Development of a Risk Prediction Model for Post-Stroke Functional Recovery Based on Clinical and Nursing Factors

This study aimed to develop and validate a risk prediction model for unfavorable functional recovery at 6 months after stroke, incorporating clinical and modifiable nursing-related predictors to support early risk stratification and individualized nursing interventions.

A retrospective cohort study included 1340 stroke patients from a tertiary hospital. Demographic, clinical, imaging, nursing, and psychosocial data were collected. The dataset was split into training and testing sets at a 7:3 ratio using outcome-stratified sampling. Univariate analysis and multivariable logistic regression were used as a predictor screening procedure. A nomogram-based risk prediction model was developed and evaluated for discrimination, calibration, and clinical utility. Patients were further stratified into low-, intermediate-, and high-risk groups based on predicted probability tertiles from the training set.

Older age, prior stroke history, comorbidity burden, longer onset-to-admission time, greater neurological deficit severity, lower Glasgow Coma Scale score, and brainstem lesions were independent predictors of unfavorable recovery. Early out-of-bed mobilization within 48 hours, higher self-management behavior score, absence of depressive symptoms, and better social support were linked to reduced risk. The model showed good discrimination and calibration in both sets. Risk stratification showed a stepwise increase in unfavorable recovery rates across the three risk groups. Decision curve analysis indicated net clinical benefit within a reasonable threshold probability range.

The developed model combines clinical and modifiable nursing-related predictors and demonstrates good predictive performance. It may serve as a practical tool for early risk stratification and targeted nursing interventions in stroke patients.