Abracadabra

Robotics Friend For Physiotherapy After Stroke

Dr. Marina Fridin, CTO

May I help you?

You? Really?

Engineering Conference, Novel Technology for people with special needs,

Ruppin Academic Center, 2014

Definition of Stroke

• Sudden brain damage• Lack of blood flow to the brain caused by a

clot or rupture of a blood vessel

Ischemic = Clot (makes up approximately 87% of all strokes)

Hemorrhagic = Bleed- Bleeding around brain- Bleeding into brain

Embolic Thrombotic

Social Assistive Robotics (SAR), scientific concept

- SAR is the class of robots that provide various types of assistance to various vulnerable populations primarily through social, rather than physical interaction.

• POPULATIONS:–Elderly:

• Post- stroke rehabilitation, Matarić, et al., 2007• Alzheimer’s disease , Tapus et..al. ,2009• Hospital delivery robot, Mutlu & Forlizzi ,2008• Nursing home residents , Wada et.al. , 2004

–Children, mainly for children with autism spectrum disorders (ASD)

• Scassellati, et al. 2007. • Robins, et al., 2005.

Factors that influence recovery following a stroke

Abracadabra can not influence

Abracadabra could influence

• Time passed before medical intervention is initiated

• What part of the brain was affected• The size of the area affected

• The patient’s age• The patient’s fitness level before the

stroke• Patient’s premorbid cognitive level

• Additional medical problems

•Patient’s emotional state/motivation level

•Family support•Environmental and social influences•Time passed since stroke occurred

•Amount of therapy received (especially in the first 12 months)

! Each stroke is different therefore it is difficult to predict the amount of recovery that will occur in the affected side.

Principles of design Stroke Rehabilitation Procedure of Abracadabra

Interdisciplinary Team participate in the design and implementation of the system

Uses Learning Theory:– Graded Levels of Task Difficulty– Opportunities for Repetition of Skill

Performance– Professional Supervision and Feedback– “Protected Practice”

Examples of Stroke Rehabilitation Interventions ABRACADABRA could participate

Functional Skills Training−Personal Care Skills−Mobility Activities−Instrumental Activities of Daily Living

Therapeutic Exercises−Flexibility−Strength−Coordination−Fitness

Visual fields – treatment−Increase awareness−Compensatory oculomotor strategies−Audio-visual stimulation−Compensatory head movement−Reading and writing

Speech therapy−Communication/language −Speech−Voice Quality−Fluency (stuttering)−Cognition/neglect−Swallowing

Cognition −Orientation (who, when, where, why)−Attention−Memory (usually with immediate or short term memory)−Problem solving−Reasoning−Insight/safety awareness

The system

• 2 versions of the robot: fully functional and restricted

• Virtual version of the robot• Connected Devices, including sensory

system for colleting of the patient data (motor/cognitive performances etc.)

• Connections to the therapeutically devices (including Virtual Reality and computer games)

• Computer-based system for rehabilitation team ( including reports of patient progress, definition of the tasks ect.)

Abracadabra for different Rehabilitation Services

The robot with full functionality:Rehabilitation unit in the hospital

Home with outpatient therapy

Long-term care facility

Community-based programs

Restricted version of the robot or it’s virtual

agentHome-bound therapy (tele-medicine)

Hello Darling, how are you today?

I can not get it? Could you move it

more forward for me?

Motor Actions: Example

Hello Robot!

Patient Personal Information

Task DefinitionRobot Activation

I am so parched, could you please

give a cup of water?

Report Next Task

Emotional StatusPerformance

Motor learning approach

- Based on the principles of

•Repetition

•Adaptation

•Appropriate feedback

•Random practice and

•Enriched environment

Particularly adjusted to the learning stage and task component

Skill acquisitionThe patient: Who? The task: What? The context: Where?

AgeExperienceMotivationMemoryAbility

Discrete/Serial/Cont.Closed/OpenGross/FineAccuracy kindProgram/Plan

ClinicalHomePresence of othersTask variability

Stage of learning

Preparation Presentation Structure

GoalsTransferContextPerformance measures

InstructionsDemonstrationGuidanceSimulatorsPart practiceMental practice

Random/Blocked

Random/Varied

Massed/Distributed

Feedback

KR/KP

Descript./Prescript

Type of feedback

Amount

Frequency

ABRACADABRA

Body Structure & Functions

Activity Participation

EnvironmentalFactors

PersonalFactors

Therapist AbracadabraHealth condition (disorder or disease)

Hands on

Conceptual Interference

Environment Taxonomy

Learning Algorithm

Hands off

Human-Robot Interaction

Motor GamesMotor Actions

Adaptation to personality, mood and motor performance

Motivation: feedback, mirroring effectTeam

Decomposition

The scheme of the ABRACADABRA modules and data flow

Low-level Perception• Kinect: Motion processing • Robot’s Video: High color saturation filters Skin-color filters Edge detection Disparity computation• Robot’s Sensors: Data filtering

Motor module• Visual-motor skills • Manipulation skills (reaching/ grasping)• Expressive skills (bodily/facial/vocalizations)

Attention module• Attention on a child with fastest/slowest reactions • Attention on a child defined by physiotherapist

Motivation module• Positive-negative reinforcement• Qualitive and quantative feedback• Empathy and mirroring effect

Behavior module• Behavior selection • Decomposition to the set of operations: body movements, operational vocabulary, sounds, emotional expressions

High-level Perception• Visual: Scene analysis and segmentation Face and eye detection Whole body labeling Subjects identification Gaze direction Emotional Recognition• Kinect: Extremities movement analysis• Sensory Data fusion

Adaptation module• Adjustment to the stage of motor learning• Mood and current success level• Personality matching: hyperactive/passive

Information analysis module• Subject’s motor behavior analysis• Subject’s cognitive behavior analysis• Monitoring of the level of interaction strength• Recognition of subjects attention state

Input• Robot: Sensors data, including Scene video• Kinect: Movement tracking• Physiotherapist: Settings Personal and Anthropometric data Functional (pathological) restrictions Tasks difficulty Session scenario components

Safety module• Avoid children in robot’s working space

Database

Self-awareness module• Error detection/Success measuring Locomotion, Falling, Manipulation• Robot localization

Personal InformationPathology characteristics: GMFCS, Altered side, Assistive deviceMotor functions : Time of sit-stand initiation, Symmetry of holding arms Child-Robot Interaction Measurements: Emotional status, ResponsivenessEnvironment: Place, # participants

Conclusions

● Every person is different and no one can be sure how quickly or how far you or your family member will progress.

● Rehab is often a long process.

● Complete recovery is not always possible, however, living an enjoyable life is still within reach.

● ABRACADABRA is feasible and promising , a new research area of social assistive technology with immeasurable potential

● Assistive robotic platforms can be used in the near future in hospitals and homes, in training and therapeutic programs that monitor, encourage, and assist their users

● ABRACADABRA may stimulate the development of new treatments for a wide variety of diseases and disorders through effective physiotherapy practice