Integrating serum globulin into the ICF framework: a novel multidimensional predictive model for 1-year mRS outcomes in acute ischemic stroke

TL;DR

This study creates a new model to predict long-term recovery in stroke patients by including blood protein levels and other factors.

Contribution

The novel integration of serum globulin into the ICF framework for predicting 1-year stroke outcomes.

Findings

A model with serum globulin and other factors predicted 1-year stroke recovery with high accuracy.

The model showed good calibration and outperformed standard strategies in decision analysis.

Abstract

The long-term functional prognosis after ischemic stroke (IS) plays a crucial role in rehabilitation planning, yet it remains challenging to predict in clinical practice. Existing prognostic models primarily focus on short-term outcomes and lack integration of multidimensional determinants. Although elevated serum globulin levels have been associated with acute neuroinflammation and short-term disability, their prognostic significance for 1-year functional outcomes within a comprehensive biopsychosocial framework has not yet been established. To address these gaps, this study aimed to develop and validate a multidimensional prognostic model that integrates serum globulin as a key inflammatory biomarker into the International Classification of Functioning, Disability and Health (ICF) framework of the World Health Organization (WHO), with the objective of predicting 1-year functional outcome in patients with acute ischemic stroke (AIS).

This prospective study consecutively enrolled 1,562 AIS patients at a Grade A tertiary hospital from 2021 to 2023; after data cleaning and screening, 1,356 cases were included for analysis. Baseline data were collected within 1 week of hospital admission. The study cohort was randomly divided into a training set (70%, n = 949) for model development and a validation set (30%, n = 407) for internal validation. The primary outcome was the patients’ functional status assessed using the modified Rankin Scale (mRS), 1-year post-admission. Predictors significant (p < 0.05) in univariate analysis within the training set were entered into backward stepwise multivariable logistic regression. The performance of the model was comprehensively evaluated using the area under the curve (AUC), Hosmer–Lemeshow test, calibration curve, and decision curve analysis (DCA).

Multivariable analysis identified six independent predictors (all p < 0.05): age, occupational status, BI, serum globulin, number of stroke episodes and NIHSS. A nomogram incorporating these predictors demonstrated excellent discrimination in both training (AUC = 0.90, 95% CI 0.88–0.93) and validation sets (AUC = 0.85, 95% CI 0.80–0.89). Calibration was good, with predicted probabilities (training: 22.23%; validation: 21.72%) closely matching the observed incidence of 22.20%, nonsignificant Hosmer–Lemeshow test results (p > 0.05), and well-aligned calibration curves. DCA confirmed the model’s superior net benefit over “treat-all” and “treat-none” strategies across clinically relevant high-risk thresholds (20–80%) in both training and validation cohorts.

This study successfully integrated serum globulin into the ICF framework and constructed a prognostic model for the 1-year prognosis after AIS. It enables early identification of high-risk individuals and personalized rehabilitation strategies to improve long-term recovery.

The online version contains supplementary material available at 10.1186/s40001-026-04004-9.

Figures