Wearable Insole Gait Tracking for Parkinson’s Patterns

ISEF Category: Embedded Systems

Subcategory: Sensors  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A tiny change in how you step can carry a lot of information. For people with Parkinson's disease, gait often shifts from day to day, and those shifts can be hard to catch early. A smart insole can turn each step into data. That gives you a way to study movement patterns without a clinic-sized setup.

What Is It?

This project studies how a wearable insole can track changes in walking patterns. The insole uses three force-sensitive resistors, or FSRs, which measure pressure, and an MPU-6050, which measures motion and tilt. Together, they act like a small sensor team inside a shoe. One sensor reads where pressure lands. Another reads how the foot moves through space. When you combine them, you can estimate gait variability, which means how much a person's steps change from one step to the next.

The Parkinson's connection comes from movement changes that can show up as tremor, shuffling, uneven pressure, or inconsistent stride timing. You are not diagnosing disease. You are building a system that can detect pattern shifts in motion data, including changes that resemble a "good day" versus a "bad day." On-device unsupervised drift detection means the system watches for data that slowly changes over time, without needing labeled examples for every case. Think of it like a smoke alarm for data patterns. It learns what normal looks like, then flags when the pattern starts to move away from that baseline.

Why This Is a Good Topic

This is a strong science fair topic because you can test real sensor performance, not just build a gadget. You can study how pressure and motion signals change under different walking conditions, then measure whether your algorithm spots drift or variability better than a simple threshold. The project connects to assistive health tech, which gives it real-world value. You also get to learn sensor fusion, signal processing, and basic anomaly detection, all of which make a project feel much more advanced.

Research Questions

• How does walking speed affect gait variability measured by three FSRs and an MPU-6050?
• What is the effect of simulated tremor on pressure distribution across the insole sensors?
• Does combining FSR data with MPU-6050 data improve detection of gait changes compared with either sensor alone?
• To what extent can an unsupervised drift-detection method distinguish baseline walking from changed walking patterns?
• Which sensor placement on the insole gives the clearest separation between stable and variable steps?
• How does step-to-step variance change between simulated good day and bad day walking conditions?

Basic Materials

• Wearable insole or shoe insert base with space for sensors.
• Three FSR sensors.
• MPU-6050 accelerometer and gyroscope module.
• Microcontroller with data logging support, such as Arduino Nano 33 BLE or ESP32.
• Breadboard and jumper wires.
• Flexible wire or conductive thread.
• Small rechargeable battery pack.
• Tape, heat-shrink tubing, or fabric channels for sensor mounting.
• Computer for data download and analysis.
• Digital scale for checking repeatability of sensor placement.

Advanced Materials

• High-precision force sensors or load cells for validation.
• Motion capture system or high-speed video for ground-truth comparison.
• Instrumented treadmill or pressure mat access.
• Biomedical-grade wearable enclosure materials.
• Calibration weights and a repeatability jig.
• Reference IMU module for cross-checking motion data.
• University microfabrication or electronics tools for custom insole layouts.
• Data acquisition board with synchronized sampling support.
• Clinical gait dataset for external comparison.
• Research computing access for model testing.

Software & Tools

• Arduino IDE: Programs the microcontroller and streams sensor data.
• Python: Cleans signals, computes gait features, and tests drift detection.
• Jupyter Notebook: Lets you document analysis, graphs, and model comparisons in one place.
• ImageJ: Measures foot placement or video markers if you validate against recorded walking footage.
• Excel or Google Sheets: Helps you inspect sensor trends, calibration curves, and step summaries quickly.

Experiment Steps

1. Define the exact gait feature you want to track, such as step symmetry, pressure balance, or signal drift over time.
2. Choose a sensor layout that matches your question, then decide which signal each sensor should measure.
3. Plan a baseline condition and one or more changed walking conditions so you can compare patterns cleanly.
4. Build a calibration plan that converts raw sensor output into repeatable measurements.
5. Select the drift-detection approach you will test, then decide what counts as a detected change.
6. Design validation tests against video, manual step counts, or another reference so your results have a ground truth.

Common Pitfalls

• Mounting the FSRs too loosely, which makes pressure readings shift when the insole flexes.
• Letting sensor wires move inside the shoe, which creates false spikes in the motion data.
• Testing only one walking style, which makes the algorithm look better than it really is.
• Comparing raw sensor counts instead of calibrated features, which hides real gait changes.
• Collecting data from too few steps or too few trials, which makes day-to-day drift look like random noise.

What Makes This Competitive

A strong version of this project goes beyond simple step counting. You would compare sensor fusion methods, test more than one drift metric, and validate the system against an external reference such as video or a pressure mat. You could also separate short-term motion noise from longer-term pattern change, which is where the analysis gets more serious. Clear controls, repeatable calibration, and a thoughtful statistical test can turn this into a project with real research depth.

Project Variations

• Test whether the insole works better for stair walking than level walking, since stairs change pressure timing and balance demands.
• Replace the Parkinson's-style simulation with fatigue-based gait changes to see whether the same drift detector still catches movement shifts.
• Compare three sensor layouts, such as toe-heavy, heel-heavy, and balanced placement, to see which one best detects gait variability.

Learn More

• NIH PubMed: Search review articles on gait variability, wearable sensors, and Parkinson's disease motor symptoms.
• NCBI Bookshelf: Read free biomedical textbook chapters on movement disorders and sensor-based measurement methods.
• NIH MedlinePlus: Find patient-friendly background on Parkinson's disease and gait changes.
• Arduino Documentation: Learn microcontroller data logging, serial output, and sensor integration basics.
• MIT OpenCourseWare: Search for free courses on embedded systems, signal processing, and biomedical instrumentation.