AUREVA: An AI-Driven Multimodal Framework for Reliable Eye Disease Screening with Uncertainty Estimation

Abstract

Eye diseases such as Conjunctivites, Uveitis, Cataract and other external eye disorders often remain diagnosed in their early stages due to limited access to ophthalmologists, language barriers and lack of awareness about symptoms. Early detection of eye diseases plays a crucial role in preventing vision impairment and blindness. Recent advancements in Artificial Intelligence have enabled automated eye disease screening using image-based and symptom-based approach. However, most existing systems rely on single-modal inputs and provides deterministic predictions without explaining model confidence, which limits their reliability and user trust in clinical things. Real-world deployment remains challenging due to limited transparency, confidence estimation, and accessibility. Language and interaction barriers remain underexplored, particularly in multilingual and low-resource settings. The proposed approach aims to improve accessibility, reliability, and early detection of eye diseases, making it suitable for deployment in remote healthcare environments.