Plantar pressure classification and feature extraction based on multiple fusion algorithms.

Area of Science:

• Biomechanics
• Gait Analysis
• Human Movement Science

Background:

• Plantar pressure distribution during walking is crucial for understanding foot function.
• Variations in pressure patterns may indicate different biomechanical characteristics.
• Optimizing classification and feature extraction is essential for accurate gait analysis.

Purpose of the Study:

• To classify and extract features of plantar pressure during the walking stance phase in healthy individuals.
• To explore the diversity of plantar pressure distribution patterns.
• To identify key indicators of plantar pressure distribution.

Main Methods:

• Utilized multiple fusion algorithms for classification and feature extraction.
• Employed Principal Component Analysis for data dimensionality reduction.
• Applied Linear Discriminant Analysis and Analysis of Variance for classification and comparison.

Main Results:

• Identified three distinct plantar pressure distribution types: thumb extension, midfoot-lateral forefoot push-off, and normal.
• Extracted four key indicators: impulses of Toe1, Meta1, Meta5, and Midfoot.
• Achieved average accuracy rates of 89.70% (original data) and 88.50% (cross-validation).

Conclusions:

• Plantar pressure distribution during walking can be reliably categorized into three types.
• Impulses in the first metatarsophalangeal joint, fifth metatarsal bone, and midfoot regions are important for classification.
• Future research should explore the link between these gait characteristics and sports injuries.