Breast cancer (BC) is the most commonly diagnosed cancer and the leading cause of cancer death among females. We aimed to develop a deep learning (DL)-based risk stratification system to predict BC patients one early using minimal features from electronic health records. We identified 8,606 patients who underwent a diagnosis of BC between January 1999 and December 2013. The CNN model was developed to predict BC one-year earlier. CNN model demonstrated great performance in predicting BC cancer. For the prediction of BC one year earlier, the areas under the receiver operating characteristic curve was 0.918. The sensitivity, specificity, and positive predictive value were 0.816, 0.848, and 0.541, respectively. The CNN model based on variables available in EHR can be a promising tool to distinguish patients at risk of BC.