Diabetes Prediction Recommender System based on Artificial Neural Networks and Sine-Cosine Optimization Algorithm

One of the most common diseases in the world is diabetes for which no certain cure has been found yet; the only promising way for these patients to survive is to take care. Fasting blood sugar (FBS) is one of the most important indicators of diabetes. But its test is not feasible for the public and requires preparations before implementation. In this study, the prediction of fasting blood sugar (FBS) is considered as a strategy for predicting diabetes for the first time. This study presents a model for prediction of FBS from other factors in blood test of the people. The proposed model, best feature is selected using sine-cosine optimization algorithm; in the second phase, uses neural network (NN) for prediction. In fact, the idea behind this study is to improve sine-cosine algorithm in selecting the features of dataset derived from diabetic patients of Isfahan city which has not been conducted so far. The prediction results of three different neural networks (with training and supervision, without supervision and semi-supervision) showed that multilayer perceptron NN managed to predict FBS with error less than 0.0017.