Smart Sleep Posture Mat for Apnea Motion Patterns

ISEF Category: Embedded Systems

Subcategory: Sensors  ·  Difficulty: Advanced  ·  Setup: School Lab  ·  Time: 1 to 2 Months

The Hook

Your body does not stay still when you sleep. Tiny shifts can tell a big story about posture, pressure, and breathing patterns. A fabric sensor mat can turn those shifts into data you can analyze. That makes this a strong project for sleep science and smart textiles.

What Is It?

This project uses a soft mat built from conductive thread and Velostat, a pressure-sensitive plastic film. When pressure changes across the mat, the sensor signal changes too. Think of it like a checkerboard that can feel where your body rests, shifts, or rolls during sleep.

You are not trying to diagnose sleep apnea. You are testing whether a sensor mat can detect motion patterns that might line up with restless sleep or breathing-related movement changes. The key idea is distributed sensing, which means many sensing points work together instead of one sensor under one spot.

Why This Is a Good Topic

This is a strong science fair topic because you can measure real signals, compare patterns, and test an algorithm you build yourself. You can ask clear questions about posture, movement, and anomaly detection without needing a hospital lab. The project connects to sleep monitoring, smart textiles, and wearable health tech, all of which have real-world use.

Research Questions

• How does sensor spacing affect the mat's ability to detect different sleep postures?
• What is the effect of using more sensing zones on posture classification accuracy?
• Does adding simple anomaly detection improve the mat's ability to flag unusual motion patterns?
• To what extent can a textile pressure mat distinguish lying on the back, side, or stomach?
• Which signal features best separate normal rolling movements from apnea-like stillness patterns?
• How does repeated use over several nights change sensor drift and posture maps?

Basic Materials

• Conductive thread for stitching sensor traces.
• Velostat sheet or roll for pressure-sensitive layers.
• Fabric or felt base for the mat.
• Conductive fabric or copper tape for contact points.
• Microcontroller with analog inputs, such as an Arduino or similar board.
• Breadboard and jumper wires.
• Resistors for forming voltage dividers.
• Digital multimeter.
• Laptop for data logging and analysis.
• Lightweight foam or cardboard frame for keeping layers aligned.

Advanced Materials

• Conductive thread with known resistance per length.
• Velostat or similar piezoresistive film.
• Stretch fabric or spacer fabric for cleaner textile integration.
• Multiplexer for expanding the number of sensing channels.
• Microcontroller with higher sampling speed and onboard storage.
• 3-axis accelerometer for motion validation.
• Reference pressure weights for calibration.
• Oscilloscope or data acquisition board.
• Thermal camera or infrared sensor for optional sleep context measurements.
• Sewing tools for repeatable textile assembly.

Software & Tools

• Arduino IDE: Programs the microcontroller and reads sensor channels.
• Python: Cleans sensor data, builds features, and tests classification models.
• Jupyter Notebook: Helps you compare posture maps and plot results step by step.
• ImageJ: Measures and compares heatmap-style images if you turn sensor data into visual maps.
• GeoGebra: Useful for quick graphing and curve fitting when you test calibration.

Experiment Steps

1. Define the sleep-related motion pattern you want to detect, such as posture changes, stillness, or sudden shifts.
2. Design the sensing grid so each zone gives you enough spatial detail without making the mat too hard to build.
3. Plan a calibration method that turns raw sensor readings into comparable pressure or contact values.
4. Choose the features you will extract from each signal window, such as peak change, duration, variance, or spatial spread.
5. Decide how you will label ground truth posture or movement, using video, timestamped notes, or a second reference sensor.
6. Set up a simple anomaly detection rule or model, then test it on held-out data to see where it fails.

Common Pitfalls

• Building a mat with too few sensing zones, which makes left-side and right-side sleep postures look the same.
• Letting conductive thread traces shift during sewing, which creates short circuits or weak contacts.
• Ignoring baseline drift in Velostat, which makes the same posture produce different readings over time.
• Training the anomaly detector on data from only one sleeper, which makes the model memorize one body instead of learning motion patterns.
• Using room light or camera angle as the main reference for posture labels, which causes ground truth errors when the sleeper moves under blankets.

What Makes This Competitive

A stronger project will do more than detect pressure. You can compare multiple sensor layouts, test a real baseline model, and report accuracy with confusion matrices, false alarm rates, and drift over time. You can also add a harder angle, like cross-person testing or an edge model that runs on the microcontroller itself. That kind of analysis shows you understand both the sensing hardware and the data side.

Project Variations

• Test whether a smaller grid of sensing cells can match a full mat for posture classification.
• Compare Velostat to another pressure-sensitive layer to see which one gives cleaner sleep maps.
• Add an accelerometer and test whether fused motion data improves anomaly detection compared with the textile mat alone.

Learn More

• NIH PubMed: Search for review articles on sleep posture, pressure sensing, and apnea-related movement patterns.
• NASA Open Science Data Repository: Look for examples of anomaly detection and sensor data analysis methods.
• MIT OpenCourseWare: Find free materials on embedded systems, signal processing, and sensor interfacing.
• USGS Publications Warehouse: Search for papers on calibration, sensor drift, and field measurement methods that transfer well to your project.
• IEEE Xplore abstracts: Read abstracts on textile sensors and pressure mapping, then use your library access for full papers if available.