Construction and validation of the risk prediction model for developing cognitive impairment in convales-cent stroke patients
10.3969/j.issn.1001-1242.2024.12.010
- VernacularTitle:恢复期脑卒中患者发生认知障碍的风险预测模型构建与评价
- Author:
Qianwen WANG
1
;
Lechang ZHAN
;
Yuting OUYANG
Author Information
1. 广州中医药大学,广东省 广州市,510120
- Publication Type:Journal Article
- Keywords:
stroke;
cognitive impairment;
risk factor;
Nomogram;
prediction model;
logistic model
- From:
Chinese Journal of Rehabilitation Medicine
2024;39(12):1810-1817
- CountryChina
- Language:Chinese
-
Abstract:
Objective:Cognitive impairment is one of the common complications of stroke,which can affect the rehabili-tation and their quality of life.It is very important to build reliable risk prediction model tools to detect post-stroke cognitive impairment(PSCI)in advance,but there is still no clinical risk prediction model for PSCI.Our aim was to identify the influencing factors of PSCI in convalescent stroke patients and construct a nomo-gram model for predicting the risk of PSCI based on these factors.Method:We retrospectively collected the demographic characteristics and clinically relevant data of convales-cent stroke patients hospitalized in Guangdong Provincial Hospital of Chinese Medicine from December 2019 to December 2022.Then we randomly divided the whole data set into the training set and the validation set according to 7:3,the former data was used to construct a nomogram model for predicting the risk of PSCI,and the latter data was used to evaluate the model performance.Univariate and multivariate logistic regression were used to analyze the factors affecting PSCI in convalescent stroke patients.Based on these factors,we used the R software to construct a PSCI risk prediction model who was visualized through a nomogram.The model performance was evaluated using the area under the curve(AUC),sensitivity,specificity,calibration curve,and decision curve analysis(DCA).Result:Our prediction model indicated that age,right hemiparesis,hypertension,coronary heart disease,hyper-homocysteinemia,Fugl-Meyer assessment scale(FMA)score,modified Barthel index(MBI)score and,mean cor-puscular hemoglobin were independent factors influencing the occurrence of PSCI in convalescent stroke pa-tients.The AUC,sensitivity and specificity of the model were 0.804,75.5％and 73.7％in the training set,and 0.737,82.9％and 62.8％in the validation set,suggesting that the model had a good discrimination.The calibra-tion curve of the training and validation sets indicated a good consistency between the prediction and the real observation.The decision curve analysis of the training and validation sets showed that the PSCI risk prediction model performed well in terms of the net clinical benefit.Conclusion:The PSCI risk prediction nomogram model constructed in this study can be personalize prediction of cognitive impairment probabilities in convalescent stroke patients,which can help healthcare providers to de-tect and treat PSCI early and improve patient outcome.
