Diabetic retinopathy is one of the most common side effects of diabetes and the primary cause of blindness (DR). The development of the sickness can be stopped if DR is discovered early. Due to differences in the distribution of medical conditions and poor labour productivity, the best window for diagnosis and treatment was lost, which causes eyesight deterioration. Neural network models can be used to categorise and diagnose DR, enhancing efficiency and reducing costs. Three hybrid model structures were created in this study—Hybrid-a, Hybrid-f, and Hybrid-c—to improve the performance of DR classification models together with an improved loss function. EfficientNetB4, EfficientNetB5, NASNetLarge, Xception, and InceptionResNetV2 CNNs were the main models. These basic models were trained using enhance cross-entropy loss and cross-entropy loss, respectively. The output of the basic models was used to train the hybrid model structures. By using enhanced cross-entropy loss, it is possible to greatly speed up the training of the basic models and enhance their performance when evaluated using various metrics. The recommended hybrid model architectures can also improve the performance of DR classification. The accuracy of DR categorization was improved by adopting hybrid model structures.

Artificial Intelligence, Computer-aided diagnosis, Hybrid Model, Diabetic retinopathy, Convolutional neural network