AI-Enhanced FES for Foot Drop: A Novel Biphasic Charge-Balanced Stimulation Approach

Abstract

Foot drop is a neuromuscular condition that impairs a patient’s ability to lift their foot while walking, leading to gait abnormalities and increased fall risk. Functional Electrical Stimulation (FES) is a rehabilitation technique that stimulates the peroneal nerve to restore normal movement. However, existing FES devices often lack precise charge balancing and timing accuracy, which can cause patient discomfort and potential tissue damage. To address these challenges, we developed a wearable, compact, and AI-driven FES system capable of generating biphasic, charge-balanced, trapezoidal pulses. The system integrates an ESP32 microcontroller, an OPA452-based stimulation circuit, and machine learning algorithms for foot lift detection. Unlike traditional threshold-based detection methods, our AI-enhanced approach adapts stimulation timing in real-time, improving synchronization with the patient’s gait. Initial testing demonstrated the system’s ability to generate precise charge-balanced stimulation pulses while accurately detecting foot-lift events. The AI-based detection method improved timing accuracy, reducing unwanted stimulation and enhancing patient comfort. The proposed system offers a technological advancement in neurorehabilitation, addressing key limitations of existing FES solutions. By integrating machine learning with FES, this work introduces a personalized, intelligent stimulation system that enhances rehabilitation outcomes for individuals with foot drop. The developed FES system achieved an average accuracy of 92.2%. Stimulation was found to be smooth between 120-220 microseconds of pulse width and 25-35 Hz of frequency.