A Hybrid TCN-Transformer Encoder Framework for Gait Classification and Rehabilitation Assessment

DOI

10.2991/978-94-6239-693-7_20How to use a DOI?

Keywords

GAN; Gound reaction force; Symmetry index; gait classification; rehabilitation

Abstract

A rigorous modelling of biomechanical pattern of lower limb fracture patients is required to handle the class imbalance between normal and defective gait, rehabilitation assessment for patients. This work uses vertical ground reaction force (vGRF) data and suggests a novel cost sensitive deep learning framework for gait classification and rehabilitation analysis based on a hybrid temporal convolutional network (TCN) and transformer encoder architecture. The data imbalance is handled using GAN which creates artificial samples during preprocessing. The global temporal context and inter phase gait interactions are recorded by the transformer encoder while the TCN block extracts local temporal gait variables with force magnitude and loading characteristics. A binary classifier is used to classify gait normal or abnormal once the learnt representations from both modules are combined to create a comprehensive gait feature space. Patients with abnormal gait are referred to rehabilitation programs and deep features are extracted from the trained model during rehabilitation sessions to evaluate the gait improvement. The performance metrics like accuracy of 0.9938, precision of 0.9978, recall of 0.9943 and F1 score of 0.9960 are obtained for gait classification. The regression metrics like RMSE of 1.1700, MAE of 0.8536 and R2 of 0.9870 are obtained from the proposed architecture for monitoring gait recovery in lower limb fracture patients.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - O. Pushpalatha
AU  - R. Premkumar
PY  - 2026
DA  - 2026/06/16
TI  - A Hybrid TCN-Transformer Encoder Framework for Gait Classification and Rehabilitation Assessment
BT  - Proceedings of the International Conference on Intelligent Systems for a Sustainable Future (ISSF 2026)
PB  - Atlantis Press
SP  - 185
EP  - 194
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6239-693-7_20
DO  - 10.2991/978-94-6239-693-7_20
ID  - Pushpalatha2026
ER  -