Delivery Date AI: Image-Based Prediction of Delivery Date from Routine Obstetric Ultrasound

About This Project

Summary

Delivery Date AI (DDAI) is software that predicts a pregnancy's delivery date from routine obstetric ultrasound. Our strategy: build direct delivery-date prediction, validate it on an external, independent dataset, publish peer-reviewed evidence before broad deployment, and integrate it into existing ultrasound workflows without new hardware, so it can reach community and rural settings.

Goals & Objectives

The project set out to give care teams a reliable delivery-date prediction for pregnancies without a dependable due date; to generate rigorous, independent clinical evidence; to evaluate performance on held-out patient data; and to deliver a solution that fits existing ultrasound workflows.

Results & Impact

In the peer-reviewed PAIR Study (5,714 patients), DDAI (V4) reached R-squared 0.92 for all births (0.95 term) with a mean absolute error of 12.90 days (10.76 term). The FDA granted De Novo DEN250007 (Feb 11, 2026), creating a new Class II device type for image-based delivery-date prediction.