Center of Mass Estimation for Impaired Gait Assessment Using Inertial Measurement Units

Area of Science:

• Biomechanics
• Rehabilitation Engineering
• Machine Learning

Background:

• Pathological gait significantly affects quality of life and independence.
• Accurate assessment of musculoskeletal control is crucial for gait rehabilitation.
• Center of Mass (CoM) kinematics provide insights into gait dynamics and pathologies.

Purpose of the Study:

• To propose and validate a novel method for estimating Center of Mass (CoM) trajectories using Inertial Measurement Units (IMUs).
• To evaluate the efficacy of a bidirectional Long Short-Term Memory neural network for gait analysis in rehabilitation.
• To determine optimal IMU placement and quantity for accurate CoM estimation.

Main Methods:

• Utilized a bidirectional Long Short-Term Memory neural network trained on IMU data from five volunteers.
• Employed a leave-one-subject-out cross-validation strategy for robust model evaluation.
• Investigated the influence of IMU number and location on CoM estimation accuracy using principal component analysis.

Main Results:

• Achieved average Root Mean Square Errors (RMSEs) of 1.44cm (ML), 1.15cm (AP), and 0.40cm (IS) for CoM estimation.
• Identified 3-5 IMUs placed on legs and medial trunk as optimal sensor configurations.
• Observed increased error in the mediolateral direction for an individual with hemiparesis, potentially due to asymmetric gait.

Conclusions:

• The proposed IMU-based CoM estimation method offers a viable framework for assessing gait deviations.
• This technology can effectively evaluate the outcomes of rehabilitation interventions for pathological gait.
• Optimized sensor placement enhances the practicality and accuracy of gait analysis in clinical settings.