基于NHANES数据库的3-苯氧基苯甲酸与心血管疾病关联研究及XGBoost预测模型构建OA

Objective:To explore the association between the typical metabolite of pyrethroid pesticides,3-phenoxybenzoic acid(3-PBA),and cardiovascular diseases(CVD),and to construct a CVD risk prediction model for sensitive populations based on the ensemble learning model XGBoost(Extreme Gradient Boosting).Method:This study was based on the data from the National Health and Nutrition Examination Survey(NHANES)of the United States from 2007 to 2012,and combined the 3-PBA concentration in urine(corrected by creatinine and logarithmic conversion),the history of CVD,and related covariates for research.Firstly,the association between 3-PBA exposure and CVD was evaluated through multiple logistic regression analysis,and stratified analysis was conducted in different age groups.Further,in the age groups where the logistic regression results showed a significant correlation with log_PBA,the SMOTE technique was used to balance the samples,and the XGBoost model was constructed to predict the CVD risk,and its accuracy rate,recall rate,precision rate,F1 score and AUC were evaluated.Result:logistic regression analysis showed that in the overall population,log_PBA had no statistical correlation with CVD,but in the elderly population aged≥65 years,it was significantly positively correlated(OR=1.55,95％CI:1.05～2.30,P<0.05).The XGBoost model constructed based on this population performed well in the test set:an accuracy rate of 80.8％,a positive recall rate of 83.1％,a positive precision rate of 79.3％,an F1 score of 81.2％,and an AUC of 0.886.Conclusion:This study shows that the level of 3-PBA in urine has a significant positive impact on the CVD risk in the elderly population,and the XGBoost model can effectively identify high-risk individuals.The results support the inclusion of environmental exposure factors in the chronic disease risk assessment system,which can provide data support for the early identification and intervention of CVD in the elderly population.