Rehabilitation Robotics What Lies Ahead? W Zev Rymer Rehabilitation Institute of Chicago 3/3/09 International Neurorehabilitation Symposium 2009 1
Outline focus on stroke recovery Robot-assisted therapy for the arm: Initial clinical results What we ve learned: 1. Practice is key to recovery; the role of robotic forces remains unclear 2. To date, benefits are small and of limited functional significance Possible directions for making robot-assisted therapy for the arm more effective in the future 1. Do we really need a robot? 2. How do we engage the patient optimally? 3. How do we select optimal tasks 3/3/09 International Neurorehabilitation Symposium 2009 2
Brief overview of first clinical tests of robots for arm movement therapy after stroke MIME - VA Palo Alto MIT - MANUS 3/3/09 ARM Guide - RIC/ U.C. Irvine Armin- ETH rehabtek International Neurorehabilitation Symposium 2009 3
Training with MIT-MANUS and MIME resulted in modest but significant gains (e.g. Fugl-Meyer score) 3/3/09 International Neurorehabilitation Symposium 2009 4
Result from first Arm Guide Pilot Study: Robot and control groups recovered about the same amount (Kahn et al. JNER 2006) o unassisted x assisted pre p <.05 post 3/3/09 5
Patton, J. L., M. E. Phillips- Stoykov, Mussa-Ivaldi rtl. (2006). "Evaluation of robotic training forces that either enhance or reduce error in chronic hemiparetic stroke survivors." Experimental Brain Research 168(3): 368-383. 6
Visits to the lab for treatment 7
What do these results suggest? Practice produces modest but measurable benefits If patients practice moving their arm more (~20 hours), they will get somewhat better recovery results (~10-20%) It is not clear that robotic forces are mandatory The role of robotic forces remains unclear We do not know what we want the robotic forces to do assist as needed, magnify errors, or both, depending on impairment severity At present, the key element of a therapy appears to be how much time is spent practicing moving the arm 3/3/09 International Neurorehabilitation Symposium 2009 8
Can We Improve These Results? Possible reasons rehabilitation robotics may not yet be transformative Trial-and-error approach is very time and resource intensive.- basically impractical We do not now key parameters of treatment dosing- We also do not know when we should start some trials in animal models suggested more cerebral damage following early training (Schallert s group 1996) Other studies suggest that the most sensitive period arises early after stroke Destruction of nervous tissue may fundamentally limit recovery We do need a theoretical and experimental framework for studying motor learning in neurological injury 3/3/09 International Neurorehabilitation Symposium 2009 9
How might we improve these results? 1. Do we really need a robot? 2. How do we optimally engage the patient? 3. How can we select better tasks 3/3/09 10
Approach 1 to improving robotic therapy: Eliminate the robot David Reinkensmeyer with Robert Sanchez, Sarah Housman (RIC), Steve Cramer, Sandhya Rao, Vu Le, Punit Shah, Tariq Rahman (AI Dupont Hospital) If practice is key, what limits a stroke patient s ability to practice? A major factor: Weakness 3/3/09 If patients are very weak, they quickly become discouraged when they try to move their arms and stop practicing Gravity creates a force threshold: if you cannot overcome this threshold, you cannot move your arm in free space Furthermore, activation of shoulder muscles invokes unhelpful muscle activation patterns, or synergies International Neurorehabilitation Symposium 2009 11
Address Weakness 3/3/09 1. Manual assistance from therapist (flexible but expensive) 2. Robotic assistance (a powerful tool, but clinically necessary?) 3. Passive arm supports (widely used, but boring and limited movements) International Neurorehabilitation Symposium 2009 12
Design Goals for T-WREX (UCI, RIC) 1. Passive device (non-robotic) Cheaper, safer, and patient must drive every movement 2. Large 3-D workspace In order to practice functional movements 3. Adjustable weight support In order to challenge patients appropriately 4. Integrated grip sensing Because the hand and arm most often work together 5. Simple virtual reality of ADLs with quantitative feedback To motivate patients and maximize functional recovery 3/3/09 International Neurorehabilitation Symposium 2009 13
T-WREX Design Based on WREX (Wilmington Robotic Exoskeleton), (Rahman et al., 2000)*. Counterbalances weight of arm using rubber bands and four-bar mechanisms T-WREX (Training-WREX) Modifications: Adult Sized Stronger Sensorized (arm and handgrip) Adjustable WREX T-WREX *Rahman T, Sample W, Seliktar R, Alexander M, Scavina M (2000) A body-powered functional upper limb orthosis. Journal of Rehabilitation Research and Development 37:675-680. 3/3/09 International Neurorehabilitation Symposium 2009 14
Initial Testing Phase 1: Does device allow very weak stroke patients to move through a greater range? (at UCI, n = 9 chronic stroke subjects) Phase 2: Does 8 weeks of exercise with the device improve arm movement ability after chronic stroke? (at UCI, n = 5) Phase 3: How does 8 weeks of training with the device compare with conventional, supervised, self-therapy? (at RIC, n =30) 3/3/09 International Neurorehabilitation Symposium 2009 15
Phase I Result: Circle Drawing (Sanchez et al. TNSRE 2006) The ability to perform coordinated arm movement persists even following 4.5 years of non-use, but is normally masked by gravity 3/3/09 International Neurorehabilitation Symposium 2009 16
Phase 2 Result: Mean Fugl-Meyer change of 5 points T-WREX 3/3/09 International Neurorehabilitation Symposium 2009 17
Phase III: Preliminary Results Using T-WREX (30 patients) 3/3/09 18
ARMEO Hocoma A.G. has developed a device based on these concepts: Armeo Goal of commercial redesign: easier adjustment Uses springs and dials and abandons perfect balance at the elbow for simplicity Device sales began in Summer 2007 3/3/09 International Neurorehabilitation Symposium 2009 19
Summary We are at a crucial stage in the development of upper extremity robotic systems Robotic systems provide some measurable benefits, but the magnitude of these benefits is still rather modest Most of these benefits can be ascribed to repetitive practice- role of robotic forces is not yet clear, except to compensate for gravity We do not have clear learning rules on which to base our interventionwe need a better theoretical framework, and ideally better animal models, at least for stroke At present it is hard to justify large-scale use of multidegree of freedom robots for upper extremity therapy in stroke survivors 3/3/09 International Neurorehabilitation Symposium 2009 20