Related Experiment Video
Updated: Nov 10, 2025
05:23
Motor Dual-Tasks for Gait Analysis and Evaluation in Post-Stroke Patients
Published on: March 11, 2021
2.6K
Externally validated model predicting gait independence after stroke showed fair performance and improved after
Anthonia J Langerak1, Alana B McCambridge2, Peter W Stubbs2
1University of Technology Sydney, Graduate School of Health, Discipline of Physiotherapy, Sydney, Australia; Utrecht University, University Medical Center Utrecht, Physical Therapy Sciences, program in Clinical Health Sciences, Utrecht, the Netherlands.
Journal of Clinical Epidemiology
|April 3, 2021
Summary
This study validated prognostic models for predicting independent gait after stroke. Updating the Australian model with leg strength significantly improved its predictive accuracy.
Area of Science:
- Neurology
- Rehabilitation Medicine
- Clinical Epidemiology
Background:
- Accurate prediction of independent gait post-stroke is crucial for patient outcomes and resource allocation.
- Recent prognostic models aim to predict gait recovery using accessible clinical predictors.
- External validation is essential to confirm the generalizability and reliability of these models.
Purpose of the Study:
- To externally validate recent prognostic models predicting independent gait in adult stroke survivors.
- To assess the performance of existing models using easily obtainable predictors.
- To explore model updating strategies for improved predictive accuracy.
Main Methods:
- A systematic literature search identified prognostic models developed within the last 10 years.
- Model predictors were assigned proxies where necessary for validation in a cohort of 957 stroke patients.
- Model performance was evaluated using discrimination (Area Under the Curve) and calibration metrics.
Main Results:
- Three models met the inclusion criteria; all exhibited a high Risk of Bias.
- The Australian model, using National Institute of Health Stroke Scale (NIHSS) and age, showed fair discrimination (AUC 0.77) and good calibration with proxy measures (Scandinavian Stroke Scale for NIHSS, Functional Independence Measure for Motor Assessment Scale walking item).
- Updating the model by incorporating paretic leg strength significantly improved predictive performance (AUC 0.82).
Conclusions:
- The externally validated Australian model, even with proxy variables, demonstrates fair predictive capability for independent gait post-stroke.
- Model performance can be enhanced by incorporating objective measures like paretic leg strength.
- Further research should focus on developing and validating robust prognostic models with readily available and objective predictors.
