A Novel Classification Methodology for Thyroid Cancer using C4.5 with Firefly Optimization Algorithm (CFOA)

Abstract

Objectives: The main objective of this research is to detect thyroid cancer in its early stages and to improve the accuracy using a novel method C4.5 with Firefly Optimization Algorithm (CFOA). This research also focuses on developing an effective machine learning-based accurate prediction model. Methods: The images in this investigation are associated to thyroid disorder with 4672 samples of people including both females and males having hypothyroidism and hyperthyroidism, as well as healthy people without thyroid disorder. The data was gathered for one year with the primary goal of classifying thyroid disease using machine learning (ML) algorithms. These data comprise of Gender, Age, Thyroid hormone (T4), Triiodothyronine (T3), Thyroid stimulating hormone (TSH), etc. The performance of this proposed method was measured using parameters such as accuracy, precision, F1-score and recall. Findings: The performance of this proposed method was assessed with the state-ofthe- art existing methods like Naive Bayes (NB) algorithm, K-Nearest Neighbor (KNN) Algorithm and Adaboost. The proposed algorithm showed maximum precision of 0.9935, recall of 0.9971, F1-score of 0.9951 and accuracy of 0.9981 respectively when compared with the existing algorithms. Novelty: A novel algorithm C4.5 with Firefly Optimization Algorithm was proposed in this paper to speed-up and to increase the effectiveness of the machine learning algorithm. Moreover, this study focuses on developing an accurate prediction model by comparing classification algorithms based on accuracy and confusion matrices and then identifying the most effective classifier based on performance. Keywords: Thyroid cancer; Classification technique; Hypothyroidism; Hyperthyroidism; Accuracy; Precision; F1score; Recall