Related Experiment Video
Updated: Jul 6, 2026
08:08
Oscillation and Reaction Board Techniques for Estimating Inertial Properties of a Below-knee Prosthesis
Published on: May 8, 2014
17.4K
Modified Extended Kalman Filter and Long Short-Term Memory-Based Framework for Reliable Stride-Length Estimation
Sensors (Basel, Switzerland)
|February 27, 2026
Summary
This study introduces a novel gait analysis framework using Long Short-Term Memory (LSTM) networks and advanced signal processing to accurately estimate stride length, improving mobility assessment for fall prevention.
Area of Science:
- Biomedical Engineering
- Wearable Technology
- Rehabilitation Science
Background:
- Gait analysis is crucial for detecting mobility issues like frailty and falls, especially in aging populations.
- Accurate spatiotemporal gait measurements are vital for early intervention but are often hindered by sensor signal noise and variability.
- Existing methods struggle with robust gait characterization, limiting their clinical utility.
Purpose of the Study:
- To develop and validate a novel stride-length estimation framework integrating advanced signal processing with Long Short-Term Memory (LSTM) networks.
- To enhance the accuracy and reliability of gait analysis for wearable systems.
- To address challenges posed by noise and signal variability in sensor-based gait monitoring.
Main Methods:
- A processing-and-estimation pipeline was developed, featuring wavelet-based denoising and cubic-spline interpolation.
- A Kalman-filtering stage with dynamic gain regulation was implemented to mitigate transient errors.
- Long Short-Term Memory (LSTM) networks were integrated for stride-length estimation, trained on data preprocessed with a Modified Extended Kalman Filter (EKF).
Main Results:
- The proposed framework significantly improved stride-length estimation accuracy, reducing the absolute mean error from 29.78% to 7.77%.
- The standard deviation of errors decreased from 20.31 to 7.17, indicating enhanced consistency.
- LSTM models trained on Modified EKF-preprocessed data achieved a Mean Absolute Error (MAE) of 0.0376 and an R² of 0.7066.
Conclusions:
- The integration of Modified EKF preprocessing with LSTM learning provides a robust and noise-resilient solution for stride-length estimation.
- This approach significantly enhances the reliability of wearable gait analysis systems.
- The framework offers valuable insights for clinical diagnostics, rehabilitation monitoring, and health management.
