Answers are grounded in this paper and cite their sources.
BACKGROUND: Functional outcomes after acute ischemic stroke (AIS) vary substantially, yet the heterogeneity of long-term trajectories and their associated baseline features remain incompletely characterized. This study aimed to identify distinct 5-year functional outcome trajectories and characterize their associated clinical, biomarker, and imaging profiles. METHODS: This longitudinal multicenter study included 9073 patients with AIS from the Third China National Stroke Registry (2015-2018). Group-based trajectory modeling was applied to repeated modified Rankin Scale assessments from discharge to 5 years. We then compared clinical, biomarker, and imaging features across the identified trajectory groups, and applied machine learning coupled with SHapley Additive Explanations (SHAP) to identify and rank the key features associated with each trajectory group. Data were split by center to compare model performance and select the best-performing model for feature interpretation. RESULTS: Four trajectories were identified: Favorable Rapid Recovery (33.8%), Favorable Steady Improvement (44.8%), Adverse Fluctuating-Worsening (15.0%), and Adverse Progressive Deterioration-Death Group (6.3%). These trajectories differed in their clinical, biomarker, and imaging profiles. LightGBM achieved the best internal validation performance (AUC = 0.711) and was selected for subsequent SHAP interpretation. The key features associated with each trajectory group included younger age and lower National Institutes of Health Stroke Scale (NIHSS) for Trajectory 1; lower coagulation markers and lacunar infarction for Trajectory 2; higher NIHSS, large vessel impairment, and low education for Trajectory 3; and advanced age, multimorbidity, and elevated inflammatory markers for Trajectory 4. Sensitivity analyses showed broadly similar feature patterns. CONCLUSIONS: Four heterogeneous 5-year functional outcome trajectories were identified after AIS, ranging from rapid and sustained recovery to progressive disability and mortality. These trajectories were characterized by distinct clinical, biomarker, and imaging profiles. Baseline features, including age, NIHSS, coagulation, inflammation, and vessel impairment, were key features associated with trajectory groups, with varying weights and directions. These findings provide a foundation for hypothesis generation in future research on risk stratification, personalized rehabilitation, and secondary prevention.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: