Diabetes Data Analysis via Gaussian Membership Functions with Deep Neural Networks

Abstract

In this article, we analyzed Pima Indians diabetes data with deep neural network. The data were fuzzified using Gaussian membership function, and also Gaussian function used for normalization. The normalized data and fuzzified data were processed with different deep neural networks. Obtained performance results were compared and performance scores showed that, results obtained with fuzzy data are more effective than the results obtained with normalized data. This new method can be used for the prediction of different medical datasets. Furthermore, it is also possible to benefit from this approach to analyze other type of datasets.