Early detection of breast cancer by extracting personalized network signatures from whole blood transcriptome

Abstract

Breast cancer is a disease with significantly different survival rates depending on the stage. Therefore, early diagnosis has long been an important issue. However, mammography and biopsy, which are traditional methods for early detection of breast cancer, are not sufficient in terms of accuracy and convenience, respectively. In order to solve this problem, studies using blood transcriptome gene expression analysis technology have been conducted. In this paper, in order to improve the insufficient performance of existing blood transcriptome gene expression analysis techniques, gene interactions that were not previously considered are considered patient-specific. Through this, the performance of the early breast cancer classification model has been dramatically improved, and by extracting individual network signatures, we propose genetic signatures that are specifically and meaningfully considered for early breast cancer patients through the analysis of patients by stage of breast cancer.