Cervical cancer affects a large portion of the female population, making the prediction of this disease using Machine Learning (ML) of utmost importance. ML algorithms can be integrated into complex, intelligent, agent-based systems that can offer decision support to resident medical doctors or even experienced medical doctors. For instance, an experienced medical doctor may diagnose a case but need expert support that related to another medical specialty. Data imbalance is frequent in healthcare data and has a negative influence on predictions made using ML algorithms. Cancer data, in general, and cervical cancer data, in particular, are frequently imbalanced. For this study, we chose a messy, real-life cervical cancer dataset available in the Kaggle repository that includes large amounts of missing and noisy values. To identify the best imbalanced technique for this medical dataset, the performances of eleven important resampling methods are compared, combined with the following state-of-the-art ML models that are frequently applied in predictive healtchare research: K-Nearest Neighbors (KNN) (with k values of 2 and 3), binary Logistic Regression (bLR), and Random Forest (RF). The studied resampling methods include seven undersampling methods and four oversampling methods. For this dataset, the imbalance ratio was 12.73, with a 95% confidence interval ranging from 9.23% to 16.22%. The obtained results show that resampling methods help improve the classification ability of prediction models applied to cervical cancer data. The applied oversampling techniques for handling imbalanced data generally outperformed the undersampling methods. The average balanced accuracy for oversampling was 77.44%, compared to 62.28% for undersampling. When detecting the minority class, oversampling achieved an average score of 60.80%, while undersampling scored 41.36%. The logistic regression classifier had the greatest impact on balanced techniques, while random forest achieved promising performance, even before applying balancing techniques. Initially, KNN2 outperformed KNN3 across all metrics, including balanced accuracy, for which KNN2 achieved 53.57%, compared to 52.71% for KNN3. However, after applying oversampling techniques, KNN3 significantly improved its balanced accuracy to 73.78%, while that of KNN2 increased to 63.89%. Additionally, KNN3 outperformed KNN2 in minority class performance, scoring 55.72% compared to KNN2’s 33.93%.

Cervical Cancer affects a large part of female population that makes the prediction of this disease based on Machine Learning (ML) by outmost importance. ML algorithms can be integrated in complex intelligent agent-based systems that can offer decision support to the resident medical doctors or even to experienced medical doctors. For instance can be mentioned the situation when an experienced medical doctor diagnose a case but he/she needs expertise support that is related to another medical specialty. Data imbalance is frequent in healthcare data and has a negative influencing effect in making predictions using ML algorithms. Cancer data generally and cervical cancer data particularly are frequently imbalanced. Based on this fact the study of data imbalance impact on diverse state-of-the-art ML prediction algorithms is important. This research subject is also motivated by the fact that in many research are presented experimental evaluations of algorithms without characterization of the data on that they have been applied. Such characterizations could give clear indication to other researchers regarding the applicability of the algorithms on their specific data. Specifically, if the data have the respective characteristics than are expectable the same performance evaluation results like those in the reported research. For the study we chosed a messy real-life Cervical Cancer dataset available in a recognized data repository included a large amount of missing and noisy values. To identify the best imbalanced technique for this medical dataset, it is compared the performance of eleven important resampling methods combined with the following state-of-the-art ML models: K-Nearest Neighbours (with k values of 2 and 3), binary Logistic Regression, and Random Forest, that are frequently applied in prediction types of researches in healthcare. The studied resampling methods includes seven undersampling methods namely Condensed Nearest Neighbour, Tomek Links, Edited Nearest Neighbours, Repeated Edited Nearest Neighbours, All K-Nearest Neighbours, NearMiss, Neighbourhood Cleaning Rule, and Instance Hardness Threshold, and four oversampling methods namely Synthetic Minority Oversampling Technique (SMOTE), Adaptive Synthetic Sampling Approach for Imbalanced Learning SMOTE, Support Vector Machine SMOTE and Borderline SMOTE. In the case of the dataset for the confidence interval with 95% confidence level was between 9.23 and 16.22, while the imbalance ratio is 12.73. The obtained results show that resampling methods help the learning models to improve the classification ability of cervical cancer. The applied oversampling techniques generally showed better results than undersampling methods. In the case of Logistic Regression classifier had the highest impact on balanced techniques, while Random Forest had promising performance, even before balancing techniques and KNN2 was generally better than KNN3.