Retinal-image AI predicted preeclampsia before diagnosis in a two-site study

Researchers writing in Nature Biotechnology report that Visionary AI uses retinal blood-vessel patterns to predict preeclampsia before symptoms. Developed using a Columbia cohort of 1,267 pregnancies, the model reached an AUC of 0.81 in a separate 66-person NYU validation group using first-trimester images, compared with 0.73 for a clinical risk calculator. The authors say larger, more diverse studies are needed to establish its screening performance.