Perbandingan Algoritma Random Forest, XGBoost, dan Logistic Regression untuk Prediksi Risiko Kekambuhan Kanker Tiroid

Thyroid cancer, although relatively rare (0.85-2.5% of all cancer cases), is of serious concern due to its higher prevalence in women and challenges in diagnosis due to limitations of conventional methods such as fine-needle aspiration biopsy and ultrasound. This study aims to predict the risk of thyroid cancer recurrence by applying random forest, XGBoost, and logistic regression methods. Classifying the recurrence of thyroid cancer using 14 dataset variables obtained from Ken Saras Hospital, which amounted to 2000 datasets. The data will be classified using 3 method models and evaluated using a confusion matrix to find the best accuracy evaluation value. Based on the evaluation results, logistic regression gets an accuracy value of 83%, and random forest and XGBoost get an accuracy of 82%. Our findings prove that machine learning approaches can serve as an effective clinical decision support system in improving diagnosis efficiency and facilitating timely medical interventions. The implementation of this in clinical practice still requires integration with comprehensive medical considerations and supervision of healthcare professionals to ensure safety. The results contribute to the development of more reliable and efficient thyroid cancer diagnostic tools.