Artificial Intelligence to diagnose low back pain using motion captures

Low back pain (LBP) is a highly common musculoskeletal condition and the leading cause of work absenteeism, yet in most cases the source of nociceptive pain is rarely identified and physical therapy duration is administered regardless. One of the most important eval- uation assessments for LBP is range of motion analysis (ROM) that consists of calculating the difference in angles between the standing position and the maximum reaching point in multiple axial exercises, at present ROM is performed with taping measure lacking precision. This graduation project is based in the application of motion capture (Mo- Cap) techniques with inertial sensor units for precise spatiotemporal measures to assess ROM exercises from healthy and clinically diagnosed patients with LBP from Imbabura, Ecuador in order to automate classification of normal or pathological ranges of movement using seven machine learning algorithms (Logistic regression, SVM, K-nearest neighbours, Decision tree, Random forest, Gradient boosting algorithm and Multilayer perceptron) to evaluate and compare the resulting metrics. All machine learning techniques obtained ac- curacy above 80% and three models obtained >90% accuracy (Support Vector Machines, Random forest, Multilayer perceptron), concluding SVM is the best performing algorithm. Obtained results share similar behavior compared to related works.

• Autor principal: Loján Cueva, Alejandro Bolívar (author)
• Format: bachelorThesis
• Idioma: eng
• Publicat: 2023
• Matèries: Rango de movimiento; Machine Learning; Lumbalgia; Range of movement; Low back pain; Motion Capture (MoCap)
• Accés en línia: http://repositorio.yachaytech.edu.ec/handle/123456789/667