Accurate COP Trajectory Estimation in Healthy and Pathological Gait Using Multimodal Instrumented Insoles and Deep...

Area of Science:

• Biomechanics
• Neurology
• Machine Learning

Background:

• Assessing center-of-pressure (COP) trajectories outside laboratory settings is crucial for understanding gait and balance changes in neurological disorders.
• Current COP measurement tools are often expensive and inaccessible, limiting widespread clinical application.
• There is a need for cost-effective and portable solutions for dynamic COP trajectory estimation.

Purpose of the Study:

• To introduce and validate a novel deep recurrent neural network (DRNN) model for estimating dynamic COP trajectories.
• To fuse data from affordable, heterogeneous insole sensors, including force sensitive resistors (FSRs) and inertial measurement units (IMUs).
• To assess the technical and convergent validity of the proposed method for out-of-the-lab ambulatory tasks.

Main Methods:

• Development of a DRNN model integrating data from an eight-cell FSR array and an IMU embedded in insoles.
• Validation against gold-standard equipment (e.g., force plates) during simulated real-world walking tasks.
• Analysis of root-mean-square errors (RMSE) in mediolateral (ML) and anteroposterior (AP) directions for healthy and neuromuscular condition groups.

Main Results:

• The DRNN model accurately estimated dynamic COP trajectories with low RMSE in both ML (0.51-0.59 cm) and AP (1.44-1.53 cm) directions.
• Technical validity was confirmed in both healthy individuals and those with neuromuscular conditions.
• COP-derived metrics demonstrated significant correlations with clinical measures of ambulatory function and lower-extremity strength in the neuromuscular group.

Conclusions:

• The proposed method offers a technically valid and potentially cost-effective approach for estimating COP trajectories using insole sensors.
• This technology shows promise for clinical applications, enabling remote monitoring of gait and balance in neurological conditions.
• The findings support the convergent validity of insole-based COP estimation for assessing functional mobility.