With AI Tool, Johns Hopkins Clinician Boosts Diabetic Retinopathy Screening to 95% Among Pediatric Patients
In 2018, the Food and Drug Administration approved IDx-DR as the first autonomous artificial intelligence (AI)-based diagnostic tool for use in any field of medicine. The system streamlines screening for diabetic retinopathy (DR), and its results require no interpretation by a doctor.
That same year, although the technology was approved only for adult patients, pediatric endocrinologist Risa Wolf led research efforts at Johns Hopkins to use IDx-DR in clinics that treat children. As the first pediatrician in the nation to use the device, she has improved access to diabetic retinopathy screenings in children because images can be taken and looked at during the same initial trip to doctor.
Before we implemented it, about 49% of the kids who needed screening actually had it done. We went to 95% after implementation. This AI system has allowed us to improve screening rates for a recommended guideline for kids with diabetes, and it has added convenience and cost savings to families.
Risa M. Wolf, M.D.
She says the system also democratizes diagnosis and care by guarding against racial bias.
“When you train an AI system, you need to make sure that the dataset is adequately diverse, that it’s been trained on the populations you’re using it on. The prospective trial for IDx-DR was very diverse, from all over the United States.
“And as a lesion-based algorithm, the device looks for specific markers of diabetic retinopathy in the back of your eyes, which makes it racially invariant — it doesn’t matter what your race is.” Wolf and her research team are working to determine if this technology can reduce disparities and make diabetic retinopathy screening more equitable.
To refer a patient call 617-447-4220.