Exploring the Parametric Impact on a Deep Learning Model and proposal of a 2-Branch CNN for Diabetic Retinopathy Classification with Case Study in IoT-Blockchain based Smart Healthcare System

Smart healthcare has changed the way how patient interacts with specialists for treatment. However, security and support various diseases are still concern such smart automated systems. One of critical namely Diabetic Retinopathy (DR), is a major person prolonged diabetes may lead to complete blindness irrespective age groups. Moreover, in recent years blockchain gained popularity providing secure communication between sender receiver. Hence, this work focus on designing blockchain-based system early detection diabetic retinopathy. DR impose complexities requires expert diagnosis, which not available everywhere. proposed model contains Computer-Aided Diagnosis (CAD) assistance symptoms disease. The CAD assist ophthalmologists DR, intensive research developing an efficient accurate that can operate without human interaction. Convolutional Neural Network (CNN) one learning models AI shown its potential computer vision applications. performance CNN depends several factors like pooling approach, activation function, rate architecture. This study provides empirical analysis these design best DR. be used develop IoT based devices detect patients. also explains importance technology development values parameters type hyperparameters chosen from 2-branch model, validated using Kaggle fundus image set. Analysis their gives outstanding model.