Machine learning techniques for independent gait recovery prediction in acute anterior circulation ischemic stroke

 Recovery prediction DOES NOTHING to get survivors recovered! Useless research, you're fired!

Machine learning techniques for independent gait recovery prediction in acute anterior circulation ischemic stroke

• Jiangping Ma &
• Yuanjie Xie 

Journal of NeuroEngineering and Rehabilitation volume 22, Article number: 19 (2025) Cite this article

• Metrics details

Abstract

Objective

This study aimed to develop and validate a machine learning-based predictive model for gait recovery in patients with acute anterior circulation ischemic stroke.

Methods

Between May and November 2023, 237 patients with acute anterior circulation ischemic stroke were enrolled. Patients were randomly divided into training and validation sets at a 7:3 ratio. Thirty-one medical characteristics were collected, and the Least Absolute Shrinkage and Selection Operator (LASSO) regression was applied to screen predictor variables. Predictive models were developed using the Random Survival Forest (RSF) and COX regression methods. The optimal model was identified based on C-index values. The SHapley Additive exPlanations (SHAP) method was employed to interpret the RSF model globally and locally.

Results

Ten predictors were identified through LASSO regression, including age, gender, periventricular white matter hyperintensities (PVWMH), Montreal Cognitive Assessment (MoCA), National Institutes of Health Stroke Scale (NIHSS), enlarged perivascular spaces in basal ganglia (BG-EPVS), lacunes, parietal infarction, basal ganglia infarction, and Timed Up & Go (TUG) test score. The C-index values of the COX regression and RSF models were 0.741 and 0.761 in the training set and 0.705 and 0.725 in the validation set, respectively. SHAP analysis of the RSF model identified BG-EPVS, TUG, MoCA, age, and PVWMH as the top five most influential predictors of gait recovery.

Conclusion

The RSF model demonstrated superior performance to the COX regression model in predicting gait recovery, offering a reliable tool for clinical decision-making regarding stroke patients’ prognoses.

Introduction

Ischemic stroke, with a lifetime global risk of 18.3%, is the third leading cause of disability among adults [1]. Stroke frequently leads to long-term and debilitating gait impairments, significantly affecting functional independence and quality of life [2]. Walking capacity is a crucial indicator of functional independence and long-term survival in stroke patients [3]. Accurate early prediction of gait disturbances in stroke patients is critical for formulating treatment plans and allocating rehabilitation resources effectively [4, 5]. Predicting gait recovery early also helps clinicians set realistic rehabilitation goals for post-discharge stroke patients. Despite its importance, gait recovery after stroke remains underexplored in predictive modeling [6,7,8]. Therefore, urgent clinical research is needed to establish robust models for predicting gait recovery in stroke patients.

Previous studies have employed COX regression models to evaluate gait recovery in stroke patients [9]. However, these methods rely on linear assumptions, limiting their ability to model the complex, non-linear relationships between prognostic variables in biological systems, thus reducing predictive accuracy. Novel solutions capable of handling these potentially non-linear variables are highly needed for accurate prognostic prediction.

Machine learning (ML) is a computational approach that uses data-driven algorithms to identify patterns and improve prediction accuracy [10]. In recent years, ML algorithms have been commonly applied to predict functional outcomes following stroke [11,12,13,14,15]. Unlike COX regression models, ML approaches can account for non-linear functions and complex variable interactions, enhancing predictive performance [16]. However, the complexity of ML models, often referred to as “black boxes,” poses interpretability challenges, which are critical in the medical field to ensure patient safety and effective treatment planning [17]. By interpreting predicted results, physicians can better understand the rationale for treatment and thus make accurate clinical decisions.

This study aimed to collect demographic features, clinical features, infarct region characteristics, and magnetic resonance imaging (MRI) features to develop predictive models for gait recovery in patients with acute anterior circulation stroke using COX regression and Random Survival Forest (RSF) models. The study also employed SHapley Additive exPlanations (SHAP) to evaluate the contribution of individual predictors, facilitating model interpretability and enabling early intervention opportunities.

More at link.