Meta Learning Gradient Boosted Neural Network Model Based Diabetes Risk Prediction with Bias Reduction Using OCT Image Attributes

Ensemble learning is defined as technique used to combine feeble techniques to develop a robust technique to maximize the algorithm efficiency. Ensemble boosting technique is combined with neural network to improve the prediction accuracy. The combined technique is optimized with fine-tuned hyper parameters, and it is cross validated to avoid overfitting. Neural network is defined as the technique used to imitate human memory that consists of number of perceptrons to accept input data from user. Input layer consists of number of input nodes based on input data given by user. The weight calculation and parameter calculation are performed in hidden layer. The output layer consists of output nodes to display the required output. This work consists of OCT retinal parameters of Myopic, CSR and normal data of to predict diabetic risk. This work discusses about optimized and cross validated gradient boosted neural network with 98% prediction accuracy.