This study presents a portable and intelligent wrist rehabilitation system that leverages custom-made flexible sensors and AI-driven motion classification to evaluate hand rehabilitation performance and degrees of freedom. The system integrates electromyographic signal acquisition, real-time motion tracking, and predictive modeling, providing a precise and adaptive rehabilitation assessment. The findings of this research include improved sensor integration, optimized AI models, and comprehensive experimental validation. Comparative performance evaluations demonstrate that the enhanced system outperforms traditional rehabilitation devices in terms of classification accuracy and usability. These advancements contribute to the development of more adaptive, accessible, and efficient rehabilitation technologies for individuals with motor impairment.