Our Work

Our Projects - Facial Palsy Rehabilitation Game

A Smartphone-Based Facial Electromyography Rehabilitation Game

Who

M GordanProject Student

Level

Undergraduate

Area

EMG, AI, ML, Digital

When

2025

Details

Facial palsy is a neurological condition that limits voluntary facial movement, often causing significant psychological, social, and emotional challenges. Traditional rehabilitation methods, such as mirror therapy or camera-based feedback, can be uncomfortable, demotivating, and ineffective, especially when muscle contractions are too subtle to see.

This project introduces an innovative solution: A smartphone-based rehabilitation game controlled by facial electromyography (fEMG) signals.

Instead of relying on visual feedback, the system detects electrical muscle activity, allowing even subtle or sub-visual movements to be used as input. The goal is to make rehabilitation:

  • More engaging (gamified)
  • More accessible (mobile-based)
  • More effective (effort-sensitive feedback)

Methods

The project integrates hardware, signal processing, machine learning, and game design into one system.

  • 1. Data Acquisition
  • 2. Signal Processing & Feature Extraction
  • 3. Machine Learning Classification
  • 4. Game Development
  • 5. User Testing

Results

Classification Performance

  • 80.1% MCC (large dataset)
  • ~83–89% MCC (participant datasets)
  • KNN outperformed all other classifiers

System Performance:

  • Total latency: ~62.6 ms
  • Within acceptable real-time gaming threshold (<100 ms)

User Feedback:

  • Overall satisfaction: 4.06 / 5
  • High scores for Enjoyment (4.5/5) and Motivation (4.25/5)
  • Challenges identified: Electrode adhesion issues, setup time and variability across users

Conclusions

This project demonstrates that: fEMG signals can be used as a reliable real-time control input, gamification significantly improves engagement in rehabilitation and a smartphone-based system is feasible and effective.