Cervical Cancer Risk Prediction Using XGboost Classifier

Machine learning is now the most recurrently used computational technique for clinical diagnosis. Many algorithms have been developed for the prediction and identification of various diseases at an early stage, which benefits in decreasing the mortality rate. Several studies indicate that even after the development of a vaccine, cervical cancer remains the prime cause of death in women when compared with other cancer types. Cancer risk predictions have been helpful for targeted screening and maximizing treatment gains. In this paper, a scalable tree boosting method, known as Extreme Gradient Boosting (XGBoost) has been used to achieve cervical cancer risk prediction. The tuned model uses regularization and handles missing data with its in-built sparse-aware and quantile algorithms. It also outperforms the available state-of-the-art machine learning models, through its parallel computational capabilities and distribution across clusters. The results enhance the chances of applying machine learning to identify the relevant disease predictors and thus become an important tool for early detection of cervical cancer.