Predicting Eye Diseases
Ryan discusses the journey of using image data to predict eye diseases like diabetic retinopathy with remarkable accuracy. Initially intended as a diagnostic approach, the inclusion of patient metadata revealed surprising predictive capabilities, leading to the realization that high accuracy in predicting gender was not a bug, but a genuine insight from the model. This discovery challenges conventional expectations about predictive modeling in healthcare.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Predicting Cardiovascular Risk Factors from Eye Images with Ryan Poplin - #122
Related Questions