Gait parameter estimation from a miniaturized ear-worn sensor using singular spectrum analysis and longest common...

Area of Science:

• Biomechanics
• Sensor Technology
• Signal Processing

Background:

• Gait analysis is crucial for assessing mobility and detecting neurological disorders.
• Traditional gait analysis methods often require specialized equipment and controlled laboratory environments.
• Developing unobtrusive, wearable sensors is essential for remote and real-world gait monitoring.

Purpose of the Study:

• To present a novel approach for gait parameter estimation using a single ear-worn sensor.
• To evaluate the accuracy of the ear-worn sensor against instrumented treadmills and high-speed cameras.
• To determine the feasibility of ear-worn sensors for assessing and quantifying gait pattern changes.

Main Methods:

• Utilized a miniaturized ear-worn sensor with a triaxial accelerometer.
• Applied singular spectrum analysis and the longest common subsequence algorithm for gait parameter estimation.
• Incorporated signal periodicity and multi-axial accelerometer data for comprehensive analysis.
• Validated key gait events (heel contact, toe off) against high-speed cameras and force-plate instrumented treadmills.

Main Results:

• The ear-worn sensor accurately extracted gait parameters, including swing, stance, and stride times.
• Absolute errors for swing, stance, and stride times were within 35.5 ±3.99 ms, 36.9 ±3.84 ms, and 17.9 ±2.29 ms, respectively (95% confidence intervals).
• The sensor demonstrated feasibility for assessing and quantifying gait pattern changes in healthy adults.

Conclusions:

• A single ear-worn sensor is a viable tool for accurate gait analysis and parameter estimation.
• The proposed method, combining singular spectrum analysis and LCS algorithm, effectively captures gait dynamics.
• This technology offers a promising, unobtrusive solution for continuous gait monitoring and assessment.