The research team conducted logistic and Cox regression according to the behavioral data of gastroesophageal reflux disease patients who had long been drinking caffeinated coffee drinks, and determined the sensitivity and mathematical rationality of AI prediction model in behavioral science, which can support the research team to build a deep learning neural network and complete the prediction of gastrointestinal tract involvement.