Across the globe, more than 460 million people live with diabetes and millions risk losing their sight to diabetic retinopathy (DR). Traditional diagnosis depends on manual image analysis by specialists, a process that is slow, costly, and often inaccessible to rural communities. Recognizing this critical gap, my doctoral research at Koneru Lakshmaiah Education Foundation set out to build a new AI architecture that could bring precision, speed, and privacy to diabetic retinopathy screening.
Building Intelligent Vision: The AI Framework
The research introduces four interconnected frameworks DRG-Net, CEFNet, SDRG-Net, and ICNN-IRDO that work together to automate the detection and grading of retinal images. Each framework tackles a unique challenge: class imbalance in datasets, weak feature representation, limited generalization, and data security in real-time IoMT environments.
DRG-Net uses graph-based feature extraction and XGBoost to improve classification accuracy.
CEFNet refines feature selection through Iterative Random Forest and enhanced balancing techniques.
SDRG-Net adds a layer of privacy with Multi-Level Color Transformation (MLCT) encryption and a disease-specific graph correlation network for precise grading.
ICNN-IRDO fuses deep learning with optimization algorithms for robust diagnosis across multiple datasets.
Results That Speak for Themselves
Tested on the EyePACS and Messidor datasets, the models achieved an unprecedented 99.7 percent accuracy and 100 percent sensitivity and specificity in grading different stages of DR. Each innovation was validated through peer-reviewed publications in Elsevier, Informatica, and IEEE, and protected by design and utility patents in India.
From Lab to Life: AI That Heals
The impact goes beyond research papers. These frameworks are being adapted into real-time clinical systems and IoMT kiosks for rural India solar-powered units that can capture eye images, analyze them locally using AI, and connect to ophthalmologists through secure cloud networks. This fusion of AI and accessibility brings preventive eye care closer to the people who need it most.
A New Era of Trustworthy Healthcare AI
What sets this research apart is its commitment to security and scalability. By combining graph-based learning, ensemble methods, and privacy-preserving encryption, architecture offers a pathway to AI solutions that clinicians can trust, and patients can embrace.
About Dr. Venkata Kotam Raju Poranki (PVK)
Dr. Poranki is a technology leader and AI researcher with over two decades of industry experience in software engineering and product innovation. Currently serving as Program Manager at Zions Bank in Utah USA, he bridges his academic work with real-world applications in secure healthcare and financial systems. His vision is to make AI-powered diagnostics affordable, ethical, and universally accessible.
