Behavioral imitation learning-based brain training simulation system

Brain Training Simulation System Based on Imitation Learning

This technology acquires non-invasive brain signals (such as EEG and NIRS), performs preprocessing and AI-based machine learning to continuously decode a patient's movement intentions, links these to the operation modes and difficulty levels of rehabilitation equipment (such as treadmills), and induces neuroplasticity through visual avatar content and neurofeedback.

Conventional bottom-up rehabilitation training struggles to encourage active patient participation, and technologies focused on single-motion recognition cannot change training modes continuously, failing to provide the sensory-motor virtuous cycle required for chronic or paralyzed patients.

This technology implements a continuous movement intention recognition algorithm based on brain signals (applying wavelet transforms and AI models), a control unit for the speed and intensity of rehabilitation equipment using state transition diagrams (S1–S5), and an evaluation system that monitors the user's training status to provide feedback on appropriate training protocols and store them in a database. It can be applied to rehabilitation training, gait assistance, and medical/welfare services, improving rehabilitation by clearly recognizing the user's operational intent using brain signals and operating the rehabilitation training accordingly.

‍

‍

Key Features:

• A user intention expression unit that operates the rehabilitation equipment and presents rehabilitation training content based on the user's movement intention recognized by the user movement intention decoding unit.
• A user movement intention decoding unit that recognizes the user's movement intention based on brain signal data processed by the brain signal acquisition and processing unit.
• A brain signal acquisition and processing unit that acquires and processes the user's brain signals using non-invasive brain activation measurement methods.
• A brain training simulation system based on imitation learning, characterized by providing neurofeedback to induce brain activity by presenting training speed visually or audibly through a monitor, which serves as a user monitoring device, in text or voice format.

‍

This invention was developed with support from the Ministry of Trade, Industry and Energy for the development of biosignal interface technology with over 90% gait intention detection accuracy for various gait rehabilitation of stroke patients, and application technology for overground gait rehabilitation robots.