Thyroid disorders represent a significant global health challenge, affecting approximately 5% of the population and necessitating precise screening to prevent risks like cardiac arrhythmias and metabolic imbalances. This research addresses the limitations of manual diagnostics—often subjective and prone to error—by introducing an Automated Thyroid Diagnostic Assistant. Leveraging the XGBoost machine learning algorithm, the system classifies patient status into four distinct categories: Normal, Primary Hypothyroid, Compensated Hypothyroid, and Hyperthyroid. The model was developed using the UCI Thyroid Disease dataset, utilizing 18 critical features, including demographics, medical history, and hormone levels (TSH, T3, and T4). To ensure clinical utility, the system is deployed via a Flask-based web interface, providing medical professionals with near real-time predictions and confidence scores.
Joseph et al. (Thu,) studied this question.