Voice Dialog System for Simulated Patient Robot and Detection of Interviewer Nodding

The objective of this study is to develop a medical interview training system in which an android robot is employed as a simulated patient (SP) to provide consistent training and a quantitative evaluation to medical students. In this study, first, to realize autonomous voice dialog by the android robot, called Android SP, in medical interview training, we analyzed the utterances of medical doctors in a preliminary medical interview experiment. Subsequently, based on the analysis result, we implemented a simple algorithm to classify the interviewer's utterance into “question” and others. Second, to quantify a part of the communication skills of the interviewer, we proposed a method to detect the interviewer's nod from a camera. Finally, the voice dialog system and nodding detection method were evaluated through a medical interview experiment with medical students. As a result, the voice dialog system can correctly classify most interviewer utterances. Nodding detection could reduce the false detection of head movements during utterances by excluding the section of the interviewer's speech activity from the target section. However, further improvements regarding voice dialog and the evaluation of interviewer skills are required to increase the feasibility of Android SP for medical interview training.