Analysis Of Risk Factors For Malignant Pulmonary Nodules And The Construction Of A Clinical Prediction Model

Objective:A predictive model of benign and malignant isolated pulmonary nodules based on imaging and clinical data and analyzed by statistical methods(univariate and multivariate)was established,aiming to provide objective model support for clinicians to help them judge the benign and malignant properties of isolated pulmonary nodules and guide them to choose the next intervention measures.Methods:Data on 143 patients with isolated pulmonary nodules diagnosed in the Fifth Affiliated Hospital of Xinjiang Medical University during February 2021 and February 2022 were collected,of which 70%were treated as the training group(n=100)and the remaining 30%as the validation group(n=43).Comparing the imaging and clinical data of patients with benign and malignant isolated pulmonary nodules,logistic regression analysis was used to screen for risk factors with independent influence on the benign and malignant signs of isolated pulmonary nodules.Use the training group data to build a mathematical model to predict the benign and malignant properties of isolated lung nodules.The operating characteristic curve of the participants(ROC curve)was drawn using the validation group data to evaluate the prediction ability of the model,and the area under the curve(AUC)was calculated with the Mayo model and the Peking University People’s Hospital(PKUPH)model.Results:Data from the training groups were analyzed using univariate regression analysis methods,Significant differences between the maximum diameter of the malignant nodules and some imaging features of the nodules and the benign pulmonary nodules were also found(P<0.05),Based on these data,A multivariate regression analysis was performed,The mathematical model of benign and malignant prediction of isolated pulmonary nodules was constructed:P=value of malignant prediction of isolated pulmonary nodules=e~x/(1+e~x),X=-15.349+1.256*nodule maximum diameter+4.021*burr sign+1.134*calcification+5.049*nodule position left upper lobe of the lung,The best cut-off value of the prediction model for bringing the validation set into this study is 0.448,Sensitivity 0.905,specificity 0.678,Area under the receiver operating characteristic(ROC)curve of the model(AUC=0.839).Conclusion:The predictive model developed in this study can be used to evaluate the malignant probability of isolated pulmonary nodules to aid in the diagnosis of the disease and the selection of subsequent interventions.The prediction model can be used as a tool to guide clinical decision-making.