January 2026: Dor & Shomron: Large-scale evaluation of AI models shows no significant advantage over Hadlock method in high-risk pregnancies
A large-scale study, led by Omer Dor, Edmond J. Safra MD-PhD student fellow, Prof. Noam Shomron, Edmond J. Safra member (Medical & Health Sciences), and Dr. Misgav Rotenstreich (Shaare Zedek Medical Center), evaluated AI models, and shows no significant advantage over the traditional Hadlock method in estimating fetal weight in high-risk pregnancies.
A large-scale study led by Omer Dor, Edmond J. Safra MD-PhD student fellow, Prof. Noam Shomron, Edmond J. Safra member (Medical & Health Sciences), and Dr. Misgav Rotenstreich (Shaare Zedek Medical Center) evaluated whether AI-based models outperform the traditional Hadlock method in estimating fetal weight in high-risk pregnancies.
While AI models showed better accuracy on average, they did not offer a meaningful advantage in extreme risk categories (below the 3rd percentile and above the 97th percentile), where obstetric decisions are most critical. In these scenarios, the Hadlock method remained equally reliable, and its clinical familiarity provides a practical benefit.
The study analyzed data from 10,000 pregnancies across three medical centers in Israel (Ichilov and Shaare Zedek) and in Canada (McMaster). The researchers recommend cautious integration of AI as a complementary tool in routine cases and advise maintaining the established Hadlock method for extreme-risk cases until further extensive clinical validation is available.
Read more here.
Related Articles
