Abstract:Objective To systematically compare the predictive efficacy of different cardiovascular disease risk assessment models in university staff and identify an optimized tool for precision health management. Methods A retrospective cohort study was conducted，including university staff who consecutively underwent health examinations at Peking University Third Hospital from January 1，2019，to December 31，2022. Ten-year cardiovascular disease risks were assessed using six prediction models：Framingham -LDL，Framingham-TC，WHO/ISH Charts，China-PAR，ICVD，and CVDRC. Inter-model agreement was analyzed via weighted Kappa tests，with carotid ultrasound findings（carotid intima-media thickness thickening and plaque formation）serving as endpoint events. Predictive performance was evaluated using receiver operating characteristic（ROC）curves. Results A total of 556 university staff members were enrolled，comprising 314 males and 242 females，with a mean age of 58.4 years. Significant differences were observed in medium-to-high-risk detection rates among models（China-PAR：73.7%；ICVD：3.7%，P<0.05）. Inter-model consistency analysis revealed the highest agreement between Framingham-LDL and Framingham-TC（Kappa=0.761），while China-PAR showed the poorest consistency with WHO/ISH Charts and ICVD（Kappa=0.082，0.022）.ROC analysis demonstrated that China - PAR achieved the highest area under the curve（AUC=0.791）in predicting carotid atherosclerosis，significantly outperforming other models （P<0.05）. Framingham-LDL （AUC=0.629） and Framingham-TC（AUC=0.624）ranked second，while the remaining models had AUC values <0.6. Conclusion The China-PAR model exhibits optimal performance in predicting atherosclerotic risk among university staff and is highly concordant with CVD phenotypes. It is recommended as the core tool for risk stratification in this population，providing a scientific foundation for targeted intervention strategies.