Abstract

Thyroid disorders represent a significant global health concern requiring accurate and timely diagnosis. While traditional machine learning models such as Support Vector Machines and Random Forest have demonstrated promising performance, their reliability is often limited by class imbalance and overlapping clinical features. This study proposes a soft voting ensemble framework that integrates Random Forest, Support Vector Machine, and Gradient Boosting classifiers to improve multiclass thyroid disorder classification. The proposed model was evaluated on a dataset containing clinical and hormonal attributes including TSH, T3, and T4 levels. Experimental results show that the ensemble approach improves classification stability and balanced performance across classes, achieving an overall accuracy of 88%, macro F1-score of 0.87, and improved detection of minority thyroid conditions compared to individual models. The findings demonstrate the potential of ensemble learning for enhancing clinical decision-support systems for thyroid disorder diagnosis.

Authors

M.S. Mir, U.H. Mir, Majid Zaman, Shabir Najar
University of Kashmir, India

Keywords

Thyroid Disorder, Soft Voting Ensemble, Machine Learning Random Forest, Support Vector Machine, Gradient Boosting