This paper presents a semi-supervised data-driven approach to identify pediatric foot deformities using foot plantar pressure measurements. Essentially, the developed approach merges the desirable features of the kernel principal components analysis as a feature extractor and the Kantorovich Distance-driven monitoring scheme for detecting pediatric foot deformities. For extending the flexibility of the proposed scheme, kernel density estimation based nonparametric decision threshold is adopted. The method is assessed via publically available data containing three types of footsteps (i.e., normal, flat, and cavus). The detection results show that the method proved promising results, thus, outperforming commonly applied monitoring schemes.
ASJC Scopus subject areas
- Electrical and Electronic Engineering
- Hardware and Architecture