Nowadays, there are more and more people start to notice their health in daily life that makes this topic become a popular topic. Many health care systems or applications are developed, but these systems only depend on calculating the amount of exercise by devices, but don't involve other situations, such as life rhythm, social condition, etc. In this paper, the authors discuss and improve these existence problems. The paper puts emphasis on current health history and the different career's life rhythm. Active mining is utilized to analyze and construct user models. This research presents a new way for exercise support; it also shows the ability and potential for robotic system to offer personalized service.