Performance evaluation of modified Gail and Gail-Rosner-Colditz models for breast cancer risk prediction in Vietnam
Main Article Content
Abstract
This retrospective study evaluated the predictive performance of a modified Gail model and a combined Gail-Rosner-Colditz model in 31,016 women participating in a breast cancer screening program, including 19 breast cancer cases. Data were split by stratified sampling into training (60%) and testing (40%) sets, and class imbalance was addressed by random undersampling combined with SMOTE. Model performance was assessed using ROC-AUC, PR-AUC, and Brier score; factors associated with breast cancer risk were analyzed using Firth’s penalized logistic regression. Adding Rosner-Colditz variables increased ROC-AUC across all three algorithms, and Weighted Kernel Logistic Regression achieved the highest ROC-AUC (0.662 ± 0.062). Age was associated with higher odds of breast cancer, whereas parity of two or three births was associated with lower odds compared with nulliparity. The combined Gail-Rosner-Colditz model showed better predictive performance than the modified Gail model, although overall predictive value remained limited because of the very low number of events.
Article Details
Keywords
Breast cancer, Gail model, Rosner-Colditz model, risk prediction, machine learning
References
2. Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81(24):1879-1886. doi:10.1093/jnci/81.24.1879.
3. Rosner B, Colditz GA. Nurses’ Health Study: log-incidence mathematical model of breast cancer incidence. J Natl Cancer Inst. 1996;88(6):359-364. doi:10.1093/jnci/88.6.359.
4. Rice MS, Tworoger SS, Hankinson SE, et al. Breast cancer risk prediction: an update to the Rosner-Colditz breast cancer incidence model. Breast Cancer Res Treat. 2017;166(1):227-240. doi:10.1007/s10549-017-4391-5.
5. Anothaisintawee T, Teerawattananon Y, Wiratkapun C, Kasamesup V, Thakkinstian A. Risk prediction models of breast cancer: a systematic review of model performances. Breast Cancer Res Treat. 2012;133(1):1-10. doi:10.1007/s10549-011-1853-z.
6. Firth D. Bias reduction of maximum likelihood estimates. Biometrika. 1993;80(1):27-38. doi:10.1093/biomet/80.1.27.
7. Heinze G, Schemper M. A solution to the problem of separation in logistic regression. Stat Med. 2002;21(16):2409-2419. doi:10.1002/sim.1047.
8. Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP. SMOTE: Synthetic Minority Over-sampling Technique. J Artif Intell Res. 2002;16:321-357. doi:10.1613/jair.953.
9. Saito T, Rehmsmeier M. The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets. PLoS One. 2015;10(3). doi:10.1371/journal.pone.0118432.
10. Min JW, Chang MC, Lee HK. Validation of risk assessment models for predicting the incidence of breast cancer in Korean women. J Breast Cancer. 2014;17(3):226-235. doi:10.4048/jbc.2014.17.3.226.