Closed-Loop Approaches and Decoding Algorithms in Bidirectional Brain-Machine Interface Systems

As brain-machine interface (BMI) systems advance toward closed-loop and bidirectional architectures, understanding the interaction between decoding algorithms and neuroplasticity is essential for clinical applications. This study aims to comprehensively review the architecture, decoding algorithms, and the role of sensory feedback in bidirectional BMI systems. This research is a descriptive-analytical review based on a systematic evaluation of articles published in PubMed and IEEE databases (from ۲۰۰۸ to ۲۰۲۶), focusing on invasive systems, state-space algorithms, and novel clinical applications. Investigations show that the transition from linear models (such as the population vector) to dynamic state-space models (Kalman filter) and deep learning has significantly increased the information transfer rate and decoding stability. Furthermore, the integration of intracortical microstimulation (ICMS) as sensory feedback not only improves closed-loop control but also leads to cortical reorganization and modulation of internal states (such as mood) through co-adaptation mechanisms. Bidirectional BMIs operate beyond mere motor assistive tools, evolving into neuroprosthetic platforms for neurorehabilitation and the regulation of limbic networks. Future challenges include the development of flexible arrays and low-power implantable processors.

کلیدواژه ها:

Brain-Machine Interface (BMI) ، Closed-Loop System ، Kalman Filter ، Intracortical Microstimulation (ICMS) ، Neuroplasticity ، Motor Rehabilitation