SCH: Model-informed patient-specific rehabilitation using robotics and neuromuscular modeling

PROJECT DESCRIPTION
1 Motivation
Stroke is a leading cause of long-term disability in the United States. Stroke survivors now constitute
around 3% of the over-20 population, with 50% of stroke-affected subjects left with impaired propulsion
on the paretic side, resulting in asymmetric movement and compromised balance [1]. The hemiparetic
gait observed in many individuals post-stroke is slower and more metabolically expensive than in healthy
individuals [2–6], and is a primary contributor to reduced community participation and quality of life [7–11].
Contemporary approaches to gait training are based on repetitive therapy often conducted on treadmills [12],
with variants including the combination of human or robotic assistance [13], body weight support [14], and
functional electrical stimulation [15].
Robotic intervention enables systematic and accurate modulation of joint-level variables, such as assis-
tance torques and joint angles/velocities. Robotics is an intriguing tool for gait training, but the capability
of using robots as tools to support locomotor learning for rehabilitation purposes has not yet been fully
demonstrated. Earlier implementations of robot-aided gait rehabilitation provided non-convincing or nega-
tive results [13, 16], as extensively quantified in a meta-analysis [17]. Currently, the effects of robot-aided
gait training in stroke have yet to exceed those achieved with conventional therapy methods [17].
We speculate that such limitations are mostly imputable to the controllers used for robot-aided gait train-
ing. The majority of robotic devices, designed specifically to rehabilitate gait, utilize one of the various
controller forms (e.g., force control, position control, or impedance control), and controller update methods
(e.g., assist-as-needed control, inter-limb coordination, or finite state machine), to ultimately promote spe-
cific features of gait kinematics [18]. The limited efficacy of these methods could be due to their lack of
targeting specific functional mechanisms of gait, which are only partially described by joint kinematics.
From an extremely reductionist perspective, walking is pushing ones' center of mass in a desired direction
while not falling. Fundamentally, walking involves three main sub-tasks: propulsion, limb advancement, and
balance [19]. Of these components, limb advancement may be based on kinematic control, but is the least
energetically demanding. Instead, the sub-tasks of propulsion and balance require precise neuromuscular
coordination, and specifically mediation of the interaction forces between the walker and ground. Despite
their fundamental importance, there have been very little efforts in rehabilitation robotics in developing
robot-aided methods to study and/or train propulsion and balance in post-stroke rehabilitation.
The overarching goal of the proposed research
Measure
is to advance the science of therapeutic engineering Walking Surfac~
for gait by identifying optimal robot interventions " .,
and therapies with specific functional outcomes. Stiffne.ss Perturbations Model
Those interventions will be developed using a new
modeling approach to target enhanced propulsion Evaluate
and balance in stroke survivors. The sense-plan-
act paradigm in robotics will be applied in a unique
way to robot-assisted model-informed rehabilita-
tion research. The proposed framework will inte-
grate robotic solutions that will allow the creation
of comprehensive models of sensorimotor mecha-
nisms of gait. These models will then inform a set
of interventions to stroke survivors, the outcomes
of which will be fed back to the developed models
to uncover and suggest novel patient-specific train-
ing strategies. The proposed approach will enable Figure 1: Proposed integrative research framework fol-
a better understanding of essential mechanisms lowing the sense-plan-act paradigm in robotics.
responsible for walking and lead to the design of
optimized and personalized post-stroke rehabilitation strategies. The overall framework of the proposed
49

Grant Number: 5R01HD111071-04
NIH Institute/Center: NIH
Principal Investigator: Panagiotis Artemiadis