Rehabilitation Robotics

Major factors causing disabilities world-wide

• Neonatal Nutrition >> Premature Births >> Cerebral Palsy, Autism, Down Syndrome, etc.

• High Blood Pressure >> Strokes

• HIV >> Dementia, Strokes

• COVID-19 >> Chronic fatigue, Cognitive Impairment

Therapy Robots

• Treat neurological disorders such as stroke and cerebral palsy

  • Function to automate and deliver autonomous or semi-autonomous therapy for the arm (or leg or joint)

  • Function to assess the level of disability and impairment remaining in a limb, arm, or leg

  • Outcome >>> reducing motor impairment, increasing function, and driving brain re-organization

• Typically function in clinics or supervised settings

Assistive Robots = Service Robots in Rehabilitation/Medical Settings

• Replace other functions or activities, or things (e.g., surveillance robots)

• Replace a lost limb (e.g., prosthetics)

• Replace the function of a paralyzed limb and do tasks instead of the limb (e.g., a wheelchair robot)

Observing Human-Human

patients w/ therapists

• study of therapist movement gives us info on

  • —> how we model the kinematics of those movements

  • —> how we develop desired trajectories

• look at the forces of interaction that therapists exert on their patients

  • —> how much force should the robot exert?

• look at the therapist's role and behavior

  • —> understand the physical cues, verbal cues, and administrative cues they are using

Capturing Roles and Cues

encode therapist activity to understand what behavior, rules, and scenarios we might see in a typical therapy session between the therapist and the patient

• Multimedia Video Analysis Software: MVTA

• 8 Videos Coded independently by 2 therapists

Patient-Therapist Dyads

therapist took on 3 behaviors

• demonstrators (while the patient observed the action)

• observer (while the patient becomes the performer)

• helper (patient is performing, but now with the therapist’s assistance)

  • at the patient’s request/or if they observe something

Therapist >> Robot

• Ideally, the robot should take on three roles as demonstrator, observer, and helper, and co-act with the patient

• Helper role is often seen in hands-on effector THERAPY ROBOTS (e.g., ADLER, Theradrive)

• Demonstrator and Observer Roles are often found in ASSISTIVE ROBOTS or SERVICE ROBOTS (e.g., Nao)

  • i.e., look at patients performing tasks —> provide feedback —> monitor

• Fluid transitioning from contact to non-contact with a patient is not often done due to huge safety concerns about soft and hard impacts

Helper role example

• permanently attached to the person

TeleMonitoring/TeleTherapy: Rehabilitation Tasks

• two rehabilitation tasks:

  • magazine stacking

  • and free movement of the stroke-affected limb

• example of a robot that is in the social role of observing

• patient is wearing a sensor

• robot’s role is to

  • observe/monitor the patient to see if they are complying with the task and

  • get information about the level of impairment as they move on

Baxter: Elder Exercise

• Demonstrator/Observer roles >>

• using camera systems and sensora

• as the robot tries to get the person to interact with it, it can know whether the person interacted or not

BAXTER: Therapy Assistant

• Observer/Helper roles >>

• sensory on the patient’s arm

• monitoring the velocity of the movement of the patient’s arm

• if there is an indication that the person is not acting within some prescribed bounds, the robot will come in and act

social robot on top of a telepresence platform

• Demonstrator/Observer role >>

• camera allows the the robot to observe the actions of the patient that it’s interacting with —> social robot can demonstrate activities that the patient can then comply with

Goal: Robot shares control with the therapist

philosophy: robots, whether they’re a therapy robot or a social robot, should act together w/ the therapist

• sharing of control

• therapist should program the robot in an appropriate way for their patient

Long-term impairments persist after stroke

Cognitive impairments (46-61%)

• language, visuospatial function, executive function, memory

Motor dysfunction (65%)

• muscle weakness, paralysis, spasticity, pain

Psychosocial issues (20-33%)

• depression, anxiety, general distress, isolation

Activities of daily living impairments (25-74%)

• bathing, eating, toileting, dressing, transfer, and mobility

Upper Arm Impairment

• Spasticity

• Muscle Weakness

• Impaired Interlimb and Intralimb Coordination

• Reduced Joint Range of Motion

• Impaired Posture

• Increased Trunk Compensation

• Inability to perform isolated arm movement

  • Flexor or Extensor Synergy

Upper Arm Stroke Rehabilitation

• Practice Weight-Bearing

• Practice Real Activities

• Practice Moving Joints

• Sensory Facilitation

• Ex: Specific Techniques

  • Constraint Induced Therapy

Design Choices for RAT (Robot Assisted Therapy system)

figure out what these things should be; variables

• what controls should you use?

• how do you motivate a patient?

• how do you customize it so that it’s different for individuals?

• what type of feedback?

• how will we make it learn?

• what type of task?

==> must be useful in increasing engagement, allow for intense therapy, result in neuroplasticity, and carry over into the world

Mechanical Device and Sensors

• Exoskeletons

• End-effector

• Single joint (e.g., only acting across the elbow)

• Multiple joints (e.g., acting across the shoulder and the elbow)

• Degrees of Freedom (i.e., the # of motors that they may have; the # of independent movements in the system that they may have)

• Planar

• 3D

• Pneumatic, etc.

(advantages to) Therapy Robots: Upper Limb

• Automate traditional therapy treatments

• Enable semi-autonomous training

  • once you program them, you can have them go

• Provide consistent, repeatable, intensive training

  • can stress your patients a little bit more in terms of repetition; e.g., doing 1000 reps

• Provide objective measures of recovery

  • quantitative ways of knowing that what you’re doing works

End-effector Robot Assisted Therapy

• attached towards the forearm/wrist complex

• can come in different configurations to support the UE (e.g., joystick)

• usually connected to some type of gaming activity

Example List of End-Effector Therapy Robots

• Cost of these robots varies from $3K (CR-2 Haptic) to $275k (G-EO)

ExoskeletonsRobot Assisted Therapy

• covering the upper arm in its entirety

• often are not just attached to the wrist; also attached at the elbow

• trying to move individual joints

Example list of some popular exoskeletons that exist that you can purchase out of the market

RUPERT IV

one of the earliest examples of exoskeletons; early designs were very heavy and bulky (but the premise of these were very encouraging)

• Device encourages the retaining of the coordination of real-life activities

• Three main links to support Shoulder, Elbow, and Wrist Movements Robot

• RUPERT IV utilizes pneumatic McKibben artificial muscle actuators

• The power supply and the control system are located in the backpack part of the robot

Myomo

less bulky exoskeletons that people can take home

MIT-MANUS/InMotion

2D - HELPER ROBOT= designed to focus on exercise in the plane (mostly upper extremity- elbow/shoulder activities)

• SCARA Manipulator

• 2 Degrees of Freedom (Planar robot)

• Backdrivable

• Low Inertia

  • you don’t feel a lot of resistance when you’re moving it

• Reaching practice

  • Point-to-Point Tracking

• Impedance Control

  • we can craft how it feels: the mass, the spring, the damping experience as you use the robot

gaming environment used to develop a training modality (a particular exercise)

• as you move around the dashed lines, you are mainly exercising the elbow

• as you move along the closer dashed lines, you are mainly exercising the shoulder

• as you move along the solid lines, you are doing more inter-limb and coordinated type reaching

Minimum Jerk Model

model that allows us to command and create trajectories

example of someone using the MIT MANUS (end-effector robot, as it is attached to the end-effector)

• there is a target that is appearing on the screen

• the person is asked to move to those targets

• how do we control the experience at the robot?

  • controlling the stiffness and the damping experience at the robot

    • patient is severely impaired —> robot should be stronger and stiffer, so that the person is guided along the path

    • patient is a lot milder in terms of their presentation —> robot should give little help, so that the person is doing it themselves; or, provide resistance/increase damping, so that the person has a more challenging experience

Examples of the measures that can be extracted from the actions that the patients are doing with the robot; can have quantitative info about…

• velocity

• acuracy

• efficiency

• smoothness

• time

***these measures have been shown to be sensitive to reduction in impairment

InMotion Therapy Robot

designed to have 1 therapist set up the system and work with the patient; advantages:

• can allow lots of repetition

• therapist can wander off if they need to (e.g., to look at another patient)

Robot-Assisted Therapy (RAT) VA Clinical Trial Results

clinical trial that compared robot therapy to non-robot therapy in 2 ways: standard therapy and intensive therapy; looked at moderate-severe functioning stroke survivors with UE impairment

• Robot Therapy:

  • 36, 1-hour, high-intensity therapy with the In Motion robot (MIT-Manus paradigm with horizontal, vertical, wrist, and hand modules)

  • ***findings: did not improve significantly more than non-robot control groups, BUT it got better as time goes on

  • ***findings: cost of robot therapy was comparable to non-robot therapies, and the price is not too bad!

• Non-Robot Therapy:

  • Standard Therapy and Intensive (same dosage)

• Subjects:

  • Moderate to severe functioning stroke survivors with upper limb impairment for at least 6 months and with lesions due to single and multiple strokes

• Robot therapy did not improve significantly more than non-robot control groups of usual care or intensive therapy, but had a modest improvement over 36 weeks

• The cost of robot therapy was comparable to the non-robot therapies (12 weeks: $9,977 RT versus $8,269 Non RT, and 36 weeks: $15,562 RT versus $15,605 and $14,343 for non RT)

***could be good in environments that don’t have a lot of clinicians, but a lot of patients that need help; robots could help deliver therapy

image

• systems have also been used for kids w/ cerebral palsy; similar results

ADLER

3D (3 degrees of freedom)- Example

HELPER ROBOT

robot with motions in 3-dimensional space, so that you can move in and out of the plane; designed for people to practice a task in their workspace, but also were connected to the robot with their hands free, so that they can interact with objects within the environment

• idea is to be more task-oriented

Task-Oriented Robot Assisted Therapy

• Design robots to support real tasks/ADLs practice

• Design to maintain engagement/motivation

• Design is patient-centered

• Design is integrated with the clinical environment

• Design control strategies that promote effective and adaptive therapy

ADLER: Activities Daily Living Exercise Robot

built on a desktop-type environment, and the different end effectors allowed the hand to be free

• HapticMaster Robot

  • RPP Positioning Robot

  • 6 Degrees of Freedom (3 active, 3 passive)

  • Position sensing at all DOF

  • 3-axis force sensing at the end-effector

  • Admittance control >> stronger

• Current Functionality

  • Administers functional unilateral robotic therapies to stroke subjects

  • Training of real-life functional tasks

  • FES Glove Expansion for Lower Functioning Subjects to do ADL Training

    • e.g., opening/closing activities for those who had a more impaired hand

ADLER: Task Examples (3DOF Example)

• Use real tasks that reflect ADLs

• Adaptable to many motivating and engaging task environments

Supervisory Structure

software that allowed therapists to design programs ahead of time and implement them, so that the robot could actually complete the task

• HERALD Software

  • HapticAPI programming environment

  • Crystal Space v0.98

  • PC-based Environment

  • The GUI shows a one-to-one mapping of the environment and the trajectory

• Programming Activities

  • On the fly, using start, end, and multiple via points to define the task

  • By uploading data from a file with real or pre-defined trajectories

  • Various models of task trajectories are programmed

  • Moves the subject in Training modes

Trajectory Planning for Reaching

to map different trajectories (e.g., reach to the cup; bring cup to mouth), had to develop different types of models (versions of the minimum jerk paradigm- the original model)

• modified ways to incorporate real tasks

Trajectory Planning ADLER (3D)

Example of projection planning for a real task within the environment

Reaching Training

• Current Training Strategies

  • Form (Guiding Forces); for people who are very impaired

    • Actively follow a desired path

      • supports movement along the path, so there would be reinforcement of movement

    • Experience forces if they deviate

    • Graded resistance along the path

    • goal: tracking the original movement of the robot

  • Function (Minimal Guiding Forces) (New); for people who are less impaired

    • Aim for the target point

      • supports target reaching

    • Permitted to deviate

    • Attracted to the target

    • Graded resistance along the path

    • goal: get to the target

• Resist

• Normal (No forces): Evaluation mode

Observed Kinematic Changes

tested Adler with patients; after 4 weeks of interacting with the system, 3x/week for ~1hr

== system can promote recovery and reduction in impairment for a drinking task (i.e., as a result of the treatment, the person was able to move across the table + move out of the plane a little bit better)

Most Frequent Observations:

• 1. Increased range of motion of the elbow and shoulder flexion and extension, which translates into the ability to lift the arm higher and extend the arm further

Most Frequent Kinematic Changes Seen:

• 1. Increased ROM of the elbow and shoulder flexion and extension

• 2. Increased speed and strength, and less time to do a mastered task

also saw that the system was useful for children in a more VR-type environment

• Integrated VR with HapticMaster Robot for Therapy; 3D of freedom

• Exercises were virtual games that adapted to motor performance

• 2 CP children tested in 9 hours of therapy in 3 weeks

• Improvements in joint ROM and reach kinematics

• Follow-up study with 9 children with CP with robot therapy alone or with other therapies

• Clinically significant improvements in active shoulder abduction and flexion as well as forearm supination.

reviews + pros of rehabilitation robotics

original reviews

• upper limb robot therapy is able to reduce motor impairment

pros of rehabilitation robotics

• As effective as high-intensity physical therapy

• Repeatable

• Adaptive

• Enable semi-autonomous training

• Provide consistent training

• Provide objective measures of recovery

• Reduce Impairment

• Increase Activity

Key Findings: Robot Assisted Therapy by ICF

Health Condition (CNS damage)

• Body Functions & Structure (Impairment)

  • Increase strength

  • Increase motor control

  • Reduce spasticity, muscle tone

  • Improve Interjoint movement

  • Training specific to target areas

• Activity (Disability)

  • •Train mainly reaching tasks, hands or grasping

  • Inconsistent carryover to real activities of daily living ADLs, but still have effects overall

• Participation (Handicap)

  • Some help in the long term – Exoskeletons

  • Not affordable for significant help at home

The economic cost of robotic rehabilitation for adult stroke patients: a systematic review

Cost Minimization: Studies compared the cost of providing robotic intervention against the cost of providing dose matched-therapy

• robots are favorable in terms of their outcomes as compared with the standard of care

• in general, robots are worth pursuing

Rehabilitation Robotics Market

growing market!

Rehabilitation Robots for Neurorehabilitation in High-, Low-, and Middle-Income Countries Current Practice, Barriers, and Future Directions

growing number of companies in Europe, North America, and Asia- high-income countries; much less in South America and Africa

Problem with current systems for LMICs

design issues:

• Mechanical complexity

• Huge sizes and masses

• High Costs

REHABILITATION 2030- a call for action

there is a global need for rehabilitation and the availability of services

• World Health Organization passed a law saying that all their member countries should prioritize rehabilitation

Rehab Landscape in HIC (high-income countries)

even in HIC, rehabilitation is increasingly occurring outside of a hospital environment in acute care

• 75% discharge with residual impairments

• 40% go home without post-acute care

Rehab taking place in the community

• At home with nursing care or a home health agency

• Nursing home

• Day-care or all-inclusive care facility (PACE)

• Assisted Living Facility

In LMICs (low and middle income countries), these issues are compounded

Rehabilitation in Low and Middle Income Countries

• Rehabilitation is often an afterthought, and the focus is on disease cure

• Rehabilitation access is low (below WHO guidelines)

• Rehabilitation care is not as specialized, and many are not trained to deliver it

  • Skilled therapists and physiatrists are often not available in large numbers inside or outside of cities;

  • Low doctor-patient ratios, e.g., in Malawi 1:88,000

  • Low therapist-patient ratios, e.g., in Botswana – a hospital ~15:800 beds

• Rehabilitation technology is not available or is not of the same quality

Some Key Guidelines

• Rehabilitation Robots In LMICs Must Be Affordable

• Rehabilitation Robots In LMICs Must Be Multipurpose

• Rehabilitation Robots In LMICs Must Be Effective

• Rehabilitation Robots In Low-Resource Settings Must Promote Community-based Rehabilitation

• Rehabilitation Robots In LMICs Must Be Appropriate

• Sustaining Rehabilitation Robots In LMICs Depends On Capacity Building

Affordability

“Robots are not affordable … but despite their high cost … a likely advantage is that automated interventions like robotic therapies require minimal input from rehabilitation professionals in terms of time and efforts.”

• affordability, noun= the extent to which something is affordable, as measured by its cost relative to the amount that the purchaser can pay

Potential strategies to reduce cost and be multipurpose?

• Use low-cost robotics/mechatronic systems

• Use cheaper materials – 3D printing/soft robots/found objects

• Use a system of low-cost robot units that can reconfigure

• Use multiple-use systems

• Use local manufacturing and resources

Mexican Theradrive

a low-cost training system

• 6-station training gym that had robotic and technology assistant systems in the gym

• ran a study confronting the standard of care vs people participating in the gym

== saw that it was cost-effective; able to maximize the impact of the therapies without diminishing the outcomes of therapies

Rehab C.A.R.E.S. Gym

community based scenario of multiple robots together that could then be placed in different areas

Haptic Theradrive (TD-3)

also developed a new robot that was stronger and able to support people with more severe impairment