According to the World Health Organization, over 2.5 billion people have vision impairments and 275 million people are blind. Over 80% of these cases were preventable. Many eye diseases are “silent-killers” of vision and current testing methods are designed for corporate hospitals and are often inaccessible for marginalized patients. Thus, diseases go undetected until it is too late.
To solve this global problem, I developed Speculor, combining low-cost portable retinal imagers using smartphones with artificial intelligence to screen diseases. Speculor builds a low-cost teleophthalmology platform serving marginalized communities. A new transfer learning method was invented and tested for the classification of glaucoma achieving the best results reported in literature. Imagers have been tested in a clinic and pick up all clinically relevant features needed for accurate diagnosis for glaucoma. It is hoped that this project will be a small step in our global efforts to eliminate preventable blindness.