Human gait is analyzed based on their walking patterns. Gait disorders can be diagnosed early with the help of fast-growing technologies. By using gait features, Parkinson disease, hemiplegic disease, and neuropathic disease can be identified. To obtain a high performance of gait analysis and classification, deep-learning techniques are presented. In this paper, Convolution Neural Network (CNN) with Long Short-Term Memory (LSTM) is proposed. The Chinese Academy of Sciences (CASIA) dataset is used to analyse the different types of gait disorders. A set of features is trained with the data from this dataset to reduce training time and remove irrelevant and noisy data. The lean and ramp angle features extracted from the dataset are considered as the prominent features for gait analysis in this work. As a result, the proposed method is capable of accurately classifying disorders and requires less computational time. In this study, we compare the experimental results with those of other machine learning algorithms. In order to assess the proposed system's performance, performance metrics such as F1 score, precision, accuracy, and recall are used. Due to its increased performance, the proposed system was able to surpass other techniques of a similar nature.

convolution neural network, CASIA dataset, long short-term memory, machine learning algorithms.