Classification of Pathological Gait Modes Using Phase Space Features and Combining Joints Information

Articles in Press

Document Type : Original Article

Authors

Department of Biomedical Engineering, Imam Reza International University, Mashhad, Razavi Khorasan, Iran.

10.22034/jasp.2025.55909.1222

Abstract

Gait patterns provide valuable medical information for making medical and clinical decisions influenced by physical characteristics, orthopedic health, and skeletal health. Weakness or injury in certain body parts can lead to an abnormal walking pattern. The primary objective of this research is to classify different pathological states using features extracted through phase space dynamics based on 3D data obtained from the joints of 25 participants. This study utilizes a freely accessible database containing six walking modes - one healthy normal and five pathological modes. Ten healthy individuals participated in the data collection process and were instructed to simulate the gaits for each of the six modes. Each participant repeated each walking type twenty times. The first step involves separately reconstructing the phase space for the following data: 1) Integration of XYZ vector information from the joints using Euclidean distance, 2) Integration of XYZ vector information from the joints using angle information, and 3) Mapping the information to a new space by combining the XYZ directions through principal component analysis. Subsequently, several geometric features are extracted from the reconstructed data. Each of the features obtained from steps 1 to 3 is inputted individually, as well as their combinations into support vector machine, k-nearest neighbor, random forest, and hybrid classifiers. The random forest classifier demonstrates the highest accuracy of 99% when considering all features.

Keywords

Main Subjects

Advanced Signal Processing

Articles in Press, Accepted Manuscript
Available Online from 06 September 2025

Files

Share

How to cite

Statistics