Estimation of Ankle Joint Moment From Plantar Pressure Through an Optimized Sensor Layout Using Genetic Algorithm and...

Area of Science:

• Biomechanics
• Wearable Technology
• Machine Learning

Background:

• Accurate ankle joint moment estimation is crucial for gait analysis but traditionally requires complex inverse dynamics modeling.
• Plantar pressure insoles offer a feasible, wearable solution for gait analysis, though high-precision moment estimation remains a challenge.
• Optimizing sensor number and placement is key to cost-effective, accurate wearable gait analysis.

Purpose of the Study:

• To develop and validate a novel algorithm combining genetic algorithm (GA) and deep forest regression (DFR) for ankle joint moment estimation.
• To optimize the number and layout of plantar pressure sensors using the GA-DFR approach.
• To achieve cost-effective, high-precision ankle joint moment estimation from plantar pressure data.

Main Methods:

• Recruited 26 healthy participants to collect gait data (motion, ground reaction forces, plantar pressure) at various speeds.
• Utilized inverse dynamics to calculate ankle joint moments for training and validation data.
• Developed a GA-DFR optimization algorithm to determine the optimal sensor configuration and estimate ankle joint moments, validated using leave-one-out cross-validation.

Main Results:

• The optimized sensor layout comprised 9 sensors, achieving high fitness.
• Excellent Pearson Correlation Coefficients were reported for sagittal (0.967 ± 0.014), coronal (0.918 ± 0.027), and transverse (0.894 ± 0.073) plane moments.
• The estimation accuracy remained consistent across different walking speeds (P > 0.05).

Conclusions:

• The proposed GA-DFR algorithm accurately estimates ankle joint moments and optimizes sensor configuration.
• This approach enables rapid and accurate ankle joint moment estimation from plantar pressure insoles.
• The study presents a trade-off approach for efficient and precise gait analysis using wearable technology.