Anindo Roy, PhD

1998   Bachelor of Technology (B.Tech.), Major: Electrical Engineering, JMI University, New Delhi, India

2000   Master of Philosophy (M.Phil.), Engineering (Major: Control Systems), University of Sussex, Brighton, Sussex, UK

2005   Doctor of Philosophy (Ph.D.), Applied Science (Major: Engineering Science and Systems), Dissertation: “Robust Stabilization of Multi-Body Biomechanical Systems: A Control Theoretic Approach”, University of Arkansas at Little Rock, Little Rock, Arkansas, USA

2006   Post-Doctoral Fellow (Neuromechanics), Georgia Institute of Technology (GeorgiaTech)/Emory University 

2009   Post-Doctoral Associate (Rehabilitation Robotics), Massachusetts Institute of Technology (MIT)

NAME: Roy, Anindo

eRA COMMONS USER NAME: ANINDOROY

POSITION TITLE: Associate Professor

EDUCATION/TRAINING

A. Personal Statement.

Background: My training and research interests are in the areas of automatic control systems, rehabilitation robotics, biomechanics modeling, human-robot interaction control, and novel methodologies including robot-based and other instrumented assessments of physical and mobility function. After training under the pioneers of impedance control and rehabilitation robotics (Neville Hogan, Hermano Krebs), my work over the past 10 years has synergized these areas to address a single overarching need: develop and deploy therapeutic robotics and interactive information technologies to restore functional independence in those with mobility disabilitiesresulting from cerebrovascular disease (stroke) and other neurologic injuries, and aging. Toward this end, I have spearheaded the engineering development and clinical testing of an ankle robot exoskeleton (ABOT) for task-oriented neurorehabilitation, in stroke and other disability conditions.

Evidence Guided Technology Development: Since 2009 my research has focused on the design and testing of ABOT-assisted seated visually guided-evoked isolated ankle interventions, as well as treadmill-integrated gait training. The former has shown benefits in paretic ankle motor control and impairments leading improved overground walking, in both chronic and sub-acute stroke The latter has included clinical testing of a gait event-triggered adaptive control system that integrates ABOT into task-oriented treadmill walking exercise in chronic stroke survivors with residual hemiparetic gait deficits, such as foot drop. This technological breakthroughallows for the first time, a deficit-adjusted robotic gait therapy--one that links robotic support to specific functional deficits of hemiparetic gait in a manner that accommodates in real-time, step-to-step variability during walking thereby ensuring robust human-robot stability and patient safety. Its unique features include: customized therapy to individual gait deficit profiles, “on-the-fly” robotic tuning, calibration with recovery profiles, and performance-based/motor learning guided robotic progression, both step-by-step and across sessions, to optimize ankle neuromotor learning and maximize locomotor function. This approach has shown unprecedented benefits in chronic stroke, reversing foot drop, restoring push-off forces, and correcting heel-first landings, in just 6 weeks. This control framework is now being advanced to build and refine a clinical-data driven Artificial Intelligence (supervised machine learning) repository, the first of its kind in any brain injured population that would enable co-operative robotic training of walking, dynamic balance and real-world, diverse activities of daily living (ADL) mobility tasks in stroke. Engineering is already underway to provide clinicians/therapists, a “suite” of task- and disease-specific adaptive controllers that will provide individualized robotic prescription strategies.

Research Trajectory: Current research foci include: (1) Clinical testing of adaptive control ankle robotics for treadmill gait training (TMR) in chronic stroke and subjects with neuro-orthopedic injuries that cause foot drop, as well as investigating comparative efficacy of TMR against other therapeutic modalities such as traditional PT and TM-alone. While my prior work has focused on the effects of ABOT interventions on gait and balance biomechanics, new data suggest that individualized robotic therapies such as TMR may also positively impact cardiometabolic fitness in chronic stroke, a key aim of the proposed R01. Hence, my research has diversified to investigate longer-term (e.g. 3- or 6 months) TMR and its effects on health and mobility function beyond gait biomechanics and ankle neuromotor control (cardiometabolic/vascular, metabolic cost, muscle and molecular and epigenetic mechanisms); (2) Modifying ABOT adaptive controller to conduct training of ADLs beyond treadmill walking, include overground mobility sub-tasks such as stepping and staircase walking; (3) Development and clinical testing of a portable, low-cost ABOT to provide patients with continued ABOT therapy outside of the clinic/lab and after cessation of clinical studies; and (4) Development and testing of scalable and cost-effective telerehabilitation technologies (IVET) to disseminate proven task-oriented exercise models from point-of-care (center) into the home for those with stroke and age-related mobility deficits. This unique technology will enable translation of our protocols for safe, effective administration, and ultimately increase access, quality, and continuity of exercise rehabilitation therapies for chronic disease management.

Education and Mentorship: My direct teaching at the University of Maryland School of Medicine has entailed the development of new courses to meet the needs of students, and complement evolving research. I introduce elements from my own research into course content, such that future practitioners, scientists, and engineers are best trained for research/academic careers. My teaching approach blends lecture content with interactive, hands-on experiences for maximum engagement and impact. In order to attract the best engineering student talent for our neuro-robotics research and as part of the UM Board of Regents mission to “bridge campuses”, I teach a senior-level elective “Assistive robotics” and a graduate elective “Rehabilitation Robotics” at the University of Maryland, College Park. My formal mentoring (HP-STAR, MARC-USTAR Program, curriculum based Independent Study) has spanned across both education levels (STEM students, postdoctoral scholars, medical students) and disciplines (engineering, medicine, allied health sciences). My mentoring approach is highly “goal driven”, with emphasis on tangible outputs in the form of publications, posters, or presentations to set clear goals for mentees, while affording objective evaluation of/by the mentor. Many of my mentees are pursuing their pre- and postdoctoral careers at prestigious places (UC Irvine-Mechanical Engineering, Brown University-Computational Neuroscience, UMCP-Electrical & Computer Engineering, and Army Research Lab).

B. Positions and Employment.

2005-2006       Postdoctoral Research Fellow, Georgia Institute of Technology, Atlanta, GA

2006-2009       Postdoctoral Associate, Massachusetts Institute of Technology (MIT), Cambridge, MA

2009-2016       Assistant Professor of Neurology, University of Maryland School of Medicine, Baltimore, MD

2012-pres        Chief Robotics Engineer, VA Maryland Health Care System, Baltimore, MD

2016-pres        Associate Professor of Neurology, University of Maryland School of Medicine, Baltimore, MD

2012-pres        Core Director, Robotics & Engineering, VA RR&D MERCE, Baltimore, MD

2014-pres        Faculty, Maryland Robotics Center, University of Maryland, College Park, MD

2015-pres        Faculty, Office of Advanced Engineering Education, University of Maryland, College Park, MD

C.    Contribution to Science

Pioneering Ankle Robotics.Upper-body rehabilitation robotics emerged as a new field with the invention of impedance control (1985), but the field of lower-limb robotics remained undeveloped. My post-doctoral research (MIT, 2006-09) pioneered the design, development, and pre-clinical testing of a modular ankle robot exoskeleton (ABOT) designed to improve ankle contributions into gait and balance functions, after disabling stroke or other neurologic conditions that affect the lower extremity. ABOT is the first lower-limb exoskeleton that is impedance controlled (“assists as-needed” – the gold standard for safe, gentle human-robot interaction), modular (therapeutically isolates and targets the affected ankle, a joint critical to mobility function), and back-drivable (“gets out of the way”, as appropriate – that is, user does not perceive the robot’s end-point inertia, and does not have to learn robot dynamics). Impedance control and back-drivable hardware enables highly compliant behavior (“minimally intrusive”) to allow and encourage volitional dynamics (even for planes of movement that are unactuated), a fundamental requirement for motor learning (ML). Thus, the ABOT is a highly versatile platform to study motor learning and a training device suited for multiple therapeutic settings (seated, supine, upright). Moreover, ABOT design distributes the device mass in a manner that does not negatively alter hemiparetic gait pattern. This body of work has broken down previous technological barriers that impeded development of suitable rehabilitation robotics for lower-limb therapy. I am the co-inventor and served as the primary investigator for this research.

a.   Roy, A., Krebs, H.I., Williams, D.J., Bever, C.T., Forrester, L.W., Macko, R.M, Hogan, N. (2009). Robot-aided Neurorehabilitation: A Robot for Ankle Rehabilitation. IEEE Trans Robotics, 25:569-582.

b.   Khanna, I.,Roy, A., Rodgers, M.M., Macko, R.M., Krebs, H.I., Forrester, L.W. (2010). Effects of Unilateral Robotic Limb Loading on Gait Characteristics in Subjects with Chronic Stroke. J Neuro Engineering and Rehabilitation, 7:23.

Robotics as a Motor Learning Platform.My studies have utilized modular robotics (ABOT) for non-ambulatory (seated), isolated ankle training in subjects with hemiparetic deficits resulting from stroke. Coupling the ABOT with a novel visually evoked and visually guided task has proven to be a potent ML paradigm. Findings from multiple studies have shown that ABOT therapy even in a non-task specific context results in generalized benefits beyond the ankle, contrary to long-held model of task-specificity of training. Seated, computer-video interfaced ABOT therapy has demonstrated, in both chronic and early sub-acute phases of stroke, improved paretic ankle motor control (speed, smoothness, accuracy of ankle targeting) and reduced ankle impairments (range of motion, passive stiffness) that carry-over to gains in independent walking function evidenced by higher speeds (20%), improved spatial-temporal gait symmetry, more stable inter-limb dynamic weight transfer, as well as more stable standing balance. In this context, my work also shows that even a single, initial exposure to seated ABOT therapy session (~1 hour) leads to rapid motor adaptations in measures of paretic ankle motor control, which are retained at 48-hour retest. This paradigm has also enabled for the first time, investigation of reward integrated ABOT therapy on neurophysiologic adaptations associated with motor learning, in chronic stroke. High reward subjects exhibited faster learning in paretic ankle motor control concomitant with reduced EEG-measured contra- and frontal-parietal coherence and reduced left-temporal spectral power, suggesting that combining explicit rewards with ABOT therapy accelerates motor learning for restoring mobility. Finally, the versatility of this protocol can be gauged from its use in disease conditions beyond stroke, such as multiple sclerosis (MS). Robot-aided ankle targeting show improved paretic ankle motor control measures similar to stroke, and are accompanied by changes in EEG measures of activation and networking, suggesting that this platform may be used to advance our understanding of the neuro-physiological mechanisms in motor learning-based recovery in persons with ankle motor deficits secondary to MS. This body of work provided the first-ever evidence of (a) Benefits of ABOT therapy in improving ankle function, (b) Translation of localized (ankle) to generalized whole-body functional benefits, (c) Short-term motor learning at the paretic ankle from a single bout of ABOT therapy, and (d) Reward-based effects of ABOT therapy in improving cortical dynamics. I have served as the primary and co-investigator for this research.

a.   Forrester, L.W., Roy, A., Krebs, H.I., Macko, R.F. (2011). Ankle training with a robotic device improves hemiparetic gait after a stroke. Neurorehabil Neural Rep, 25:369-377.
b.   Forrester, L.W., Roy, A., Krywonis, A., Kehs, G., Krebs, H.I., Macko, R.F. (2014). Modular ankle robotics in early sub-acute stroke: A randomized controlled pilot study, Neurorehabil Neural Rep, 28:678-687.
c.   Roy, A., Forrester, L.W., Macko, R.F. (2011). Short-term ankle motor performance with ankle robotics training in chronic hemiparetic stroke. J Rehabil Res Dev, 48:417-430.
d.   Goodman, R.N., Rietschel, J.C., Roy, A., Jung, B.C., Diaz, J., Macko, R.F., Forrester, L.W. (2014). Increased motivation during ankle robotic training enhances motor control and cortical efficiency in chronic hemiparetic stroke. J Rehabil Res Dev, 51:213-228.

Adaptive Co-Robotics for Locomotor Rehabilitation. Task-oriented robotic gait therapy has not yet shown efficacy and has been negatively viewed by the AHA/ASA, DOD, and VA. This view is not without merit – despite contemporary lower-limb exoskeletons being technologically sophisticated, their design and underlying operation have generally not been aligned with fundamental tenets of ML. Further, hemiparetic gait with ankle motor deficits such as foot drop are considered refractory to therapeutic isolation (hence, treatment), managed with devices (cane or other assistive device, ankle foot orthotics-AFO) that confer safety when worn, but do not produce recovery; instead, reinforce abnormal compensations, or learned disuse. To integrate robotics into task-oriented locomotor practice safely and effectively necessitated a completely new approach and technological breakthroughs. My research pioneered the field of adaptive co-robotics for neuro-rehabilitation, defined here as co-operative interactive learning between a patient and a robotic device. Adaptive co-robotics in TM walking exercise (TMR) can safely accommodate the heterogeneity of hemiparetic gait, dynamically shape robotic actuation with the patient’s capacity and recovery to individualize functional locomotor learning for improving and promoting safe and efficient mobility. Further, adaptive controllers can operate in different modes to meet the changing needs and deficit profiles of patients, while providing high-volume practice at varying degrees of assist in a variety of ML contexts (goal-oriented, progressive challenge, performance feedback). At the heart of adaptive ankle co-robotics is my invention of an event-triggered, gait sub-task control system that enables precise timing of robotic assistance via real-time information from bilateral micro-switches embedded inside patient’s shoes to key functional gait deficits, to prevent human-robot destabilization and ensure safety. Accordingly, the ABOT is controlled and actuated in a manner to deliver ankle torques during critical instants, each with unique functional needs (“deficit-adjusted” approach): Concentric plantar-flexor torque to enable terminal stance push-off propulsion and concentric dorsi-flexor torque to facilitate foot-floor swing clearance. Robotic assist levels are matched to entry deficit severity and systematically progressed based on performance and data-driven mathematical models, a major advancement in robotic gait therapy that now enables truly individualized robotic prescriptions to assure individualized optimality in neuromotor functional performance (e.g. maximize floor clearance for those with foot drop). Our clinical findings show that 6-week TMR reverses foot drop, restores push-off, and corrects foot landing, with durable effects at 6 weeks after cessation of training, causing 85% users to self-discard their AFOs. No therapies have durably reversed hemiparetic gait deficits and improved gait patterning to an extent as TMR, leading to unprecedented benefits in independent mobility function. I am the inventor of the adaptive co-robotics control system, robotics progression algorithms/formulae, and have served as co-investigator for the clinical study.

a.   Forrester LW, Roy A, Hafer-Macko C, Krebs HI, Macko RF. Task-Specific Ankle Robotics Gait Training After Stroke: A Randomized Pilot Study,” J NeuroEngineering & Rehabilitation, 13:51, 2016.
b.   Roy A, Forrester LW, Macko F. Methods and Apparatus For Providing Deficit-Adjusted Adaptive Robotic Assistance For Gait Training. US Patent Pending, 14/549,370.
c.   Roy, A., Krebs, H.I., Barton, J.E., Macko, R.F., Forrester, L.W. (2013). Anklebot-Assisted Locomotor Training After Stroke: A Novel Deficit-Adjusted Control Approach,” In: Proc. IEEE Int. Conf. on Robotics and Automation, 2167-2174.
d.   Roy, A., Krebs, H.I., Macko, N.R., Macko, R.F., Forrester, L.W. (2014). Facilitating Push-Off Propulsion: A Biomechanical Model for Ankle Robotics Assistance for Plantarflexion Gait Training, In: Proc. IEEE Int. Conf. on Biomedical Robotics and Biomechatronics, 656-663.

Robotics as a Clinical Measurement Instrument.Adequate ankle stiffness is critical to control forward body momentum and for “shock absorption” during walking. Hence, an accurate estimate of ankle stiffness is important for locomotor rehabilitation, potentially providing a measure of recovery and a quantitative basis to design treatment protocols. My study was the first ever to use ankle robotics to measure passive ankle stiffness in the frontal plane (inversion-eversion) in chronic hemiparetic stroke, of significant value given that frontal-plane mechanics is important in the maintenance of balance and prevention of injury under a variety of conditions. We found that frontal-plane passive stiffness of the paretic ankle is highly anisotropic (higher in inversion and lower in eversion) and significantly higher in chronic stroke than in age-matched healthy adults of a similar cohort. I conceived this study and served as the primary investigator.

a.   Roy, A., Krebs, H.I., Patterson, S.L., Bever, C.T., Forrester, L.W., Macko, R.F., Hogan. (2011). Measurement of passive ankle stiffness in subjects with chronic hemiparesis using a novel ankle robot. Journal of Neurophysiology, 105:2132-2149.
b.   Roy, A., Krebs, H.I., Patterson, S.L., Judkins, T.N., Khanna, I., Forrester, L.W., Macko, R.F, Hogan, N. (2007). Measurement of human ankle stiffness using the anklebot. In: Proceedings of the IEEE International Conference on Rehabilitation Robotics, 356-63.
c.   Roy, A., Forrester, L.W., Macko, R.F., Krebs, H.I. (2013). Changes in passive ankle stiffness and its effects on gait function in people with chronic stroke. J Rehabil. Res Dev, 50:555-72.

Telerehabilitation Technologies:Our exercise programs have evolved to emphasize aerobic fitness and muscular endurance within the framework of movement quality practice to facilitate functional mobility and cardiovascular health for stroke survivors. This approach has been successfully used to safely and effectively progress complexity and intensity across a broad range of stroke deficit profiles. However, most of our protocols targeting adaptations in diverse outcome categories have been center-based, in part due to absence of scalable technologies, needed to safely and effectively administer home-based therapies. To remedy this, I have developed (co-inventor) a novel telerehabilitation tool (IVET) is a web/tablet-based application that “delivers” customized home exercise prescriptions emphasizing multi-segmental motor control and balance to promote long-term improvements in mobility, activities of daily life (ADLs) and fall risk. IVET is designed to improve safety, compliance, quality and sustainability of movement exercise practice in the home and delivers a high quality exercise stimulus using fewer resources in a way that is more convenient for participants. New published data show that users have high enthusiasm for IVET and are able to learn its features quickly and safely. Ongoing enhancements include integration of off-the-shelve wearables with IVET to capture physiological and movement measures.

a.   Hafer-Macko C, Naumes J, Macko RF, Roy, A. Technology Platform for Tele-Rehabilitation Implementation in Mysathenia Gravis at the Point-Of-Care. In: Proceedings of the IEEE-NIH Special Topics Conf. Health Care Innovations & Point-of-Care Technologies, 2016 (In-Press).
Macko RF, Forrester T, Francis P, Nelson G, Hafer-Macko C, Roy A. Interactive Video Exercise Tele-Rehabilitation (IVET) for Stroke Care in Jamaica. In: Proceedings of IEEE-NIH Special Topics Conf. Health Care Innovations & Point-of-Care Technologies, 2016 (In-Press).

Complete List of Published Work in MyBibliography: http://www.ncbi.nlm.nih.gov/sites/myncbi/163RDNLat4V5P/bibliography/49505203/public/?sort=date&direction=ascending.

D.             Research Support

Current Research Support

1. VA RR&D Merit 1I01RX001699-01A1                               Roy (Co-PI)                                                            01/01/15 - 12/31/18

Adaptive ankle robot control system to reduce foot-drop in chronic stroke

The goal of this randomized controlled trial to compare the effectiveness of 6 weeks of treadmill aerobic exercise with and without ankle robotics as a means to improve gait and balance functions in chronic stroke.

2. VA RR&D B9215-C                                   Wittenberg (PI)                                                            01/01/11 - 12/31/17                

Center for task-oriented exercise & robotics in neurological disease

The focus of this center is to optimize functional recovery in individuals with mobility impairment due to stroke and other neurological conditions using a multi-systems approach investigating models of task-oriented exercise, robotics, and behavioral modification.

Completed Research Support

ORH N05-FY14Q1-S2-P01050                                              Macko (PI)                                                            10/01/15 - 09/30/16

Interactive Video Exercise Tele-rehabilitation (IVET)

This grant uses a “train-the-trainer” model for supervised group stroke exercise classes to develop a home Interactive Video Exercise Tele-rehabilitation (IVET) program embedded on smart devices.

ORH N05-FY15Q1-S1-P01504                                              Macko (PI)                                                            10/01/14 - 09/30/16

Exercise + MOVE for Chronic Disease Management of Rural Veterans

 Abell Foundation Award                                                                    Roy, Macko (Co-PI)                                  07/01/15 - 06/30/16

Bio-Based Software Engine for Adaptive Control of Modular Robots for Clinical Neuro-Rehabilitation

The goal of this project was to develop a portable ankle robot with a deficit adjusted control system to enable home-based ankle therapy for chronic stroke survivors.

VA RR&D Merit Pilot 1-IO1-RX000592-01                 Forrester (PI)                                                                     07/01/12 - 06/30/14

Developing a brain-machine interface for an ankle robot

This goal of this pilot project was to use non-invasive EEG to decode ankle movements performed in an ankle robot, first in a group of older nondisabled persons and then in a group of chronic stroke subjects.

VA Merit Review Pilot 1-I01-RX000351-01                      Forrester (PI)                                                              07/01/11 - 05/30/14

Ankle robotics training after stroke: effects on gait and balance

This study investigated two modalities (seated vs. treadmill) for using an impedance controlled ankle robot to improve gait and balance function among stroke survivors with chronic lower extremity hemiparesis.

Rehabilitation, Robotics, Recovery, Motor Control, Stroke, Biomechanics

1. Roy, A., Krebs, H.I., Williams, D.J., Bever, C.T., Forrester, L.W., Macko, R.M, Hogan, N. (2009). Robot-aided Neurorehabilitation: A Robot for Ankle Rehabilitation. IEEE Trans Robotics, 25:569-582.
2. Roy, A., Krebs, H.I., Patterson, S.L., Bever, C.T., Forrester, L.W., Macko, R.F., Hogan. (2011). Measurement of passive ankle stiffness in subjects with chronic hemiparesis using a novel ankle robot. Journal of Neurophysiology, 105:2132-2149.
3. Forrester LW, Roy A, Hafer-Macko C, Krebs HI, Macko RF. Task-Specific Ankle Robotics Gait Training After Stroke: A Randomized Pilot Study,” J NeuroEngineering & Rehabilitation, 13:51, 2016.
4. Forrester, L.W., Roy, A., Krebs, H.I., Macko, R.F. (2011). Ankle training with a robotic device improves hemiparetic gait after a stroke. Neurorehabil Neural Rep, 25:369-377.