely and accurate PAS diagnosis. Researchers anticipate that the model's application as a screening tool may contribute significantly to a reduction in PAS-related maternal morbidity and mortality.

The research, titled "AI-based ultrasound screening for early, accurate identification of placenta accreta spectrum" (Oral abstract #39), is scheduled for publication in the February 2026 issue of PREGNANCY, the official journal of SMFM.