Construction of a Risk Prediction Model of Swallowing Dysfunction During Recovery from Acute Ischemic Stroke

Objective: This study aimed to analyze risk factors for post-stroke dysphagia during the recovery phase of acute ischemic stroke and to construct a preliminary risk prediction model for patients who were eligible for bedside swallowing assessment. Methods: A retrospective analysis was performed for patients with ischemic stroke who were continuously admitted to Shanghai Fourth People’s Hospital. Swallowing function was assessed using the water swallowing test. Univariate and multivariate analyses were used to screen variables, and a predictive model was constructed using binary logistic regression. The model was visualized using a nomogram. Calibration curves and receiver operating characteristic curves were employed to assess the model’s accuracy and predictive value, while decision curve analysis was used to examine the model’s clinical relevance. Results: Overall, 209 ischemic stroke patients were included in the analysis. Univariate and multivariate logistic regression analyses identified higher NIHSS score and lower triglyceride level as independent factors associated with swallowing dysfunction (p-value < 0.05); the full model (NIHSS + TG) yielded AUCs of 0.685 (95% CI: 0.523–0.783) in the training set and 0.679 (95% CI: 0.606–0.700) in the test set, with corresponding sensitivities of 0.52 and 0.61, and specificities of 0.78 and 0.70, respectively. Compared with NIHSS alone (AUC 0.652, 95% CI: 0.518–0.746), the AUC improvement was not significant (DeLong p = 0.09), but the net reclassification improvement was 0.26 (95% CI: 0.02–0.49). Conclusions: This preliminary derivation model, based on NIHSS and triglyceride levels, demonstrates modest discrimination but provides incremental reclassification value over NIHSS alone for estimating the risk of PSD in cooperative patients. The model should be considered exploratory, and external validation is required before any clinical application.