Abstract
While sarcopenia is a common complication after stroke, few tools exist to identify patients at high risk. This study aimed to develop and validate a risk prediction model for sarcopenia among stroke patients. Clinical data were prospectively collected from 313 stroke patients across 2 medical centers between October 2024 and March 2025. A total of 251 patients from Kunming First People's Hospital formed the training cohort, while 62 patients from Kunming Yanan Hospital comprised the external validation cohort. Independent predictors of sarcopenia were identified using univariate and multivariate logistic regression analyses. A nomogram was constructed based on the final model and validated externally. Body mass index (BMI), serum albumin level, nasogastric tube placement, NIH Stroke Scale (NIHSS) score, and history of stroke were identified as independent predictors (P < .05). The model showed good discriminative ability, with areas under the receiver operating characteristic curve (AUC) of 0.891 (95% CI: 0.848-0.891) for the training set and 0.743 (95% CI: 0.610-0.743) for the validation set. Calibration curves demonstrated good agreement between predicted and observed outcomes in both cohorts. This risk prediction model may assist clinicians in identifying stroke patients at high risk for sarcopenia, facilitating early intervention and improving clinical outcomes.