I-SCREEN: Development of an AI-based infrastructure for community-wide screening and prediction of progression in age-related macular degeneration providing accessible shared care.

OBJECTIVES

This work describes the design and methodological framework of the I-SCREEN project, which aims to develop an artificial intelligence (AI)-based infrastructure utilising optical coherence tomography (OCT) for early detection of AMD and assessment of progression risk.

METHODS

The pan-European project is conducted across clinics and optometry/optician practices in six European countries. I-SCREEN encompasses seven work packages covering community-based AMD identification, clinical follow-up, AI development and project dissemination. Three interconnected clinical studies are carried out by optometry/optician practices (PYRENEES) and ophthalmology clinics (SUDETES and APENNINES).

RESULTS

The PYRENEES study is a prospective, cross-sectional study evaluating the feasibility of detecting subclinical AMD in optometry/optician practices under ophthalmologist supervision via telemedicine. A robust screening network comprising 28 community-based optometry/optician practices and 7 ophthalmology clinics has been established. Patients with suspected non-neovascular AMD are referred to partnered clinics. In the hospital setting, patients with early or intermediate AMD are followed in the longitudinal SUDETES study, while patients with non-foveal geographic atrophy are invited to take part in the APENNINES study. Data obtained inform AI development for community-based AMD detection and monitoring. Predictive modelling will further enable personalised risk assessments.

CONCLUSIONS

I-SCREEN brings together multidisciplinary experts across Europe to establish an AI-driven shared care model for AMD detection and monitoring. By combining high-quality OCT imaging from community practices with longitudinal clinical studies, the initiative provides novel insights into early AMD progression and  establishes a foundation for innovative AI-based detection and prediction throughout the real-world population.