AI-Driven diabetic retinopathy analysis: A multi agent decision fusion approach

Abstract

Retinopathy caused by diabetes is still a primary factor contributing to vision impairment in the world today, and prompt treatment depends on early detection and precise diagnosis. Despite their effectiveness, traditional machine learning and deep learning-based methods frequently have problems such poor generalization across a variety of patient data, restricted interpretability, and static decision-making. In order to improve autonomy in making decisions, dynamic flexibility, and background comprehension of retinal fundus pictures, this study presents an Agentic-AI-Powered Diabetic Retinopathy Analysis Framework that makes use of clever learning systems based on agents. Adaptive feature learning and real-time analysis using patient-specific changes are made possible by the new integration of DR detection systems that incorporate agentic intelligence principles, autonomy, reactivity, and proactivity. DR detection systems that incorporate, autonomy, reactivity, and proactivity. For reliable classification, the suggested AI system combines a coordinated multi-agent ensemble of transformer-based and convolutional networks by a layer for decision fusion. Categorization accuracy, Interpretability of the model and effectiveness of decision fusion layer is evaluated. classification precision (up to 95.8%), increased model effectiveness using lower computing high above. This study demonstrates the revolutionary potential applications of agentic AI in medical imaging, opening in the door to learn more independent moreover comprehensible clinical decision-making tools.