An Ensemble Classification Techniques Based On ‘Ml’ Model For Automatic Diabetic Retinopathy Detection

Abstract

Diabetic Retinopathy causes the blindness, is commonly known as vision destruction which cannot cure with surgery, through the spectacles or medication completely. A retinal tissue of eye is damaged, in the condition of blindness. Through the automatic computational mechanism, diabetic retinopathy severity is detected in present days hence the early stage of retina damage can be recognized before attacking the blindness. Health records are having uncovering patterns which provides a number of data regarding various diseases diagnosis to medical practitioners by using the important role of machine learning. Sometimes the disease detection accuracy is reduces because of health records sensitivity. An ensemble based machine learning (ML) model is proposed in the paper and diabetic retinopathy dataset is uses the machine learning algorithms as Decision Tree Classifier, Random Forest, K-Nearest Neighbour, Adaboost Classifier, J48graft classifier and Logistic Regression for experiment. The ensemble based machine learning model gives the best performance in terms of sensitivity and accuracy than the individual machine learning algorithms. The proposed work of Diabetic Retinopathy (DR) uses the Image dataset collection which is predefined from Kaggle. The accuracy of proposed work is 96.34% and having the improvement in the performance when compared with previous works. Ophthalmologists society is very much helped by this proposed ensemble based retinopathy detection.