Efficient Ensemble Machine Learning Framework for Early Prediction of Thyroid Cancer Using Clinical Data

Thyroid cancer is considered as one of the global health burden that occur due to the abnormal growth of cells in the thyroid gland. One of the remedies to improve survival rates and provide successful treatment is to diagnose the disease at an early stage. The traditional diagnostic approach depends on the clinical findings and imaging-based techniques. Hence, it is recommended to adopt a data-driven approach by incorporating Machine Learning (ML) for the early prediction of diseases. According to the literature findings, machine learning applied to predict thyroid cancer using clinical data has demonstrated high precision in predicting the disease and has also improved personalized treatments. The XGBoost is an ML model that can handle complex features in the dataset by analyzing the feature interaction, and LightGBM will work efficiently on an imbalanced dataset. In the proposed research work, an ensemble-based machine learning framework has been implemented by combining the Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting (LightGBM). The main objective of the proposed work is to predict thyroid cancer with higher accuracy by applying an ensemble-based machine learning approach. The proposed work also aims to improve the model bias and generalization by utilizing the capabilities of two powerful ensemble algorithms. The experimental results illustrated that the proposed ensemble framework achieved greater accuracy, precision, recall, and AUC score compared to traditional ML algorithms. Hence, the proposed ensemble framework can be effectively used to predict thyroid cancer disease early.