Scratch Detection Bracelet for Eczema

By /

Scratch Detection Bracelet for Eczema

ISEF Category: Biomedical Engineering

Ready to Turn This Idea Into a Real Project?

This guide was put together with the help of AI research tools to give you a solid starting point.But a competitive science fair project lives in the details: refining your research question, fine-tuning your variables, analyzing your data, and presenting your findings like a seasoned scientist.

For next steps tailored to your interests, skill level, and timeline, work one-on-one with a MehtA+ mentor. Learn more about MehtA+ Science & Engineering Research Mentorship →

Subcategory: Biomedical Devices  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

Eczema sufferers scratch in their sleep without ever remembering. The scratching wrecks the skin and the next day starts the cycle. A wristband that recognizes scratching motions versus other movements would let patients and doctors measure therapy by counts, not by guesses.

What Is It?

Atopic dermatitis is the medical name for eczema. Scratching is its defining behavior and the main driver of skin damage. Current outcome scores depend on memory.

An IMU detects wrist motion. A surface EMG sensor adds muscle activation. Combined, they give a clear signature for scratching vs. typing vs. waving.

An LSTM (long short-term memory) model classifies windows of sensor data. Trained on a self-collected and family-collected dataset, the LSTM separates scratching from other motions and counts events per night.

Why This Is a Good Topic

Wearable behavior detection is approachable and the problem is real. Hardware costs are modest. You will learn time-series ML, label-collection design, and clinical-outcome interpretation.

Research Questions

  • How does adding EMG to IMU change classifier precision?
  • What is the effect of window length on scratch detection?
  • Does the LSTM beat a random-forest baseline?
  • To what extent does wrist position affect signal?
  • Which feature dominates the LSTM decisions?
  • How does sleep stage affect signal quality?
  • What is the effect of therapy on detected scratch counts?

Basic Materials

  • MPU-6050 IMU.
  • MyoWare EMG sensor.
  • ESP32 development board.
  • LiPo battery.
  • Soft fabric wristband.
  • Sleep-diary template.
  • Informed-consent form for participants.

Advanced Materials

  • Clinical actigraphy device.
  • Sleep-stage monitor.
  • Multi-channel EMG amplifier.
  • Dermatology mentor.

Software & Tools

  • PyTorch: Trains the LSTM.
  • Python (NumPy, Pandas): Aggregates and labels data.
  • OpenCV: Syncs video ground truth.
  • Matplotlib: Plots scratch counts vs. therapy phase.

Experiment Steps

  1. Lock the bracelet hardware and wrist placement.
  2. Design a labeling protocol with video ground truth.
  3. Decide window length, features, and target classes.
  4. Split data by subject for cross-validation.
  5. Train and report precision and recall on scratching events.
  6. Compare scratch counts during therapy on vs. off periods.

Common Pitfalls

  • Reporting accuracy alone on imbalanced classes.
  • Skipping subject-wise splits.
  • Letting wrist position vary, shifting axes.
  • Trusting self-reported labels without video.
  • Forgetting to validate battery life across a full night.

What Makes This Competitive

A competitive entry uses leave-one-subject-out validation, reports precision and recall on rare scratching events, and validates against a sleep-diary or video ground truth. Comparing therapy on vs. off using the device adds clinical meaning.

Project Variations

  • Compare the wristband to a thigh-mounted sensor.
  • Add a thermal sensor for skin-temperature change.
  • Run the device on parents of eczema kids to test caregiver burden.

Learn More

  • PubMed: Search nocturnal scratching detection reviews.
  • NIH PubMed Central: Open-access atopic dermatitis papers.
  • PyTorch tutorials: Free LSTM examples.
  • National Eczema Association: Open educational materials.
  • MIT OpenCourseWare: Course 6.S191 Introduction to Deep Learning.
Shopping Cart