Closed-Loop Neuroprosthetics: Real-Time Intention Decoding with Bidirectional Feedback

Abstract: This study investigates a closed-loop neuroprosthetic framework that couples real-time intention decoding with bidirectional sensory feedback, targeting dexterous upper-limb control under clinically realistic constraints of latency, stability, and interpretability. The proposed framework formalizes intention decoding as a continuous regression problem and integrates an explicit feedback encoding stage that translates decoded kinematics into compact haptic command streams suitable for w