Empowered Artificial Intelligence Approach using Intuitionistic Fuzzy based Deep Neural Network for Chronic Diseases Prediction

Managing the increasing prevalence of chronic illnesses is a major worldwide health concern. It's important to understand the medical problem, its stages, types, and available treatments. The most important component in the early diagnosis of chronic diseases might enhance the diagnostic procedure more successfully. Medical professionals can diagnose a process with the help of numerous artificial intelligence tools. However, ambiguity, irrelevance, and inconsistency in the disease dataset have an impact on the classification accuracy. Conventional machine learning techniques are inadequate for managing scenarios of uncertainty that arise in datasets related to chronic illnesses. To anticipate chronic diseases, each characteristic in the suggested work is represented with a hesitation index. In order to compute the hesitation degree as a more informative procedure in determining the association among features, the belongingness and non-belongingness aspects of the intuitionistic fuzzy concept are induced in the prediction process. This work effectively manages the large chronic illness dataset by developing an intuitionistic fuzzy hesitation index-based similarity measure to weed out unimportant and strongly associated characteristics and enhance the classification model. Utilizing the scored feature information, the deep neural network modifies its hyper parameter to address the overfitting issues that frequently arise in DNNs. The performance of the empowered intuitionistic fuzzy-based deep neural network (EIF-DNN) is analyzed using five distinct chronic conditions, including hypertension, diabetes, hepatitis, cancer, and stroke. Comparing the proposed EIF-DNN against the current versions of multi-layered networks, the simulation results show that it produces promising results in predicting chronic diseases at an early stage.