Enhanced Personalized Clinical Treatment: Regression Models’ Applications to Breast Cancer Patient Management

Abstract

Incorporation of statistical predictive models into clinical practice enhances personalized treatment planning and patient management. This study compares the performance of six parametric regression models, namely Weibull, Exponential, Log-logistic, Lognormal, Gamma, and Gompertz, in identifying the prognostic factors affecting the survival of breast cancer patients. Survival models were estimated for a dataset of 686 breast cancer patients from the German Breast Cancer Study (GBCS). The following comparison criteria were used to compare the survival models: Akaike Information Criterion, Bayesian Information Criterion, Log-Likelihood, and Deviance. While all the models yielded similar outcomes, the model selection criteria indicated that the Lognormal model had the best fit for the data. All models identified the same set of significant predictors, suggesting that tumor size, tumor grade, number of lymph nodes involved, and progesterone receptor count significantly influenced survival. The findings from this study show the lognormal model to be the best fit for survival and the importance of early detection and testing to improve the prognosis of breast cancer patients.