Objective: This study aimed to develop a robust clinical prediction model for Poststroke cognitive impairment (PSCI) within 6 months following acute ischemic stroke (AIS) and subsequently validate its effectiveness.
Methods: A total of 386 AIS patients were divided into the PSCI group (174 cases) and the cognitively normal (CN) group (212 cases) based on the occurrence of PSCI. These patients were further categorized into two cohorts: 270 AIS patients in the training set, 116 AIS patients in the validation set. Multifactor logistic regression analysis was performed to identify independent predictors, which were then included in the prediction model for further analysis and validation. The performance of the prediction model was evaluated using the area under the receiver operating characteristic curve (AUC-ROC), calibration plots analyses to assess discrimination, calibration ability, respectively.
Results: Based on the selected variables (smoking, alcohol consumption, female gender, low education level, NIHSS score at admission, stroke progression, high systolic blood pressure, diabetes, atrial fibrillation, coronary heart disease, low-density lipoprotein cholesterol, β2-microglobulin, and Lp-PLA2), a clinical prediction model for the occurrence of PSCI within 6 months in AIS patients was constructed. The AUC-ROC of the model was 0.862, 0.806 in the training, validation sets, respectively. Calibration curve analyses and Hosmer-Lemeshow goodness-of-fit tests, along with other validation metrics, further demonstrated the model's good predictive performance.
Conclusion: The model exhibits high discriminative ability for PSCI and has substantial guiding value for clinical decision-making. However, further optimization of the model is required with multicenter data to enhance its robustness and applicability.
Keywords: Acute ischemic stroke; Atrial fibrillation; Clinical prediction model; Poststroke cognitive impairment.
Copyright © 2024 Elsevier Inc. All rights reserved.