Breast Cancer Prognosis

An accurate breast cancer prognosis, or breast cancer survivability prediction, is important as it often guides the treatment course of action, ability to claim additional financial support from the government, actions of the patient and family, and more [1]. Predicting breast cancer survivability is commonly done using clinical features. TNM staging, the globally accepted standard used to describe cancer, was devised more than 60 years ago and only looks at three features: size of the tumor, number of regional lymph nodes with cancer, and the spread of cancer to other parts of the body. With the advent of affordable genomic sequencing and acceleration of findings in molecular biology in the past decade, molecular features may be practical to improve breast cancer prognosis. Molecular diagnostics for cancer therapy decisionmaking have shown initial promising clinical results. This has lead to a flood of published reports of signatures predictive of breast cancer phenotypes, and several molecular diagnostic tests for cancer therapy decisionmaking have gained regulatory approval in recent years [2, 3]. However, there is no consensus for the most accurate computational methods and models to predict breast cancer survivability. In addition, it is unclear that incorporating molecular data as a complement or replacement for traditional clinical diagnostic tools adds any value [4]. Therefore, it is necessary to objectively assess whether genomic data currently provides value beyond traditional clinical diagnosis tools. To aid in efforts to solve this problem, we predicted breast cancer survivability with machine learning techniques as part of the DREAM Breast Cancer Prognosis Challenge. The ultimate goal of the challenge is to objectively compare many computational algorithms through providing a common training dataset in an effort to find the best features for breast cancer prognosis. The dataset provided contains standard clinical measurements in addition to genomic information, thus allowing genomic information to be compared with standard clinical features. 2 Objective