Application of Two-Class Neural Network-based classification model to predict the onset of Thyroid Disease

A machine learning technique to diagnose thyroid disease via proper analysis is a major classification problem. The thyroid organ is an important part of our body. It helps to control our metabolism. Less amount of thyroid hormone causes hypothyroidism, and more amount of thyroid hormone causes hyperthyroidism. Therefore, the current work objective was to build a machine learning-based classification model to classify samples with thyroid disease from a publically available dataset. The classes were labeled as healthy and thyroid disease with many explanatory variables. A class balancer, namely Synthetic Minority Oversampling Technique (SMOTE), was used to balance the minority class (thyroid disease) in the dataset. In this work, filter-based feature selection algorithms, specifically mutual information in conjunction with a two-class Neural Network (NN) classifier, was used with Azure Machine Learning tools to construct a predictive model. Our proposed two-class NN model build using selected features in association with SMOTE performed better than other recent ML models with an F1Score (0.982), precision (0.968), recall (0.995), and accuracy (0.981), respectively.