A Hybrid Supervised Machine Learning Classifier System for Breast Cancer Prognosis Using Feature Selection and Data Imbalance Handling Approaches

Solanki, Yogendra Singh; Chakrabarti, Prąsun; Jasiński, Michał; Leonowicz, Zbigniew; Bolshev, Vadim; Виноградов, А. П.; Jasińska, Elżbieta; Goňo, Radomír; Nami, Mohammad

doi:10.3390/electronics10060699

Cited by 46 publications

(29 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Solanki et al also investigated the prediction of benign or malignant BC using selected ML techniques. The results showed that the J-48 yielded the best classification performance with an accuracy of 98.83% (49).…”

Section: Discussionmentioning

confidence: 99%

Comparing Data Mining Algorithms for Breast Cancer Diagnosis

et al. 2022

View full text Add to dashboard Cite

Background: Early screening and diagnosis of breast cancer (BC) is critical for improving the quality of care and reducing the mortality rate. Objectives: This study aimed to construct and compare the performance of several machine learning (ML) algorithms in predicting BC. Methods: This descriptive and applied study included 1,052 samples (442 BC and 710 non-BC) with 30 features related to positive and negative BC diagnoses. The data mining (DM) process was implemented using the selected algorithm, including J-48 and random forest (RF) decision tree (DT), multilayer perceptron (MLP), Naïve Bayes (NB), Adaboost (AB), and logistics regression (LR) classifier. Then, we obtained the best algorithm by comparing their performances using the confusion matrix and area under the receiver operator characteristics (ROC) curve (AUC). Finally, we adopted the best model for BC prognosis. Results: The results of evaluating various DM algorithms revealed that the J-48 DT algorithm had the best performance (AUC = 0.922), followed by the AB, MLP, LR, and RF algorithms (AUC: 0.899, 0819, 0.716, and 0.703, respectively). Also, the NB algorithm achieved the lowest performance in this regard (AUC = 0.669). Conclusions: The ML presents a reasonable level of accuracy for an early diagnosis and screening of breast malignancies. Also, the empirical results showed that the J-48 DT algorithm yielded higher performance than other classifiers.

show abstract

Section: Discussionmentioning

confidence: 99%

Comparing Data Mining Algorithms for Breast Cancer Diagnosis

et al. 2022

View full text Add to dashboard Cite

show abstract

“…On the other hand, the over-sampling method is based on SMOTE. It has been widely used as a baseline over-sampling method for breast cancer datasets [14][15][16][17]. The percentage of synthetic instances was set to make the two datasets become balanced datasets where the malignant and benign classes contain the same numbers of data samples.…”

Section: The Feature Selection and Over-sampling Methodsmentioning

confidence: 99%

“…According to Fernandez et al [16], SMOTE over-sampling can benefit from the use of feature selection, where feature selection is performed over the class imbalanced dataset to select a subset feature of it, and then the reduced dataset is over-sampled to make it contain the same size of the data samples as in the majority and minority classes. Recently, Solanki et al [17] propose the contrary procedure that SMOTE be performed first to rebalance the breast cancer dataset, and then wrapper-based feature selection methods can be applied to reduce the feature dimensions.…”

Section: Introductionmentioning

confidence: 99%

On Combining Feature Selection and Over-Sampling Techniques for Breast Cancer Prediction

et al. 2021

View full text Add to dashboard Cite

Breast cancer prediction datasets are usually class imbalanced, where the number of data samples in the malignant and benign patient classes are significantly different. Over-sampling techniques can be used to re-balance the datasets to construct more effective prediction models. Moreover, some related studies have considered feature selection to remove irrelevant features from the datasets for further performance improvement. However, since the order of combining feature selection and over-sampling can result in different training sets to construct the prediction model, it is unknown which order performs better. In this paper, the information gain (IG) and genetic algorithm (GA) feature selection methods and the synthetic minority over-sampling technique (SMOTE) are used for different combinations. The experimental results based on two breast cancer datasets show that the combination of feature selection and over-sampling outperform the single usage of either feature selection and over-sampling for the highly class imbalanced datasets. In particular, performing IG first and SMOTE second is the better choice. For other datasets with a small class imbalance ratio and a smaller number of features, performing SMOTE is enough to construct an effective prediction model.

show abstract

“…Machine learning (ML) has achieved very good results for prediction and timely treatment of various diseases [22,23]. For instance, Solanki et al [24], proposed methods for improving the performance of ML classification models, namely a support vector machine (SVM), a decision tree, and a multilayer perceptron (MLP), i.e., a feed-forward artificial neural network (ANN), for the prognosis of breast cancer. ML models can be utilized also to predict blood glucose levels, based on collected medical data and various human body's health indicators.…”

Section: Related Workmentioning

confidence: 99%

RG Hyperparameter Optimization Approach for Improved Indirect Prediction of Blood Glucose Levels by Boosting Ensemble Learning

Wang

Zhang

et al. 2021

Electronics

View full text Add to dashboard Cite

This paper proposes an RG hyperparameter optimization approach, based on a sequential use of random search (R) and grid search (G), for improving the blood glucose level prediction of boosting ensemble learning models. An indirect prediction of blood glucose levels in patients is performed, based on historical medical data collected by means of physical examination methods, using 40 human body’s health indicators. The conducted experiments with real clinical data proved that the proposed RG double optimization approach helps improve the prediction performance of four state-of-the-art boosting ensemble learning models enriched by it, achieving 1.47% to 24.40% MSE improvement and 0.75% to 11.54% RMSE improvement.

show abstract

A Hybrid Supervised Machine Learning Classifier System for Breast Cancer Prognosis Using Feature Selection and Data Imbalance Handling Approaches

Cited by 46 publications

References 29 publications

Comparing Data Mining Algorithms for Breast Cancer Diagnosis

Comparing Data Mining Algorithms for Breast Cancer Diagnosis

On Combining Feature Selection and Over-Sampling Techniques for Breast Cancer Prediction

RG Hyperparameter Optimization Approach for Improved Indirect Prediction of Blood Glucose Levels by Boosting Ensemble Learning

Contact Info

Product

Resources

About