Development and Internal Validation of a Clinical Prediction Model for Trousseau Syndrome in Patients Presenting with Acute Ischemic Stroke

Abstract Background To identify clinical factors associated with Trousseau syndrome presenting as acute ischemic stroke (AIS-TS) and to develop and validate a risk prediction model for AIS-TS, providing a basis for early clinical identification and risk stratification. Methods This retrospective study included 51 consecutive patients with AIS-TS who were admitted to the Affiliated Hospital of Jiangsu University between July 2018 and December 2025 as the case group. A control group comprising 204 patients with metastatic solid malignancies without thromboembolic events was selected using 1:4 matching based on admission period. A total of 21 clinical variables were collected, including demographic characteristics, tumor types, and laboratory parameters. Candidate predictors were screened using univariate logistic regression, followed by LASSO regression for dimensionality reduction. A comprehensive scoring approach integrating univariate P-values, random forest variable importance, absolute LASSO coefficients, and odds ratio deviations was applied to determine the final predictive variables. A nomogram was constructed based on the selected variables and internally validated using Bootstrap resampling. Model performance was assessed using the calibration C-index, calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC). An online prediction calculator was developed using the R Shiny platform to facilitate individualized risk assessment. Results Univariate logistic regression identified significant differences between groups in age, white blood cell count, platelet count, C-reactive protein (CRP), D-dimer, neutrophil count, carbohydrate antigen 50 (CA50), CA125, CA15-3, CA19-9, and hypertension (all P < 0.05). After collinearity assessment and LASSO regression screening, four predictors—D-dimer, platelet count, neutrophil count, and hypertension—were retained. Multivariate logistic regression demonstrated that elevated D-dimer levels (OR = 1.094, 95% CI: 1.039–1.153, P < 0.001), increased neutrophil count (OR = 1.123, 95% CI: 1.002–1.259, P = 0.047), and hypertension (OR = 2.375, 95% CI: 1.189–4.745, P = 0.014) were associated with an increased risk of AIS-TS, whereas lower platelet counts were also associated with higher AIS-TS risk (OR = 0.993, 95% CI: 0.989–0.998, P = 0.005). Bootstrap internal validation yielded a calibrated C-index of 0.773 (95% CI: 0.72–0.87), and calibration curves demonstrated good agreement between predicted and observed probabilities. DCA and CIC indicated favorable clinical net benefit across certain threshold probability ranges. An online AIS-TS risk prediction calculator was successfully developed using the Shiny platform. Conclusion D-dimer, neutrophil count, platelet count, and hypertension are important factors associated with AIS-TS. The developed prediction model demonstrates good discrimination, calibration, and clinical utility, and may facilitate early identification and risk stratification of patients at increased risk of AIS-TS. The Shiny-based online calculator further improves model accessibility and enables convenient individualized risk assessment for patients with AIS-TS.