The accurate determination of gestational age (GA) is crucial for effective prenatal care, particularly in low- and middle-income countries (LMICs) where resources are often limited, and access to skilled sonographers and high-specification ultrasound machines is scarce. A recent prospective, two-country diagnostic accuracy study with results published in the Journal of the American Medical Association has demonstrated the potential of an AI-enabled ultrasonography tool to bridge this gap.
The study aimed to evaluate the accuracy of an AI-enabled ultrasonography tool operated by novice clinicians, who received just one day of training, in estimating GA. The tool's performance was compared with that of credentialed sonographers using standard fetal biometry on high-specification machines.
In the primary evaluation window (14-27 weeks), the AI tool provided GA estimates that were statistically equivalent to those of credentialed sonographers. This period is particularly significant because it encompasses the time frame when the majority of first antenatal visits occur in LMICs. In the secondary window (28-36 weeks), the tool maintained its accuracy, effectively distinguishing meaningful variations in fetal size that arise due to pathological or constitutional factors.
However, the tool's performance was inconclusive after term (≥37 weeks of gestation), leading researchers to recommend against using it for GA estimation at term. The study also revealed that the tool's accuracy was consistent across different geographies, such as Zambia and North Carolina, and among participants with high body mass index (BMI), a group for whom ultrasonography can be more challenging.
As noted in Eurekalert by the corresponding author, Dr. Jeffrey S. A. Stringer, “Our study demonstrates that an AI-enabled, portable ultrasound device can estimate gestational age as accurately as an expert sonographer using an expensive, high-specification machine. This high degree of accuracy was obtained even though the users of the device had no formal training in sonography."
The AI-enabled ultrasonography tool enables novice users to achieve GA estimates comparable to those of expert sonographers, potentially expanding access to essential prenatal care. This capability aligns with the World Health Organization's goal of universal GA estimation for all pregnant individuals, addressing a critical gap in LMICs where access to skilled sonographers and high-specification machines is limited.
By bringing advanced diagnostic capacity to settings with high clinical disease burdens, this research represents a major step forward in improving maternal and fetal health outcomes.
Sources: Eurekalert, Journal of the American Medical Association
