No abstract

No abstract

No abstract

Breast cancer is the second most common cancer in women in the United States. With the revolution of the machine learning era, many researchers currently aim to find pathways and develop tools that may help to detect breast cancer early on in its development. We propose a novel generalized logistic dependent model with considerations of the dependence among selected biomarkers for breast cancer detection based on a set of nine biomarker predictors such as age, glucose, BMI, resistin, HOMA, MCP-1, leptin, insulin, and adiponectin. Our research findings demonstrate that the proposed model has the potential to predict breast cancer in women just based on five biomarkers, ie, glucose, age, BMI, resistin, and MCP-1. We also compare our model results to several other machine-learning modeling approaches including SVM, logistic regression, random forest, and multiple regression analyses using various training data sets (60%, 70%, 80% of all data) and all the dataset. It shows that the inclusion of the dependence among those five predictors in the proposed model is worth the extra model complexity and effort for achieving a significant accuracy prediction level of breast cancer detection in women. Further work in broader validation of the conclusion of our study and exploring the ability for artificial intelligence (AI) to be able to bolster these predictions based on biomarkers are also discussed. KEYWORDS BC-dependent model, biomarkers, breast cancer detection, generalized logistic dependent function, machine learning modeling 1 Concurrency Computat Pract Exper. 2020;32:e5467. wileyonlinelibrary.com/journal/cpe

No abstract