The Construction And Validation Of Predictive Model And Scoring System For Vascular Embolism In Cardiac Myxoma Patients

Objective:To explore the risk factors of vascular embolism?VE?in patients with cardiac myxoma?CM?,and establish predictive model and scoring system,evaluate and verify the model and scoring system.And provide a research basis for the prevention of vascular embolization events.Methods:A total of 300 CM patients was enrolled into this study between January 2000and January 2020 in the second affiliated hospital of Nanchang university.The selected patients were randomly divided into derivation group?210 cases,70%?and validation group?90 cases,30%?.Patients with VE were included in the case group,the remaining patients were included in the control group.Collect clinical data,preoperative hematological data and preoperative echocardiographic data of all CM patients.Univariate analysis was carried out for all variables,and significant variables in univariate analysis were included in multivariate analysis,independent risk factors were screened out and a binary logistic regression model was established.A scoring system was developed based on regression coefficients from the final multivariate model.Calibration and discrimination of the logistic regression model and scoring system were assessed by Hosmer-Lemeshow?H-L?test and the area under the receiver operating characteristic curve?AUC?respectively both in the derivation group.According to the model and scoring system,the prediction probability and total risk score of each patient in the validation group were calculated.Calibration and discrimination of model and scoring system were assessed by H-L test and the AUC.The AUC of the logistic regression model and scoring system was compared by U test.The AUC of derivation group and validation group were compared by U test.Results:1.Multivariate analysis showed that:age<45years,smoking,atrial fibrillation,tumor morphology and tumor structure are independent risk factors for VE in CM patients.2.The prediction model established was:logitP=-4.929+1.255X₁+1.041X₂+2.079X₃+1.679X₄+1.781X₅?X₁=age,X₂=smoking,X₃=atrial fibrillation,X₄=tumor morphology,X₅=tumor structure?.The H-L test showed that P>0.05,which indicates that the model fits well.The AUC of the model is 0.849,which indicates that the model has a good prediction effect.3.A scoring system based on the above 5 variables,score ranging from 0-5points.Among them were younger than 45 years?1 point?,smoking?1 point?,atrial fibrillation?1 point?,morphology?1 point?,and structure?1 point?.The AUC of the scoring system was 0.821,indicating that the scoring system had a good prediction effect.2 points was the best diagnostic cutoff value,the sensitivity,specificity,consistent rate was 85.71%,63.10%,67.62%.4.In the validation group,model showed good calibration and discriminative power.The H-L test's P value was more than 0.05 and the AUC was 0.785.The scoring system also showed good discriminative power,and the AUC was 0.766.2points was the best diagnostic cutoff value,the sensitivity,specificity,consistent rate was 77.28%,66.18%,68.89%.5.The U test of the AUC was not statistically significant between logistic regression model and scoring system in the derivation group and validation group?all P>0.05?.6.The U test of the AUC was not statistically significant between derivation group and validation group in the logistic regression model and scoring system?all P>0.05?.Conclusion:1.Age<45years,smoking,atrial fibrillation,tumor morphology,and tumor structure are independent risk factors for VE in patients with CM.2.The prediction model established was:logitP=-4.929+1.255X₁+1.041X₂+2.079X₃+1.679X₄+1.781X₅?X₁=age,X₂=smoking,X₃=atrial fibrillation,X₄=tumor morphology,X₅=tumor structure?.A scoring system based on the above 5 variables,score ranging from 0-5 points.Among them were younger than 45 years?1 point?,smoking?1 point?,atrial fibrillation?1 point?,morphology?1 point?,and structure?1point?.3.The risk prediction model and scoring system established in this study have good fit and prediction effect,and have good clinical application value,and provide a research basis for the prevention of vascular embolization events.