Therapy Tiles
PLAY-BASED THERAPY FOR CHILDREN WITH HEMIPARETIC CEREBRAL PALSY
A modular tile board that turns constraint-induced movement therapy into something a two-year-old actually wants to do.
Context
Carnegie Mellon · BioMedical Engineering · 2025
Timeline
[Timeline]
Team
10 students, mechanical and biomedical engineering
My Role
Industrial design, CAD, prototyping and testing, user research
Tools
SolidWorks, FEA, Arduino, 3D printing
Focus
Assistive design, pediatric therapy, physical computing, material safety
The Problem
Constraint-induced movement therapy works by restraining a child’s stronger arm so the weaker one has to do the work. Clinically it’s effective. In practice it means casting the arm of a two-year-old for three weeks.
1 in 345 children is diagnosed with cerebral palsy, and around 40% present with hemiparesis. The therapy that helps them most is also the therapy they resist hardest.
Three failures kept surfacing: restrictive casts cause frustration and abandonment, there’s almost no motivation to continue therapy at home, and neither parents nor therapists get feedback on whether it’s working.
Make the harder hand the more rewarding choice.
Research
01
The neuroplasticity window is 2 to 5 years old. Narrow, and it closes—design for toddlers, not children generally.
02
CIMT’s real target is grip release, not grip strength. Opening the hand is the hard part.
03
You cannot legally constrain the unaffected hand. Make the affected hand more rewarding instead.
04
The child has to be able to walk away. Any design that traps a child fails on autonomy and engagement.
Principles
Reward, don’t restrain. The tile only responds to the affected hand via RFID—no cast, no brace.
Design for release. Every interaction ends in an open hand, not a closed one.
Make quitting allowed. The child can disengage without being punished.
Survive a toddler. Wipeable, non-toxic, under 5 lbs; no small pieces or sharp radii.
Concept Selection
Reachie, a companion robot the child follows with the affected hand
Therapy Tiles, a modular board of interchangeable activity tiles
Tiles won: therapists can reconfigure challenge from one session to the next.
Form, Build & Test
FORM · Hexagonal geometry enables gap-free modularity.
INTERACTION · Five embedded 30 mm buttons.
SYSTEM · Arduino, pressure sensors, RFID bracelet.
FORCE TARGETS
Feedback calibrated for little hands.
3.5 N
Low activation
7.1 N
Normal activation
DROP TEST · SIX-FOOT DROPS · FEA
4 orientations tested · 2 passed · corner impacts remain the open problem.
Orientation
Max stress
Factor of safety
Flat bottom face
46.79 MPa
1.28
Flat side face
42.13 MPa
1.42
Bottom vertex
335 MPa
0.179
Top vertex
172.8 MPa
Material & Electronics
MATERIAL
ABS
≈60 MPa compressive strength against 0.06 MPa expected contact pressure. Durable, impact-resistant, chemically cleanable, biocompatible, printable, and cost-conscious.
ELECTRONICS · Arduino, pressure sensors, LEDs, speaker, dedicated PCB, and RFID bracelet.
Physical Prototyping
Tiles were 3D printed, handled, and tested with classmates and faculty.
EDITABLE NOTE
What handling the physical prototype taught me that CAD didn’t.
Validation Planned → Not Yet Executed
A full validation plan was written but could not be completed within the course timeline. Classmate and faculty handling did not substitute for clinical testing.
CIMT-trained therapist testing to assess clinical benefit.
Testing with healthy children and children with hemiparetic CP for age-appropriateness.
Caregiver testing.
Mechanical and chemical limit testing for voltage, strength, and resistance.
Final Design
Interchangeable tile modules
RFID bracelet
Pressure and touch sensors
LED and speaker feedback
Left- and right-handed use
Wipeable ABS shell
Target Skills
DEXTERITY
GRIP STRENGTH
REACTION TIME
System / Use
The RFID bracelet wakes the board when the affected hand comes near. Lights and sound draw the child in; they press, grip, and release across tiles targeting different skills. Sensors log contact time, pressure, and movement speed for therapists, while new tile sets can be configured per session.
Outcome & Reflection
ENGINEERING LIMITATION
The design passes flat drops and fails on corners. A factor of safety of 0.179 at the bottom vertex means a tile dropped on its point from table height would likely crack. The honest fix isn’t more material—it’s a geometry change: increase the vertex fillet radius, or overmold the corners in TPE to improve grip and reduce impact noise.
MY REFLECTION
[Add personal reflection here — what surprised me, what I would change, and what I learned.]