Background: Crouch Gait is a fairly common pathological gait adaptation seen in children with Cerebral Palsy, characterized by inadequacy of plantar flexion, knee extension and hip extension in the stance phase of gait. From a kinetic perspective crouch gait severely impairs the ability to generate thrusting forces through the extensor muscles of stance extremity which is necessary for forward progression. Broader goal of this study was to develop and test a wearable robotic exoskeletal device for children with crouch gait, where a knee actuator provides extensor thrust. Artificial Intelligence (AI) models which control the torque output of the knee actuator, needed to be trained to activate, control, or organize the output. Objectives: Rhythmical/repetitive patterns, like the phases of normal gait, needed to be identified in crouch gait, which could be used to train AI models Methodology: A total of 30 children with cerebral palsy were screened of which 7 children aged between 7 and 14 years, who presented with classical crouch gait were recruited. Kinesiological data was collected through accelerometric and goniometric placed bilaterally on thigh, leg, and knee, as well as pressure sensors embedded on to bilateral insoles at the fore-foot and hind foot, while the children walked over a 5 meter course. All sensor data was relayed through an Arduino micro-controller unit for analog to digital conversion and stored on a laptop computer. All Accelerometer and goniometric data was plotted against time on MS Excel to identify specific patterns or motifs. Results: Key kinematic motifs could be identified in crouch gait, which bestowed semblance of rhythmicity, and cyclical pattern, which was difficult to identify under normal observation. Conclusion: Accelerometric, as well as goniometric data suggest the presence of inherent kinematic patterns in the same. These could be used for training AI models for robotic exoskeletal devices.
Read full abstract