X García-Massó1, P Serra-Añó2, L M Gonzalez3, Y Ye-Lin4, G Prats-Boluda4, J Garcia-Casado4. 1. Departamento de Didáctica de la Expresión Musical, Plástica y Corporal, Universidad de Valencia, Valencia, Spain. 2. Departamento de Fisioterapia, Universidad de Valencia, Valencia, Spain. 3. Departamento de Educación Física y Deportiva, Universidad de Valencia, Valencia, Spain. 4. Grupo de Bioelectrónica (I3BH), Universitat Politècnica de València, Camino de Vera s/n Ed.8B, Valencia, Spain.
Abstract
STUDY DESIGN: This was a cross-sectional study. OBJECTIVES: The main objective of this study was to develop and test classification algorithms based on machine learning using accelerometers to identify the activity type performed by manual wheelchair users with spinal cord injury (SCI). SETTING: The study was conducted in the Physical Therapy department and the Physical Education and Sports department of the University of Valencia. METHODS: A total of 20 volunteers were asked to perform 10 physical activities, lying down, body transfers, moving items, mopping, working on a computer, watching TV, arm-ergometer exercises, passive propulsion, slow propulsion and fast propulsion, while fitted with four accelerometers placed on both wrists, chest and waist. The activities were grouped into five categories: sedentary, locomotion, housework, body transfers and moderate physical activity. Different machine learning algorithms were used to develop individual and group activity classifiers from the acceleration data for different combinations of number and position of the accelerometers. RESULTS: We found that although the accuracy of the classifiers for individual activities was moderate (55-72%), with higher values for a greater number of accelerometers, grouped activities were correctly classified in a high percentage of cases (83.2-93.6%). CONCLUSIONS: With only two accelerometers and the quadratic discriminant analysis algorithm we achieved a reasonably accurate group activity recognition system (>90%). Such a system with the minimum of intervention would be a valuable tool for studying physical activity in individuals with SCI.
STUDY DESIGN: This was a cross-sectional study. OBJECTIVES: The main objective of this study was to develop and test classification algorithms based on machine learning using accelerometers to identify the activity type performed by manual wheelchair users with spinal cord injury (SCI). SETTING: The study was conducted in the Physical Therapy department and the Physical Education and Sports department of the University of Valencia. METHODS: A total of 20 volunteers were asked to perform 10 physical activities, lying down, body transfers, moving items, mopping, working on a computer, watching TV, arm-ergometer exercises, passive propulsion, slow propulsion and fast propulsion, while fitted with four accelerometers placed on both wrists, chest and waist. The activities were grouped into five categories: sedentary, locomotion, housework, body transfers and moderate physical activity. Different machine learning algorithms were used to develop individual and group activity classifiers from the acceleration data for different combinations of number and position of the accelerometers. RESULTS: We found that although the accuracy of the classifiers for individual activities was moderate (55-72%), with higher values for a greater number of accelerometers, grouped activities were correctly classified in a high percentage of cases (83.2-93.6%). CONCLUSIONS: With only two accelerometers and the quadratic discriminant analysis algorithm we achieved a reasonably accurate group activity recognition system (>90%). Such a system with the minimum of intervention would be a valuable tool for studying physical activity in individuals with SCI.
Authors: X García-Massó; P Serra-Añó; L M García-Raffi; E A Sánchez-Pérez; J López-Pascual; L M Gonzalez Journal: Spinal Cord Date: 2013-09-03 Impact factor: 2.772
Authors: Richard A Washburn; Weimo Zhu; Edward McAuley; Michael Frogley; Stephen F Figoni Journal: Arch Phys Med Rehabil Date: 2002-02 Impact factor: 3.966
Authors: François Routhier; Josiane Lettre; William C Miller; Jaimie F Borisoff; Kate Keetch; Ian M Mitchell; CanWheel Research Team Journal: Assist Technol Date: 2017-01-04
Authors: Kiley L Armstrong; Lisa M Lombardo; Kevin M Foglyano; Musa L Audu; Ronald J Triolo Journal: J Neuroeng Rehabil Date: 2018-03-12 Impact factor: 4.262
Authors: Wiebe H K de Vries; Sabrina Amrein; Ursina Arnet; Laura Mayrhuber; Cristina Ehrmann; H E J Veeger Journal: Sensors (Basel) Date: 2022-09-29 Impact factor: 3.847
Authors: Emma Fortune; Beth A Cloud-Biebl; Stefan I Madansingh; Che G Ngufor; Meegan G Van Straaten; Brianna M Goodwin; Dennis H Murphree; Kristin D Zhao; Melissa M Morrow Journal: J Electromyogr Kinesiol Date: 2019-07-17 Impact factor: 2.368
Authors: Marika T Leving; Henricus L D Horemans; Riemer J K Vegter; Sonja de Groot; Johannes B J Bussmann; Lucas H V van der Woude Journal: PLoS One Date: 2018-04-11 Impact factor: 3.240