PRAGMATIC TECHNIQUES FOR THYROID DISEASE CLASSIFICATION USING MACHINE LEARNING

Abstract

The thyroid gland, a vital regulator of the body's metabolism, releases hormones into the bloodstream to control bodily functions. Hypothyroidism and hyperthyroidism are two hormone disorders associated with the thyroid. Factors such as iodine deficiency, autoimmune conditions, and inflammation can contribute to thyroid issues. Diagnosis involves a blood test, but data noise is common. Implementing data cleaning techniques simplifies analytics, allowing for a clearer assessment of the patient's risk of developing thyroid disease. In this study, machine learning techniques were employed for the classification of thyroid diseases using a dataset from Kaggle. The thyroid's crucial role in essential functions underscores the significance of early disease diagnosis. Multiple machine learning algorithms were implemented, and their performance was assessed through confusion matrices. Pragmatic techniques were applied for data optimization, contributing to the development of robust models. A comprehensive analysis compared various metrics, revealing that the XGBoost Classifier (XGB) demonstrated the highest effectiveness with 99.46% accuracy. The study highlights the importance of tailoring techniques to dataset characteristics, available resources, and expertise. Addi-tionally, comparisons with established models showcase notable improvements in results and procedures