Hybrid Imputation and Machine Learning: A New Paradigm for Cervical Cancer Risk Factor Analysis

Cervical cancer remains a significant public health challenge, with early detection and risk assessment being critical for effective management. This study aims to enhance the accuracy of cervical cancer risk factor classification by addressing the issue of missing data through innovative imputation methods. This research implemented two imputation techniques: Weighted Median Imputation and a Hybrid approach combining K-Nearest Neighbors (KNN) and Regression. Using a dataset from the UCI repository, these methods are applied to preprocess the data before training multiple machines learning models, including Random Forest and Logistic Regression. Our evaluation metrics included accuracy, mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE), and R-squared (R2), enabling a comprehensive comparison of model performance. The results demonstrated that both imputation techniques significantly improved model accuracy, with the Hybrid method yielding superior results. Additionally, feature importance analysis revealed key predictors of cervical cancer risk. This research highlights effective strategies for handling missing data in healthcare datasets and contributes to the development of robust predictive models for cervical cancer risk assessment. The findings underscore the importance of advanced imputation methods in enhancing the reliability of machine learning applications in medical diagnostics.