An SVM-based discrimination method of tracheal-intubation skill between experts and novices

摘要

Our research aims at objective evaluation in endotracheal intubation techniques. We have aimed at the establishment of decision methods of tracheal intubation skill level of medical doctors. As a preliminary approach to achieving the goal, we proposed a discrimination method between experts and novices. We obtained the doctor's full-body motion data by using a motion capture suit. This motion data were discriminated by a machine learning technique, i.e., an SVM classifier. Considering the importance of the movement of tracheal intubation, we employed the velocity, acceleration, and angular velocity as the feature vector for the SVM. As a result, we could classify with an average accuracy rate of 97.6%. It shows the effectiveness of our proposed method.