Classification of Gait in Children Suffering from Cerebral Palsy using Machine Learning and Deep Learning Techniques

Classification of abnormal walking patterns are required for developing customised therapies for children with cerebral palsy (CP). Hip, ankle, and knee joint angles calculated with respect to sagittal and coronal planes are used in the proposed work to evaluate a number of machine learning (ML) and deep learning (DL) models. The main goal is to assess how well various algorithms work on abnormal gait data, where the data has been improved by converting time series data into heatmap images and using methods like zooming and rotation on converted images to increase model robustness. The augmented gait data is classified using seven different models: Random Forest (RF), Support Vector Machines (SVM), Long Short-Term Memory (LSTM) networks, Convolutional Neural Networks (CNNs), k-Nearest Neighbours (KNN) and ResNet. The CNN model achieved the highest classification accuracy of 86.26