Deep learning-driven prediction of chemotherapy response in breast cancer: a pathway toward precision medicine

Breast cancer is a major life-threatening disease that increases mortality and decreases life quality worldwide, with increasing cases in developing countries like Pakistan. Subtypes of breast cancer and late diagnosis both contribute to lower survival rates. This research uses machine learning techniques to characterize breast cancer from histopathological reports and mammograms to detect chemotherapy responses. This study identifies critical characteristics from mammograms using image processing and computer models, which showed strong discriminating power in differentiating breast cancer tumors. Mammograms and clinical data from a cancer hospital were assembled to process machine-learning models designed for high accuracy and sensitivity. The unprocessed mammograms and data were used and classified into specific groups for subsequent processing. The processed dataset will be useful for early assessment of therapy response in breast cancer patients in the future. The highest accuracy score achieved by the machine learning model is 82%.