Clinical Interventions in Aging (Aug 2024)
Development of a Predictive Nomogram for Intra-Hospital Mortality in Acute Ischemic Stroke Patients Using LASSO Regression
Abstract
Li Zhou,1,* Youlin Wu,1,2,* Jiani Wang,1 Haiyun Wu,1 Yongjun Tan,1 Xia Chen,1,3 Xiaosong Song,1,4 Yu Ren,1 Qin Yang1 1Department of Neurology, the First Affiliated Hospital of Chongqing Medical University, Chongqing, People’s Republic of China; 2Department of Neurology, Chongzhou People’s Hospital, Sichuan, People’s Republic of China; 3Department of Neurology, the Seventh People’s Hospital of Chongqing, Chongqing, People’s Republic of China; 4Department of Neurology, the Ninth People’s Hospital of Chongqing, Chongqing, People’s Republic of China*These authors contributed equally to this workCorrespondence: Qin Yang, Department of Neurology, the First Affiliated Hospital of Chongqing Medical University, 1 Youyi Road, Yuzhong District, Chongqing, 400016, People’s Republic of China, Tel +86-023-89012008, Fax +86-023-68811487, Email [email protected] and Purpose: Ischemic stroke is a leading cause of mortality and disability globally, necessitating accurate prediction of intra-hospital mortality (IHM) for improved patient care. This study aimed to develop a practical nomogram for personalized IHM risk prediction in ischemic stroke patients.Methods: A retrospective study of 422 ischemic stroke patients (April 2020 - December 2021) from Chongqing Medical University’s First Affiliated Hospital was conducted, with patients divided into training (n=295) and validation (n=127) groups. Data on demographics, comorbidities, stroke risk factors, and lab results were collected. Stroke severity was assessed using NIHSS, and stroke types were classified by TOAST criteria. Least absolute shrinkage and selection operator (LASSO) regression was employed for predictor selection and nomogram construction, with evaluation through ROC curves, calibration curves, and decision curve analysis.Results: LASSO regression and multivariate logistic regression identified four independent IHM predictors: age, admission NIHSS score, chronic obstructive pulmonary disease (COPD) diagnosis, and white blood cell count (WBC). A highly accurate nomogram based on these variables exhibited excellent predictive performance, with AUCs of 0.958 (training) and 0.962 (validation), sensitivities of 93.2% and 95.7%, and specificities of 93.1% and 90.9%, respectively. Calibration curves and decision curve analysis validated its clinical applicability.Conclusion: Age, admission NIHSS score, COPD history, and WBC were identified as independent IHM predictors in ischemic stroke patients. The developed nomogram demonstrated high predictive accuracy and practical utility for mortality risk estimation. External validation and prospective studies are warranted for further confirmation of its clinical efficacy.Keywords: ischemic stroke, nomogram, predictors, lasso, intra-hospital mortality