Simple, Mobile-based Artificial Intelligence Algorithm in the detection of Diabetic Retinopathy (SMART) study

What is it about?

The SMART study is prospective, observational study that evaluated the performance of an ‘offline’ smart phone based artificial intelligence algorithm for the diagnosis of diabetic retinopathy - Medios AI.

The Medios AI is an artificial intelligence algorithm integrated into the Remidio fundus-on-phone (FOP) camera which runs on the iPhone. Images captured on the FOP are analyzed by the software to provide a binary diagnosis of diabetic retinopathy (DR) or no DR. The technology is novel because the algorithm can run ‘offline’ (i.e. without internet) due to the high-performance capabilities of the iPhone on which it runs. 

Our study compared the performance of the AI against the diagnosis of 5 retina specialists using non-mydriatic retinal images captured from 900 individuals with diabetes with the FOP camera. 
We observed that the sensitivity and specificity of the Medios AI for detecting RDR (referable DR) which is defined as moderate non-proliferative DR (NPDR) or more severe disease, or the presence of diabetic macular edema (DME) was 93% and 92.5%. The sensitivity and specificity of the AI for the detection of all DR was 83.3% and 95.5%. 

Our study thus found that the Medios AI has a high sensitivity to detect RDR and a high specificity to rule out DR.

Featured Image