Predictive Analysis of Thyroid Cancer Recurrence using Clinical and Pathological Factors
Abstract
Thyroid cancer is one of the most common endocrine malignancies worldwide, and its recurrence remains a significant clinical challenge. This study aims to develop predictive models to identify patients at risk of thyroid cancer recurrence based on demographic, clinical, and pathological variables. Utilizing a dataset of 383 thyroid cancer patients, we applied data preprocessing, exploratory analysis, and machine learning models, including logistic regression and random forest classification. Our findings demonstrate that pathology type, tumor staging, focality, and response to initial treatment are among the most predictive features. These insights can support early intervention strategies and improved patient outcomes.
Keywords
Download Options
Introduction
Thyroid cancer recurrence significantly impacts long-term prognosis and quality of life. While survival rates are generally high, recurrent disease may necessitate repeated treatments and can lead to increased morbidity. Therefore, early identification of high-risk patients is essential.
This study investigates the predictive capacity of various demographic (age, gender), behavioral (smoking, radiation exposure), clinical (physical exam, staging), and pathological (tumor type, focality) features in forecasting recurrence using machine learning techniques.
Conclusion
This research demonstrates the utility of machine learning models in identifying thyroid cancer patients at high risk of recurrence. Random Forest proved particularly effective due to its capacity to model complex interactions. These findings can assist clinicians in designing individualized follow-up strategies, enhancing patient care, and optimizing healthcare resources.