Socially interactive agents for robotic neurorehabilitation training: conceptualization and proof-of-concept study

Area of Science:

• Neuroscience
• Robotics
• Artificial Intelligence

Background:

• Neurorehabilitation requires intensive therapy for functional recovery.
• Limited availability of neurorehabilitation professionals restricts care.
• Robotic devices offer potential but lack human interaction and motivation.

Purpose of the Study:

• To introduce an AI-based system for personalized, out-of-hospital neurorehabilitation assistance.
• To integrate a socially interactive agent into a robotic framework to enhance user engagement.
• To test the feasibility of the system with healthy participants.

Main Methods:

• Developed an AI system comprising a rehabilitation device, affective signal classification, and a socially interactive agent.
• Designed the system for autonomous home-based training with virtual coaching.
• Integrated an interactive agent into a neurorehabilitation robotic framework.

Main Results:

• Participants adapted well to the AI-assisted neurorehabilitation system.
• The interactive agent positively influenced user engagement during exercises.
• The agent did not cause distraction, suggesting effective integration.

Conclusions:

• The AI-based system shows promise for delivering effective neurorehabilitation outside clinical settings.
• Socially interactive agents can successfully enhance engagement in robotic-assisted therapy.
• This approach addresses the challenge of professional availability in neurorehabilitation.