A Brain–Machine Interface Framework for Motor Imagery Classification Using Functional Connectivity and Ensemble Learning: Toward Real-Time Applications in Healthcare and Industry

The application of the brain–machine interface (BMI) can be beneficial for healthcare and industrial applications, such as assistive technologies and control systems. Nevertheless, the lack of comprehensive EEG datasets that record numerous upper and lower limb motions presents a substantial difficulty in the development of motor imaging (MI)-based BMI. In this study, we used the MILimb dataset, which has a variety of motor imagery data. Then, a feature extraction approach was developed, incorporating functional connectivity using phase locking value (PLV) and time-domain features. The extracted features were subsequently input into an ensemble learning model to classify different motor imagery movements in comparison with the eyes-open baseline condition. Consequently, the proposed method significantly outperformed the conventional approach with classification accuracy ranging from 93.98 to 98.99%. Furthermore, the developed model was successfully tested using online classification to control a drone in real time, demonstrating its robustness and efficacy in dynamic conditions to demonstrate real-world applicability. Furthermore, these findings demonstrate the potential of proposed approach to bridge the gap between offline analysis and real-time BMI systems, supporting valuable applications in healthcare and industry.