The robust performance of some retrospective validation studies of deep learning (DL) for diabetic retinopathy (DR) screening on color retinal photographs has generated enthusiasm for using artificial intelligence (AI) in healthcare, not only in ophthalmology.

The steps of applying AI in healthcare, including DR screening, can be followed from in silico evaluation, offline retrospective evaluation, small-scale and large-scale prospective online evaluation, and post-market surveillance, comparable to the preclinical and subsequent phases of studies on new drugs.

Many factors, other than its diagnostic performance, should be in consideration for deployment of AI for DR screening. These challenges include workflow issues, such as mydriasis to lower ungradable cases; technical issues, such as integration in electronic health record system, ethical issues, such as data privacy and security, and acceptance of personnel and patients.

The deployment of AI for DR screening should follow the governance model for AI in healthcare which outlines four main components: fairness, transparency, trustworthiness, and accountability.