The Ethics of Rehabilitation in Virtual Reality: the role of Self-Avatars and Deep Learning

Medical Rehabilitation systems are constantly improving according to the technological evolution, and the use of virtual reality for assistive purposes is being investigated in research. One particular case of rehabilitation regards the hands and upper limbs of hemiplegic post-stroke patients. While studying possible ways to support this population, we discuss the ethical issues that arise from the use of extended reality technologies, in particular self-avatars, and the use of artificial intelligence, in particular deep learning, in neuro-cognitive applications. We present a rehabilitation approach, based on the digital embodiment of virtual limbs, in which the movements of the self-avatars are modified and optimized by the system, to lead the patients to the performance of the natural actions they lost. The ethical discussion starts from the policy of the representation of the self in virtual environments, and additional issues arise when the users have disabilities. A learning system, based on a convolutional neural network, allows the personalization of the parameters of the therapy in the long term. The collection and analysis of physiological data is also discussed, again in a scenario that involves vulnerable users.