For the development of processed food for hypoglcemia, it is important to construct model system to confirm factors that reduce the glycemic index in real food. This study was to investigate the relationship between the gel model system and real food the high amylose type of segoami using response surface methodology. The independent variables were concentration (X1; 10, 15, 20, 25, and 30%) and steaming time (X2; 30, 35, 40, 45, and 50 min). The predicted glycemic index (Y1) was analyzed as a dependent variable. The regression of pGI was 0.7343, indicating that the model fits the data well in the prediction test. The predict glycemic index of the gel was in the range of 71.38 ~ 83.78, depending on the gel preparation conditions and predicted optimum condition was 23.7% gel concentration and 43.8 minutes of gelatinization. The predict glycemic index of rice gel were decreased with an increase of gel concentration, rather than the steaming time. In order to confirm the effect on in vitro digestibility in real food, the real food (Garraedduk) was prepared by applying the optimal conditions. As a result, there was no significant difference between the predicted value (77.1) and the experiment value (76.6). This result showed that the in vitro digestibility in real food can be predicted by applying the gel model system.