Advanced Machine Learning Techniques for Precise Thyroid Disease Classification: A Novel Approach

Abstract

Thyroid disease refers to disorders affecting the thyroid gland’s hormone production, leading to imbalances that can cause major health issues like heart disease, stroke, and infertility if timely treatment is not received. Early diagnosis and treatment of thyroid disease are important to prevent complications. To categorize patients as having normal thyroid function, hyperthyroidism, or hypothyroidism, multiple machine learning models such as SVM, Decision Tree Classifier, Random Forest Classifier, XGBoost, and Light Gradient Boosting (LGBM) are used. Thyroid hormone levels and other clinical data from patient records will be used to train the algorithms. A range of machine learning approaches such as feature selection, preprocessing, handling unbalanced data, modeling, and assessment are applied and models are fine-tuned with hyperparameters. Random Forest Classifier has produced the highest classification accuracy, precision, recall, and f1-score of 99.39\%. The important knowledge gathered from this study may be used to create a thorough foundation for machine-learning systems that anticipate thyroid diseases.