Towards An Accurate Stacked Ensemble Learning Model For Thyroid Earlier Detection

Thyroid disease is one of the most common endocrine disorders worldwide. However, thyroid conditions can be challenging to diagnose because symptoms are very similar to those of other diseases. A proper diagnosis depends on clinical examination and many blood tests involving a large amount of complex data that is difficult to interpret. Early thyroid detection is crucial since it significantly reduces complications and minimizes death risk. The main objective of this study is to create an accurate framework for improving the diagnostic accuracy of thyroid diseases. For this purpose, we propose a three-stage approach based on dimensionality reduction using feature selection, data sampling to handle the data-imbalance problem, and stacked ensemble learning instead of a single machine learning algorithm to give the final prediction. This research shows that the proposed approach can diagnose thyroid disease more accurately than existing techniques, achieving 99.49% of precision and 99.46% in terms of F1-score.