GAIT PATTERN RECOGNITION IN CEREBRAL PALSY PATIENTS USING NEURAL NETWORK MODELLING

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Rehabilitation Science

Background:

• Interpreting 3D gait analysis data for cerebral palsy is complex and time-consuming.
• Neural networks learn from data to predict unknown patterns.
• This study aimed to develop models for recognizing cerebral palsy gait patterns.

Purpose of the Study:

• To create and evaluate neural network models for classifying cerebral palsy gait.
• To differentiate between normal, pre-treatment, and post-treatment gait patterns.
• To assess the accuracy of different neural network models in gait pattern recognition.

Main Methods:

• Gait data from 28 cerebral palsy patients and 26 healthy controls were analyzed.
• Vicon Nexus system captured kinematic and kinetic parameters of lower limb joints.
• Neural network models were trained on 70% of data and tested on 30%.

Main Results:

• Models utilizing all parameters or joint angles/moments achieved ~95% accuracy in gait pattern recognition.
• Models using joint power and moments showed variable success rates between 70-90%.

Conclusions:

• Neural network models demonstrate high accuracy in identifying cerebral palsy gait patterns.
• These models show potential for clinical application in gait analysis for cerebral palsy patients.