Hedged Predictions for Traditional Chinese Chronic Gastritis Diagnosis with Confidence Machine

Traditional Chinese chronic gastritis diagnosis focuses on producing an accurate classifier and uncovering the predictive confidence for individual instance. Transductive confidence machine (TCM), which is a novel framework that provides hedged prediction coupled with valid confidence. In the framework of TCM, the efficiency of prediction depends on the nonconformity measure of samples. This paper incorporates random forests (RF) to propose a new TCM algorithm named TCM-RF. Our method benefits from the more precise and robust nonconformity measure. A case study of traditional Chinese chronic gastritis demonstrates that TCM-RF is feasible and effective.