The Application of Artificial Intelligence in Epiretinal Membrane from Detection to Prognosis Prediction

Epiretinal membrane (ERM) is a prevalent vitreoretinal interface disorder characterized by fibro-cellular proliferation on the inner retinal surface. The prevalence increases markedly with age, and progressive membrane contraction distorts retinal architecture, producing metamorphopsia and irreversible visual impairment thereby significantly impairing patients' quality of life. This highlights the need for more accurate and efficient diagnostic and therapeutic strategies. Artificial intelligence (AI) has emerged as a promising tool to address such challenges. This review summarizes and evaluates recent studies on the applications of AI in ERM management, identifying current limitations and future research directions.

A comprehensive literature search was conducted in PubMed, Web of Science, Embase, and Cochrane Library databases, focusing on studies related to AI and ERM published over the past decade.

The findings indicate that while research on treatment and prognosis prediction remains limited, and new technologies are expected in image processing, current AI models demonstrate substantial potential in the ERM management, especially in detection and diagnosis.

Nonetheless, current evidence is constrained by challenges such as single-center design, limited external validation across devices or ethnicities, insufficient multimodal imaging, and lack of health-economic or workflow integration data. Future multicenter prospective studies, federated learning platforms, publicly annotated imaging dataset and cost-effectiveness analyses are warranted to develop robust, generalizable models that can be seamlessly integrated into clinical workflows, substantially optimizing ERM management and delivering tangible benefits to both patients and clinical practice.