Advanced Machine Learning in Prediction of Second Primary Cancer in Colorectal Cancer

Colorectal cancer (CRC) ranked third among most commonly diagnosed cancers worldwide. The onset of second primary cancer (SPC) is an important indicator in treating CRC. We tried to use the advanced machine learning method in order to find the factors of SPC. Patients with CRC from three medical centers were identified from cancer registries in Taiwan. The classifier of A Library for Support Vector Machines (LIBSVM) and Reduced Error Pruning Tree (REPTree) were applied to analyze the relationship of clinical features with category by constructing the optimized model of every classified issue. Machine learning can be used to rank the factor affecting the secondary primary malignancy. In the clinical practice, physician should be of aware the possibility of cancer recurrence and routine checkups for early second primary malignancy detection is recommended. The accuracy rate of the may need more big data. The machine learning method is feasible in detecting/predicting potential second primary cancer in the future.