Smartphone Gait Asymmetry for Neuropathy Screening

ISEF Category: Biomedical and Health Sciences

Subcategory: Pathophysiology  ·  Difficulty: Intermediate  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

Your feet can show nerve trouble before pain does. A single phone video can catch tiny left-right walking differences that your eye would miss, like hearing a song that is almost, but not quite, in tune. That makes gait asymmetry a smart science fair topic because you can turn everyday walking into numbers. You also get a real health question with a simple camera setup.

What Is It?

Gait asymmetry means your left and right sides do not move in the same way when you walk. One step may last a little longer, one knee may bend more, or one hip may shift farther. In people with diabetic peripheral neuropathy, nerve damage in the feet and legs can change balance, timing, and stride control. Think of walking like rowing with two oars. If one oar pulls a little differently, the boat still moves, but the path wobbles.

MediaPipe pose estimation reads body landmarks from video, such as hips, knees, ankles, and shoulders. You can turn those points into measurements like step timing, stride length, and side-to-side sway. Your project asks whether one rear-facing smartphone video can reveal patterns that line up with published gait-lab norms. That gives you a simple camera, a real medical question, and a clear path to data analysis.

Why This Is a Good Topic

This topic works well because you can measure it from video, compare groups, and test one variable at a time. It connects to a real need, early signs of diabetic nerve damage can raise fall risk before symptoms get obvious. You can learn video tracking, data cleaning, basic statistics, and how to compare your numbers with published norms.

Research Questions

• How does step-time asymmetry change when a walker uses a normal pace versus a faster pace?
• What is the effect of footwear type on MediaPipe-based stride symmetry in the same person?
• Does trunk sway measured from a rear-facing video separate walkers whose scores match published gait-lab norms from those who do not?
• To what extent do step-length and step-time asymmetry agree with each other?
• Which landmark set, hips only or hips plus ankles, gives the most repeatable asymmetry score?
• How does adding a second trial change the stability of the gait estimate?

Basic Materials

• Smartphone with rear camera and video recording
• Tripod or fixed phone stand
• Measuring tape
• Painter's tape or floor markers
• Long, flat walkway with even lighting
• Laptop or desktop computer
• Spreadsheet software
• Printed consent and data log sheets

Advanced Materials

• 3D motion-capture system with calibration tools
• Force plates or an instrumented walkway
• Wearable inertial measurement units
• Reflective markers or a lab-grade markerless tracking setup
• Electromyography system
• Clinical neuropathy screening tools approved by the lab
• Anthropometric measurement kit
• High-speed camera

Software & Tools

• MediaPipe: Tracks body landmarks from each video frame and gives you coordinates for gait measures.
• Python: Cleans landmark data, calculates asymmetry metrics, and automates repeated analysis.
• Jupyter Notebook: Keeps your code, notes, plots, and results together in one place.
• OpenCV: Reads video files, checks frames, and helps you crop or inspect the walk path.
• R: Runs statistics and builds clear plots for group comparisons and repeatability checks.

Experiment Steps

1. Define the gait metric you will measure first, such as step-time asymmetry, stride length difference, or trunk sway.
2. Lock down one video setup, then keep camera angle, walking path, and lighting the same across every trial.
3. Choose a comparison plan, such as healthy peers, published norms, or both, and match subjects on simple factors like height or age range.
4. Build a tracking workflow that turns each video into repeatable numbers and flags bad frames before analysis.
5. Decide how you will test the signal, using repeatability checks, effect sizes, or a simple classifier rather than only eyeballing the plots.

Common Pitfalls

• Letting the walker drift toward or away from the camera, which changes apparent step length and body angles.
• Tracking feet with MediaPipe alone, which can fail when shoes overlap or the lower body leaves the frame.
• Mixing leftward and rightward walking trials without labeling direction, which flips asymmetry signs.
• Comparing your video data to gait-lab norms built from different speeds, ages, or camera setups, which makes the benchmark unfair.
• Treating one noisy walk as the whole result, which hides trial-to-trial variation and weakens the screen.

What Makes This Competitive

A strong version goes beyond "can I measure gait?". You compare several asymmetry metrics, test repeatability across trials, and check whether the signal still holds after matching for height, pace, or footwear. Strong entries also use clear controls, careful error analysis, and a fair comparison to published norms or a small local reference group. If you can show which metric works best and where the method breaks, the project feels much more like research than a demo.

Project Variations

• Compare barefoot walking with sneaker walking to see whether footwear changes asymmetry scores.
• Replace step-time asymmetry with trunk-sway analysis and test whether upper-body motion gives a clearer signal.
• Use a small local reference group, then compare your results with published gait-lab norms and report where they agree or differ.

Learn More

• PubMed: Search review articles on diabetic peripheral neuropathy, gait asymmetry, and fall risk.
• NIH National Institute of Diabetes and Digestive and Kidney Diseases: Read overview pages on diabetic neuropathy and nerve damage.
• CDC Diabetes Data and Statistics: Find background on diabetes burden and complication rates.
• Gait & Posture: Search the journal for studies on gait asymmetry, diabetic walking patterns, and screening.
• MediaPipe Documentation: Find pose-landmark definitions and examples for video tracking.
• MedlinePlus: Read patient-friendly background on diabetic neuropathy and related symptoms in the U.S. National Library of Medicine.