Effectiveness Evaluation of a Regional Comprehensive Prevention and Control Network,along with Epidemiological Characteristics and Risk Prediction Model Construction for Ischemic Stroke in Shanxi Province
WANG Zhijun
WANG Xin
DING Zhibin
PANG Qian
LIU Xiaolei
YANG Libai
ZHAO Hongping
LI Xinyi
Abstract：Objective:To investigate the epidemiological dynamics of ischemic stroke(IS)in Shanxi Province,quantify the effects of major risk factors,develop and validate a localized machine learning-based risk prediction model,and evaluate the preliminary effectiveness of an innovative regional comprehensive prevention and control network,thereby providing high-level evidence for targeted public health strategies.Methods:A population-based retrospective analysis of data from the Shanxi Health Information Platform(2015-2023)was conducted to describe the spatiotemporal distribution of standardized ischemic stroke incidence,prevalence,and mortality.A 1:1.2 frequency-matched case-control study(3 528 IS patients,4 150 controls)with multivariate Logistic regression to assess the impact of traditional and lifestyle-related risk factors on IS risk.Development of an IS risk prediction model with a Gradient Boosting Decision Tree(GBDT)algorithm based on case-control data,with its discrimination and calibration were assessed through 10-fold cross-validation and an independent test set.Finally,a quasi-experimental pre-post design was used to evaluate the one-year effectiveness of a"three-tier linkage,information synergy,and integrated prevention-treatment"network piloted in Taiyuan and Datong.Results:From 2015 to 2023,the standardized IS incidence in Shanxi increased from 185.6 to 240.2 per 100 000,and prevalence rose from 1 350.2 to 1 898.1 per 100 000,indicating a continuously growing disease burden.Multivariate analysis identified hypertension as the strongest independent risk factor(OR=3.52,95%CI 3.18-3.90),followed by atrial fibrillation or valvular heart disease(OR=3.15,95%CI 2.48-4.00)and family history of stroke(OR=2.45,95%CI 2.15-2.79).The GBDT risk prediction model showed excellent performance on the test set,with an area under the receiver operating characteristic(ROC)curve(AUC)of 0.923,it is superior to the traditional Logistic regression model.After the network pilot,the regional mean door-to-needle time(DNT)decreased significantly from 75 to 52 minutes,the intravenous thrombolysis rate within 4.5 hours increased from 12.5%to 18.9%,and the standardized management rate of high-risk populations improved from 35.6%to 62.3%(P＜0.01).Conclusion:The IS disease burden in Shanxi Province is rapidly increasing,with a risk profile characterized by distinct regional features.The localized machine learning-based prediction model can effectively improve early screening.The regional comprehensive control network constructed in this study has proven highly effective in optimizing treatment efficiency and strengthening primary prevention,offering a replicable,systematic solution of major chronic disease challenges.
Keywords：ischemic strokeepidemiologyrisk factorsmachine learningprediction modelprevention and control systemShanxi Province
Publication Date：2026-02-25
Online Publishing Date：2026-03-16(First online date of this platform, not the publication date of the document)
Pages：6( 608-613 )