Cervical Cancer Prediction Using Machine Learning

Abstract

Cervical cancer is a significant global health burden with high morbidity, mortality, and economic loss. However, resource limitations, lack of qualified healthcare professionals, and inadequate information makes it challenging to prevent and cure cervical cancer in countries like Ethiopia. Early detection and treatment are crucial for improving patient outcomes and reducing disease burden. Over diagnosis is a major issue in healthcare, resulting from abnormal cancer prediction results. The main objective of this study is to develop a predictive model for cervical cancer by using machine learning techniques. For predicting cervical cancer, bagging ensemble learning technique is employed and the data is collected from University of Gondar Referral Hospital, Gondar Polyclinic, Debark Hospital by including 6240 patients demographic, habit, and medical record. The collected data was then preprocessed, which included removing missing values, handling outliers, and handling imbalanced data, which transform the data into a suitable format for analysis. By using a chi-square feature selection technique and SMOTE data balancing technique to make sure our model is trained on relevant features and balanced data by conduct experiments based on the selected bagging ensemble model with DT as a base learner and compare with other selected classification models LR, RF, and Xgboost with an accuracy of 92.39%, 61.68%, 79.51%, and 71.52% respectively. The developed predictive model is used to develop the artifact for demonstration to potential users. This is done by integrating the bes