|
Abstract:
|
In a fraction of patients of breast cancer, the relapse after surgery is incurable. However, the accurate prognostic can significantly lower the risk of relapse by administering systemic chemotherapy. Thus, predicting the risk of relapse becomes critical. In this study, a newly developed permanental classification method was applied in predicting the risk of relapse using the microarray gene expression profiles. A risk score based on the prediction was also proposed to formulating the risk of relapse. The performance was assessed using independent testing datasets and the comparisons with other similar methods, including support vector machine (SVM) and random forests were also conducted, in which the receiver operating characteristic (ROC) area under curve (AUC) as well as the testing error rate were calculated. The proposed risk score based on the permanental approach is providing a practical measure of the risk of relapse, which would contribute to the accurate prognostic, as well as the precise assignment of systemic chemotherapy.
|
