Diabetic disease prediction using deep learning paradigm

Diabetic patients are dominating large population of today’s world. Young citizens as well as senior citizens are suffered from this chronic disease. After technology revolution and internet expansion, more flexibility is granted to fields alike medical applications in terms of disease diagnosis and treatment. Individual developers and companies are now granted accesses to data where intelligent systems can be developed. This paper is detailing the process of neural network based diabetic diseases diagnosis depending on historical diabetic data analysis. The problem of machine learning reliability in health care services was realized in this project and was addressed. Ordinary feed forward neural network is used as corner stone for classifying the diabetic data. The proposed modification using artificial bee colony (ABC) for neural network optimization has yielded best prediction result of 95.2% accuracy. Results are compared by using other algorithms such as principle component analysis (PCA), k-nearest neighbour (KNN), support vector machine (SVM) and Bagging Classification.