 This study proposes a novel approach to improve the accuracy of instrumented footwear for ambulatory gate analysis. Transduction learning models were developed to reduce measurement errors associated with conventional techniques while maintaining computational simplicity. These models were tested on a group of elderly individuals with varying degrees of gate and balance impairment. Results showed significant improvements in both spatial and temporal parameters, suggesting the potential of transduction learning models for improving the accuracy of instrumented footwear for ambulatory gate analysis. This article was authored by Huang Hejiang, Tuan T. H. Duong, Aishwini K. Rao and others.