Generalization of AI-Based Gestational Age Assessment Using Blind Sweep Ultrasonography

To evaluate the generalizability and performance of AI models for gestational age estimation from blind sweep ultrasonography scans.

Study Design: A prospective, multicenter observational diagnostic study conducted at Northwestern Medicine (Chicago, Illinois) and Jacaranda Health (Nairobi, Kenya).
Participants: Participants included pregnant individuals aged 18 and older with gestational ages between 16 weeks 0 days and 39 weeks 6 days.
Data Collection: Data collection spanned from March 2022 to September 2024, with follow-up tracking ultrasonography findings and birth outcomes.
Model Training: The original neural network model was trained on the Fetal Age Machine Learning Initiative (FAMLI) cohort and fine-tuned using data from the Chicago site.

AI models can generalize to new clinical settings when used by novice operators with different ultrasonography devices.
The study followed STARD reporting guidelines and included a diverse distribution of gestational ages.

Accurate gestational age determination is essential for effective obstetric care and management of pregnancy-related conditions.

The study's cohort size was determined by a resource-defined enrollment period.
The fine-tuning process was conducted exclusively on data from Chicago, which may limit generalizability.

The study aims to validate AI models for gestational age estimation.