Background: Breast cancer (BC) is one of the most common malignancies in women. Early diagnosis of BC and metastasis among the patients based on an accurate system can increase survival of the patients to > 86%. This study aimed to compare the performance of six machine learning techniques two traditional methods for the prediction of BC survival and metastasis. Methods: We used a dataset that include the records of 550 breast cancer patients. Naive Bayes (NB), Random Forest (RF), AdaBoost, Support Vector Machine (SVM), Least-square SVM (LSSVM) and Adabag, Logistic Regression (LR) and Linear Discriminant Analysis were used for the prediction of breast cancer survival and metastasis. The performance of the used techniques was evaluated with sensitivity, specificity, likelihood ratio and total accuracy. Results: Out of 550 patients, 83.4% were alive and 85% did not experience metastasis. In prediction of survival, the average specificity of all techniques was >=3D 94% and the SVM and LDA have greater sensitivity (73%) in comparison to other techniques. The greater total accuracy (93%) belonged to the SVM and LDA. For metastasis prediction, the RF had the highest specificity (98%), the NB had highest sensitivity (36%) and the LR and LDA had the highest total accuracy (86%). Conclusions: Our finding showed that the SVM outperformed other machine learning methods in prediction of survival of the patients in terms of several criteria. Nevertheless, the LDA technique as a classical method showed similar performance.
Page:
293---299
Related
Batch download
Cited By
noting
Similar Literature
Submit Feedback
Please wait while the file you selected is being converted