Validation of an aspiration risk prediction model for Parkinson's disease based on nomogram: a single-center study.

Objective: This study aims to develop and validate a predictive model for aspiration risk in patients with Parkinson's disease (PD). Methods: A total of 160 inpatients with PD were enrolled (December 2022 to December 2023) from the Neurology Department of the Affiliated Drum Tower Hospital. Of 33 candidate variables, univariate analysis and Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression were used to identify key predictors and construct a clinical nomogram. Model discrimination and calibration were assessed using receiver operating characteristic (ROC) curves and calibration plots. Results: Univariate analysis and LASSO regression reduced the 33 variables to four core predictors: history of choking cough (odds ratio (OR) = 11.427; 95% confidence interval (CI) [2.187-59.709]), abnormal water-swallowing test results (OR = 4.262, 95% CI [1.496-12.140]), reduced facial expression (OR = 2.929, 95% CI [1.055-8.134]), and Barthel Index (OR = 0.972, 95% CI [0.950-0.995]). The area under the curve (AUC) values of the model were 0.882 (optimism-adjusted) and 0.950 for the training and testing sets, respectively. Calibration and decision curve analyses further validated the high performance and clinical utility of this model. Conclusion: This nomogram effectively stratified aspiration risk in patients with PD, facilitating earlier detection and intervention. Future studies including more clinical variables and larger multicenter cohorts are required to enhance the predictive accuracy and generalizability of the model.