We hear often about the distressingly high incidence of Type 2 diabetes in the U.S. In a 2017 report, the CDC estimated 9.4% of the population or about 30.3 million Americans have diabetes. The World Health Organization’s 2016 Global Report on Diabetes estimated 422 million people lived with diabetes in 2014. Diabetic retinopathy (DR) or diabetic blindness is one of many secondary complications that threaten patients. Regular retinal screenings are usually part of patient care. The screenings traditionally require an office visit to sit in front of an expensive fundus camera that captures images of the back of the eye, so that a physician can look for abnormalities.

The Madras Diabetes Research Foundation, an affiliate of the University of Madras Medical College, recently published results of a study using a smartphone device with artificial intelligence (AI) software to detect diabetic retinopathy in the journal Eye. The researchers used the Remidio ‘Fundus on Phone’ (FOP) smartphone device with Eyenuk’s EyeArt AI DR screening software. While larger than most smartphone-enabled devices, the FOP is small enough for clinicians to carry to outreach clinics or for home visits. The Fundus on Phone, which could easily be mistaken for a surveyor’s transit, holds a smartphone securely in front of its viewfinder. In the study, retina images of 296 Type 2 patients were evaluated both by ophthalmologists and by the AI software. The ophthalmologists detected DR in 191 patients and of that group, found sight-threatening diabetic retinopathy (STDR) in 112 patients. The automated EyeArt software found DR in 203 patients and STDR in 146 patients. With statistically significant success in detecting both the presence and specificity of diabetic retinopathy, the researchers declared the FOP and AI software combination a highly sensitive DR and STDR tool, although the possibility of too many false positives cannot be ignored.

The Madras Diabetes Research Foundation isn’t done testing the smartphone/AI DR solution. Additional studies and larger sample sizes are necessary. But on the basis of research and experiments to date, it appears the smartphone solution may be an acceptable tool for first level retinal screening for diabetes patients. After initial screenings, patients with detected DR would be referred to ophthalmologists where potential false positive readings could be eliminated.