A mind-driven VR serious game : design, implementation, and evaluation of a BCI-controlled application based on motor imagery in virtual reality for therapeutic use in ADHD patients

Abstract

Psychological diseases are causing an increasing number of health issues for human society, affecting individuals, health organisations, as well as the entire healthcare system.Therefore, and according to predictions from the World Health Organisation (WHO),psychological diseases will become the main concern for global healthcare issues by theyear 2030. One of these major health conditions is known as attention deficit hyperactivity disorder (ADHD), affecting approximately 9.8% of all children aged three to 17 livingin the United States of America [ 1 ]. Compared to this, the world-wide prevalence ofADHD ranging from 5 to 10 percent, matching 30 to 40% of all psychiatric and neurologic referrals for children and adolescents mental health outpatient services [2, 3 ]. This worldwide psychological disorder is classified as a debilitating condition characterised bypersistent and developmentally inappropriate behaviours such as overactivity, inattention and impulsivity [ 4 ]. All these facts and predictions highlight the importance of effective pharmacological as well as non-pharmacological treatment approaches, such as therapeutic interventions [1 ]. Therefore, the aim of this work is to propose a virtual reality seriousgame driven by someone’s mind, leading to a steering mechanism based on motor imagery tasks as well as user’s concentration level. By integrating both mental tasks into a gamified setting, weakened neural connections within the cortex are intended to be strengthened and the concentration ability of the ADHD patients to be improved.For this purpose, a comprehensive literature review was conducted, the current state ofthe art and research gaps were identified, and four experts were consulted to define specificrequirements for serious games focusing on the therapy of ADHD patients. Based on 20 identified requirements, the serious game „Formula Mind“ was developed, simulating a simple racing game. For the implementation, electroencephalography (EEG) data including eight EEG channels were acquired and labelled according to the performed motor imagery (MI) task. Subsequently, a deep learning EEGNet model was trained,achieving an accuracy of 89%. By establishing online classification of measured motorimagery tasks during the gameplay, users are able to perform turns. Between the MI measurements, users control the speed of the racing car with their concentration level. After the implementation phase, an evaluation of the game including five external participants was conducted, providing initial positive insights about the gameplay and the design.