Feasibility of a Center of Mass Based Fuzzy-Logic Phase Detection Algorithm for Post-Spinal Cord Injury Gait

Current approaches for restoring walking post spinal cord injury (SCI) use feedforward systems to apply neuromuscular stimulation to lower extremities to realize stepping. However, abnormal gait patterns still persist. During neurotypical walking, center of mass (CoM) follows well-defined trajectories and reflects whole-body movement that has the potential to be exploited to better control stimulation and improve gait performance. Fuzzy logic algorithms (FLA) mimic human reasoning and provide robust frameworks for noisy inputs typical of information from biological systems. Thus, we examined the feasibility of CoM based FLA to accurately detect four phases of the gait cycle off-line and determine its potential for controlling stimulation to improve walking mechanics post-SCI. Five neurotypical subjects participated in a session of overground walking to develop a CoM based FLA with five inputs that accounted for gait periodicity at various gait speeds. We optimized the FLA with a Genetics Algorithm and completed a cross validation before verifying the system offline with data from one neurotypical person and three individuals with SCI by computing goodness indices (G). The FLA performed well for double support and swing phase ( ${G}\le 0.52$ ) across participants. This study supports the feasibility of utilizing components of the CoM kinematics as features for gait phase detection during walking post-SCI.