AI model improves early detection of placenta accreta
Latest news on WhatsApp أجدد الأخبار على واتسابResearchers have developed an AI based tool designed to improve prediction of placenta accreta spectrum (PAS), a dangerous pregnancy complication that can trigger severe bleeding after delivery and, in extreme cases, lead to hysterectomy or death, and is frequently missed by current screening approaches.
In early testing presented at the Society for Maternal-Fetal Medicine (SMFM) meeting in Las Vegas, the team evaluated the model using ultrasound data from 113 high risk pregnancies and reported that it identified all PAS cases in the sample, with two false positives and no false negatives.
Why it matters
Developers noted that advance diagnosis can be achieved in only a minority of cases, sometimes around 30%, because PAS can be missed on ultrasound. SMFM also highlighted that roughly half of cases are diagnosed during pregnancy, underscoring the gap the tool aims to address.
How the model works
The AI combines ultrasound imaging with clinical risk factors, including prior cesarean deliveries and placenta previa status, which can increase PAS risk.
Participants delivered at Texas Children’s Hospital between 2018 and 2025, with mean gestational age at ultrasound around 31 weeks.
Early performance signals
Reuters reported the model correctly identified 75% of non PAS pregnancies, with an 82% positive predictive rate in the sample and a negative result that matched PAS absence across the tested group.
Researchers said the approach could evolve into a practical screening aid that flags patients for referral to more specialized imaging and delivery planning.
Next step
The team said a prospective, real world study is needed to validate performance before broader clinical use.
