Predictive Modeling for Thyroid Cancer Diagnosis Using XGBoost: A Data-Driven Approach with Enhanced Interpretability and Accuracy  
Author Tarun Jaiswal

 
Co-Author(s) Sujata Dash; Ganpati Panda

 
Abstract This research explores the use of the XGBoost classifier to predict thyroid cancer using clinical data. Thyroid cancer is rapidly increasing, and early diagnosis is vital for effective treatment. The model classifies patients as benign or malignant, using variables like age, TSH levels, and hormones, with high accuracy. Key innovations include advanced feature selection to reduce dimensionality while maintaining accuracy and an adaptive thresholding approach that improves sensitivity for detecting malignant cases. The findings suggest that XGBoost, with these modifications, can enhance early diagnosis, support clinical decision-making, and reduce unnecessary biopsies.

 
Keywords Thyroid Cancer, XGBoost, Machine Learning, Cancer Diagnosis, Feature Selection, Adaptive Thresholding, Medical Prediction
   
    Article #:  DSBFI25- 16
 
Proceedings of 3rd ISSAT International Conference on Data Science in Business, Finance and Industry
January 6-8, 2025 - Da Nang, Vietnam