학술논문

An Intelligent Virtual-Reality System With Multi-Model Sensing for Cue-Elicited Craving in Patients With Methamphetamine Use Disorder
Document Type
Periodical
Source
IEEE Transactions on Biomedical Engineering IEEE Trans. Biomed. Eng. Biomedical Engineering, IEEE Transactions on. 68(7):2270-2280 Jul, 2021
Subject
Bioengineering
Computing and Processing
Components, Circuits, Devices and Systems
Communication, Networking and Broadcast Technologies
Drugs
Multiuser detection
Electrocardiography
Gaze tracking
Sensors
Medical treatment
Physiology
Methamphetamine
virtual reality (VR)
machine learning
Heart Rate Variability (HRV)
electrocardiogram (ECG)
Galvanic Skin Response (GSR)
eye tracking
Language
ISSN
0018-9294
1558-2531
Abstract
Methamphetamine abuse is getting worse amongst the younger population. While there is methadone or buprenorphine harm-reduction treatment for heroin addicts, there is no drug treatment for addicts with methamphetamine use disorder (MUD). Recently, non-medication treatment, such as the cue-elicited craving method integrated with biofeedback, has been widely used. Further, virtual reality (VR) is proposed to simulate an immersive virtual environment for cue-elicited craving in therapy. In this study, we developed a VR system equipped with flavor simulation for the purpose of inducing cravings for MUD patients in therapy. The VR system was integrated with multi-model sensors, such as an electrocardiogram (ECG), galvanic skin response (GSR) and eye tracking to measure various physiological responses from MUD patients in the virtual environment. The goal of the study was to validate the effectiveness of the proposed VR system in inducing the craving of MUD patients via the physiological data. Clinical trials were performed with 20 MUD patients and 11 healthy subjects. VR stimulation was applied to each subject and the physiological data was measured at the time of pre-VR stimulation and post-VR stimulation. A variety of features were extracted from the raw data of heart rate variability (HRV), GSR and eye tracking. The results of statistical analysis found that quite a few features of HRV, GSR and eye tracking had significant differences between pre-VR stimulation and post-VR stimulation in MUD patients but not in healthy subjects. Also, the data of post-VR stimulation showed a significant difference between MUD patients and healthy subjects. Correlation analysis was made and several features between HRV and GSR were found to be correlated. Further, several machine learning methods were applied and showed that the classification accuracy between MUD and healthy subjects at post-VR stimulation attained to 89.8%. In conclusion, the proposed VR system was validated to effectively induce the drug craving in MUD patients.