An Efficient JSH-FCM-Based Thyroid Disease Detection Using Ash-Ann with Stage Classification via a Fuzzy Rule-Based Approach

Abstract

One of the most misunderstood and undiagnosed diseases is termed Thyroid Disease (TD), which is a subset of endocrinology. It emerges at the edge of the thyroid gland due to the abnormal development of thyroid tissue. Owing to the lack of awareness and early diagnosis, TD is a critical problem in underdeveloped countries. For TD diagnosis, various theoretical works have been introduced; still, in the early diagnosis of TD, accurate prediction of the thyroid data is a significant problem. Thus, by utilizing Altered SigHyper activation-centric Artificial Neural Network (ANN) (ASH-ANN) with various stage classifications, an effectual Jaccard Similarity and He-initialization induced Fuzzy C-Means (FCM) (JSH-FCM) clustering-centric TD detection system is proposed by means of a fuzzy rule-centric methodology. Initially, for accurate detection, the thyroid dataset is gathered and the data is pre-processed. Next, by JSH-FCM clustering, the age-centric clustering is carried out. After that, by utilizing Pearson Correlation-amalgamated Principal Component Analysis ((PC)2A), Feature Extraction (FE) and feature selection is conducted. Moreover, to detect the TD kind, an ASH-ANN classifier is wielded. Finally, for differentiating the stages of TD, the fuzzy rule is employed. The experimental outcomes depict that the proposed system achieved superior performance with an accuracy of 97.32% when weighed against the prevailing system; in addition, the stages of TD are differentiated precisely.