Improving Equity in Healthcare: Machine Learning-Based Thyroid Disease Classification

Abstract

The thyroid, an important component of the endocrine system at the tip of the neck, plays an important role in the production of thyroxine, which is essential for overall health Disturbances in thyroid hormone production can lead to insufficient or excess levels. The use of large amounts of data in the healthcare industry has become increasingly complex, necessitating the use of machine learning. Our research focuses on thyroid disease, using a combination of machine learning and unbiased samples. Our main goal is to classify thyroid diseases into hypothyroidism, regular, and hyperthyroidism using machine learning. We used a real-world data set from Kaggle, split into 70% for training and 30% for testing. This division allows one to search for measures of accuracy and unbiasedness, especially with respect to logistic regression values. When reweighting methods were implemented, we saw an increase in accuracy and fairness metrics. Our study demonstrates the effectiveness of machine learning models with unbiased priority, yielding an accuracy of 84.58% with an appropriate value of -0.007. The importance of our work extends to the application of artificial intelligence (AI) in healthcare. By using AI algorithms to identify patterns in data, we demonstrate the potential to enhance medical research and treatment outcomes. Furthermore, our inclusion of justice considerations in model construction highlights the ethical considerations in the use of AI. It emphasizes the importance of fair and transparent decision-making in health care systems. This is consistent with broader AI research objectives, which aim to develop technologies that not only maximize accuracy but also maintain principles of fairness and accountability in their implementation.