Related Experiment Video
Updated: Jan 9, 2026
Author Spotlight: A Novel Setup to Conduct Naturalistic Laboratory Experiments with Real Human Actors in Scenarios
Published on: August 4, 2023
Comparing Patient Simulation With a Humanoid Robot or a Human Actor in Terms of Training Success and Acceptance:
Patricia Schwarz1, Sandra Hellmers1, Sebastian Spanknebel2
1Assistance Systems and Medical Device Technology, Department of Health Services Research, Carl von Ossietzky Universität Oldenburg, Oldenburg, Germany.
Background:
The addition of simulated patients to medical and nursing training makes it possible to create a link between theory and practice. This makes what has been learned more realistic and allows the complexity and multilayered nature of many illnesses to be reflected in a real-life setting. However, the selection, training, and supervision of actors as simulated patients is time consuming and expensive. In this study, we investigated how differently students and nurses perceive 2 different methods of patient simulation.
Objective:
The aim of this pilot study was to investigate whether patient behavior simulated by a humanoid robot is comparable to patient simulation by actors in videos in terms of training success and user acceptance. Participants were asked to recognize the symptoms presented by the humanoid robot and make a diagnosis. For comparison purposes, we asked a second group of participants to make a diagnosis based on a video featuring a human patient actor.
Methods:
We asked the participants (medical students and nursing staff; N=21) to conduct a psychopathological assessment. Group 1 (n=11) used the humanoid robot as a patient simulator, and group 2 (n=10) watched the identical symptoms in a video with a human actor as patient.
Results:
The participants had a mean age of 28.7 (SD 3.5) years. The students were in their sixth semester and had, on average, 7.6 (SD 3.3) years of professional experience in the medical field. The correct diagnosis was made 90% (9/10) of the time based on the video with the human patient actor and 91% (10/11) of the time based on the robot. One participant in each group made the wrong diagnosis, constituting a total error rate of 10% (2/21). In general, participants with the humanoid robot as patient simulator felt more confident that their diagnosis was correct compared to those with the human actor as patient (humanoid robot: 9/11, 82% were neutral to very confident and 2/11, 18% were uncertain to very uncertain; human actor: 9/10, 90% were neutral to very confident and 1/10, 10% were uncertain or unsure).
Conclusions:
The simulations of the human actor in the video were judged to be more realistic overall than those of the humanoid robot as patient simulator. However, the differences between the simulation methods in relation to the result (diagnosis) were very small. The results of our pilot study show a good performance of the robot in the simulation of selected psychiatric patient cases. We conclude that a humanoid robot could be a useful addition to patient simulators in medical education and discuss future directions.
