Using machine learning to identify pediatric ophthalmologists

Abstract

This cross-sectional study used data from the American Academy of Ophthalmology IRIS Registry (Intelligent Research in Sight) and machine learning algorithms to identify pediatric ophthalmologists based on physician coding patterns. A random forest model achieved an area under the receiver operating characteristic curve of 0.98, sensitivity of 0.98, and specificity of 0.88 when classifying pediatric eye specialists in the test validation cohort. Algorithm-based approaches to identify pediatric ophthalmologists using procedure codes may offer new avenues to determine the scope, scale, and trajectory of pediatric eye care delivery.