부산대학교 도서관

In this paper, a self-training system was developed for assisting the student nurses to improve their skills of transferring patients from a bed to a wheelchair. The twenty evaluation items corresponding to the skills were defined through the theory of nursing care and the experience of nursing teachers. In addition, the evaluation parameters of each item were defined. The system employed two depth camera sensors, Kinect, to measure the movement of trainee and patient. Several color marks were attached on their joints and the 3D position of these color marks were calculated by the proposed method which uses the depth data and the RGB color data. Furthermore, to automatically evaluate, a motion recognition method based on time sequence and the features of the 3D trajectory of patient's head were proposed. Using this system, we determined the thresholds of the evaluation items' parameters by a preliminary experiment in which nursing teachers and student nurses participated. Finally, an experiment to validate the system was conducted. Performance of 5 student nurses was evaluated by the teachers and system, respectively. Using the teacher's evaluation as the benchmark, the accuracy of the system is up to 85 %.