It is an important issue to determine amino acid requirement in human in nutritional science. The indicator amino acid oxidation technique is often employed to the issue. An indicator amino acid is measured repeatedly from each participant at intake of various dosage of the target amino acid to determine the requirement. Change-point models are often employed to determine the requirement. Carry-over effects due to repeated measurement may influence on estimation of the change-point. In this paper, a model is introduced to evaluate the carry-over effect and to estimate the change-point adjusting for the carry-over effect. A simple inference procedure is introduced based on an approximation of the change-point model. It can be easily executed using a standard software for the non-linear mixed effect models. We apply the model to a dataset analyzed in a previous study to estimate the requirement of lysine in human. A simulation study is also conducted to evaluate the performance of the proposed method with small sample size.