Figure 1.

Outline of the proposed BC risk prediction system using ML. In the training phase, networks of interacting genetic and demographic risk factors for BC are identified. These networks of features are then used to predict whether an unlabelled individual is a cancer case or a healthy control in the testing phase. This study provides two examples showing that a combination of interacting genetic variants (SNPs) with BC risk factors related to both familial history and oestrogen metabolism can increase BC risk prediction accuracy.