Nathan CongdonNicolas JaccardDr. Nicolas JaccardGabriella Lanouette

Project Orbis International, Inc

Unknown

60DC-607F-8DA2-827C-36FF-D86A

This project aims at developing and validating an open-access application for AI-assisted screening of Diabetic Retinopathy in low- to middle-income countries based on the screening guidelines introduced by the UK National Screening Committee. The vast majority of adults with Diabetes live in low- and middle-income countries (LMIC). One in three people with Diabetes will be affected by Diabetic Retinopathy (DR), and one in ten will be impacted by sight loss as a result. DR impacts active, working-age adults, and as thus can be economically devastative for both affected individuals and society as a whole. While the UK is leading the way in terms of DR screening, most LMICs do not have screening programmes in place, due in part a severe shortage of specialists. Methods based on recent advances in Artificial Intelligence can reliably detect sight-threatening DR with levels of performance comparable to Human experts. However, most commercially available solutions are based on the International Clinical Diabetic Retinopathy (ICDR) severity scale, which is incompatible with UK National Screening Committee (NSC) guidelines. As a result, those AI solutions cannot be used in LIMICs that adopted NSC guidelines for their screening programmes or as standard of care for DR. Moreover, the cost associated with the procurement of commercial AI solutions can be prohibitive; communities that have the most to gain from the AI revolution are the least likely to benefit from it. In this two-year project, we set out to develop an application leveraging recent advances in AI to facilitate DR screening in LMICs in a way that is compatible with UK NHSC Guidelines. This will be made possible by the vast amount of data made available by INSIGHT. The application will be integrated with our existing Cybersight AI platform, and will be made freely available to users in low- to middle-income countries.

Cybersight AI is currently accessible free-of-charge to healthcare professionals in 140+ LMICs. While it is being adopted as clinical decision support for Diabetic Retinopathy (DR) grading, incompatibility with UK NSC precludes implementation in many countries and regions. This also holds true for the majority of commercially available solutions. If the project is successful, Cybersight AI will support automated DR grading based on NSC guidelines, and will thus be used as part of screening programmes where Orbis operates. As a result, it is expected that the public will benefit from the work carried out as part of this project. In addition, the development process and results will be documented and published, establishing a baseline for automated DR grading based on UK NSC guidelines.

18/11/2022

N/A

- Only population-level statistics will be presented for patient demographics - If Fundus photographs are shown to illustrate various results, all efforts will be made to ensure that no metadata is associated with the actual image file, and that no patient information was printed within the photograph