File
B. Bischof1, S. Amsuess2, I. Popovic3, T. Fuchsberger4, A. Hahn3
1Otto Bock Healthcare Products GmbH, Orthopaedic Technology, Vienna, Austria, 2Otto Bock Healthcare Products GmbH, Systems Engineering, Vienna, Austria, 3Otto Bock Healthcare Products GmbH, Clinical Research & Services, Vienna, Austria, 4AG-Clinics, Department of Plastic, Reconstructive, Aesthetic and Hand Surgery, Traunstein, Germany
Background: Myoelectric prosthesis control is based on a control of a single prosthesis degree of freedom such as closing and opening of the hand.
Special switching methods (such as myosignals) are used to control different grips.
If wrist rotation is desired, amputees mostly use co-contraction of their forearm muscles to switch into this mode; the same signals (for open and close) are then used to control the wrist rotator.
Users often rate switching to different modes as slow, cumbersome and not intuitive.
Special switching methods (such as myosignals) are used to control different grips.
If wrist rotation is desired, amputees mostly use co-contraction of their forearm muscles to switch into this mode; the same signals (for open and close) are then used to control the wrist rotator.
Users often rate switching to different modes as slow, cumbersome and not intuitive.
Purpose: Investigating biofeedback training on a mobile app in order to fully leverage Pattern Recognition (PR) potential to control upper limb prostheses.
Methods: (1) 9 transradial amputees transitioning from conventional control (CC) → Patern recognition control, recruited at study 3 sites
(2) Biofeedback training with Android App for home use
(3) Home-use 1 month of feedback and pattern recognition controlled prosthesis
(4) 7 male and 2 female participants with a balanced amputation side between left and right
(5) Biofeedback was implemented as a visualization of 8 EMG electrode amplitudes depicted on a Radar plot. 4 movements (open, close, rotation up and down) were trained, stored and given as reference patterns for the user to replicate.
(2) Biofeedback training with Android App for home use
(3) Home-use 1 month of feedback and pattern recognition controlled prosthesis
(4) 7 male and 2 female participants with a balanced amputation side between left and right
(5) Biofeedback was implemented as a visualization of 8 EMG electrode amplitudes depicted on a Radar plot. 4 movements (open, close, rotation up and down) were trained, stored and given as reference patterns for the user to replicate.
Results: • Participants improved on average more on complex tasks (clothespin reloc. test up: -18%, down: -34% completion time) than on simple grasping tasks (modified Box&Blocks test: +21% completion time)
• Participants rated the ease of understanding the control aided by the described biofeedback control with 1.0±0.0 after initial training session (ca. 0.5 hrs) and 1.4±0.52 after one month of in-depth at home use on a 1-5 likert scale (1 best, 5 worst)
• Using the DASH score improvement in daily living was observed in 4 out of 8 patients for the 2nd follow-up with PR and for 1/3 of the patients for the follow-up after 1 month of use pattern recognition.
• Participants rated the ease of understanding the control aided by the described biofeedback control with 1.0±0.0 after initial training session (ca. 0.5 hrs) and 1.4±0.52 after one month of in-depth at home use on a 1-5 likert scale (1 best, 5 worst)
• Using the DASH score improvement in daily living was observed in 4 out of 8 patients for the 2nd follow-up with PR and for 1/3 of the patients for the follow-up after 1 month of use pattern recognition.
Conclusion(s): • Pattern Recognition improved the unilateral gross manual dexterity and ability to control two DoFs
• Training was positively rated by patients and CPOs.
• Longer patient accommodation time might minimize mild problems in fine and gross motor movements observed during the first month of pattern recognition home-use
• Training was positively rated by patients and CPOs.
• Longer patient accommodation time might minimize mild problems in fine and gross motor movements observed during the first month of pattern recognition home-use
Implications: Biofeedback training on a mobile app in combination with pattern recognition to control a myoelectric prosthesis helps participants to visualize their forearm muscle activity while thinking on specific phantom movements. This facilitates the education support for therapists and other qualified specialists.
Funding, acknowledgements: Financial support of Otto Bock Healthcare Products GmbH
Keywords: Pattern Recognition, Biofeedback, Prosthesis control
Topic: Innovative technology: robotics
Did this work require ethics approval? Yes
Institution: Otto Bock Healthcare Products GmbH, Vienna
Committee: University & Faculty of Medicine Tübingen, Germany
Ethics number: 301/2016MPG
All authors, affiliations and abstracts have been published as submitted.
