Construction of risk prediction model for delayed encephalopathy in acute carbon monoxide poisoning based on SMOTE algo-rithm
TAN Qianmei
YANG Jing
LI Qiuping
Abstract：Objective To build a risk prediction model for delayed encephalopathy after acute carbon monoxide poisoning(ACMP)based on the SMOTE algorithm.Methods A total of 180 ACMP patients were grouped according to whether or not they developed delayed encephalopathy after ACMP.The clinical data of the patients were analyzed to screen the influencing factors of delayed encephalopathy after ACMP,and then the original dataset of the influencing factors was reconstructed using the SMOTE algorithm in order to obtain its risk warning model and to validate the predictive efficacy of the model.Results Thirty two of 180 patients had delayed encephalopathy after ACMP,with an incidence rate of 17.78%.Age＞60 years,abnormal head MRI,no hyperbaric oxygen therapy,GCS score＜9 points,coma time＞24 h and CK＞200 U/L were risk factors for delayed encephalopathy after ACMP(P＜0.05).The original data early warning model(P1 =1.491X1+1.429X2+1.053X3+1.055X4+1.292X5+1.330X6-4.157)and the early warning model based on SMOTE algorithm(P2=1.433X1+1.229X2+1.215X3+1.063X4+1.402X5+1.106X6-5.224)were built.The area under AUC curve of the original data prediction model was 0.805,which was lower than the risk prediction model based on SMOTE algorithm 0.834.Conclusion Age＞60 years old,abnormal brain MRI,no hyper-baric oxygen therapy,GCS score＜9 points,coma time＞24 h,and CK＞200 U/L are risk factors for delayed encephalopathy after ACMP.The risk prediction model based on SMOTE algorithm has high predictive efficiency.
Keywords：SMOTE algorithmcarbon monoxide poisoningrisk factorsdelayed encephalopathyprediction
Publication Date：2023-09-10
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 760-763,780 )