Revolutionizing Maternal Health: Could AI Imaging Spot Pregnancy Health Risks Earlier?

Pioneering Proactive Pregnancy Care: Mount Sinai's AI Vision

Imagine a future where serious pregnancy complications are identified not just early, butproactively, giving expectant parents and healthcare providers invaluable time to prepare. This groundbreaking vision is becoming a reality thanks to the innovative work at Mount Sinai, one of the nation's leading teaching hospitals. They're leveraging advanced AI tools to transform how we assess pregnancy risks, pushing detection much earlier in the care journey.

The focus is clear: empowering better outcomes by spotting potential issues sooner. This includes identifying risks for Placenta Accreta Spectrum (PAS) even before conception, and detecting congenital heart defects (CHD) during routine mid-trimester scans. The central question driving this research is:Could AI imaging spot pregnancy health risks earlier, and if so, what impact could that have on maternal and fetal well-being?

The Critical Need for Early Detection

Placenta accreta spectrum (PAS) is a severe complication where the placenta grows unusually deeply into the uterine wall. This makes delivery incredibly complex, high-risk, and demanding on hospital resources. Similarly, congenital heart defects (CHD) are among the most common birth defects, requiring specialized care. Both PAS and CHD are associated with significant health problems for mothers and babies, substantial costs, and intensive resource requirements for health systems.

By moving risk identification "upstream" – identifying potential issues much earlier – doors open to earlier counseling for families, targeted surveillance strategies, and the ability to plan deliveries in specialized hospitals equipped for high-risk cases. This proactive approach aims to shift from reactive emergency care to carefully planned, optimal interventions.

AI's Transformative Role: Predicting & Visualizing Risks

Mount Sinai specialists recently showcased their progress at the 2026 SMFM Annual Pregnancy Meeting, unveiling an AI-assisted workflow for detecting severe CHD from fetal ultrasound and sophisticated machine learning models designed to predict PAS risk using preconception electronic medical record (EMR) data.

This work isn't isolated; it’s part of a broader strategy to integrate diverse data – clinical, social, and operational – for truly data-informed pregnancy care. Studies on social vulnerability, gun violence exposure, and labor management also contribute to this holistic approach.

Predicting Placenta Accreta Spectrum (PAS) Before Conception

In a compelling case-control study analyzing 118,890 deliveries from 2013 to 2023, Mount Sinai found that PAS occurred in a small but significant 0.23% of cases. Despite its rarity, PAS carried alarmingly high risks for severe maternal morbidity and mortality, underscoring the strategic value of precise preconception risk stratification.

Remarkably, the AI analysis uncovered a previously unrecognized risk factor: having anemia before pregnancy. This discovery stands alongside known risks like:

The implications are profound. Since anemia is a potentially modifiable condition, health systems could leverage this insight to reduce risks. By routing patients into nutritional support programs, offering specialized consultations, or providing preconception counseling, the goal is to prevent emergency deliveries and enable planned care in hospitals with specialized expertise.

To achieve these predictions, Mount Sinai's team trained multiple machine learning models using extensive pre-pregnancy EMR data. This dataset included demographics, obstetric and surgical history, vital signs, lab results, and more. An XGBoost model demonstrated superior predictive power, achieving an area under the ROC curve (AUC) of 0.86, outperforming traditional logistic regression (0.76 AUC). While a Random Forest model offered the highest sensitivity (91% – catching most true cases), logistic regression excelled in specificity (91% – minimizing false alarms). This highlights the crucial trade-offs healthcare organizations must consider when optimizing AI models for different clinical priorities.

Enhancing Fetal Heart Scans with AI-Augmented Imaging

Beyond data analytics, AI is also revolutionizing imaging. At Mount Sinai West, software from BrightHeart has been deployed to bolster fetal ultrasound screening for major CHD across a multicenter dataset. This is where the direct impact ofAI imaging could spot pregnancy health risks earlierin a visual and immediate way.

A study involving 200 second-trimester ultrasounds from 11 medical centers in two countries yielded impressive results:

Seven gynecologists and seven maternal-fetal medicine specialists reviewed each exam both with and without AI, demonstrating its value for both generalists performing initial screenings and subspecialists handling complex cases. This cutting-edge technology is currently undergoing real-world evaluation in a prenatal diagnostic center, actively flagging suspicious findings for severe CHD within standard screening workflows.

Navigating the Future: Governance and Integration

Implementing such advanced technology is not without its challenges. Mount Sinai rightly emphasizes the critical need for rigorous validation on diverse patient populations, careful stewardship of large retrospective datasets, and continuous monitoring for potential biases. This includes scrutinizing how social indices, such as composite social vulnerability scores and neighborhood gun violence exposure, are incorporated into the models.

The institution also stresses the importance of clear clinical sponsorship, with measurable metrics tied directly to morbidity reduction, cost efficiency, and workflow improvements. Furthermore, a deliberate plan is essential to scale these promising single-center pilots into system-wide decision support tools.

A New Era for Maternal-Fetal Health

By synergistically combining EMR-driven preconception risk prediction for PAS with AI-augmented fetal cardiac imaging, Mount Sinai is fundamentally redefining when and how pregnancy risks are identified. The early results are undeniably promising, pointing to tangible gains in diagnostic accuracy, operational efficiency, and the proactive planning of care.

However, realizing this potential fully depends on health systems' commitment to matching impressive technical performance with robust governance frameworks and careful, thoughtful integration into existing clinical workflows. This holistic approach ensures that AI truly serves to enhance human expertise, ultimately shaping a healthier future for generations to come.