Establishment And Verification Of A Mathematical Prediction Model To Evaluate The Probability Of Malignancy In Solitary Pulmonary Nodules

Objective:To estimate the probability of malignancy in patients with solitary pulmonary nodule(SPN)to establish a mathematical prediction model,and to verify the accuracy of the model,and provide guidance for the diagnosis and follow-up intervention of SPNs.Methods:A retrospective research in Qingdao Municipal Hospital Group collected 356 patients(186 males and 170 females)as group A with specific pathological diagnosis of solitary pulmonary nodules from January 2005 to December 2014 were used to establish the mathematical prediction model.Clinical data contained patient's essential information(age and gender),present illness(symptom and course),past medical history(previous cancer history and pulmonary disease),personal history(smoking history and smoking index),family history of tumor.Radiological data of SPN contained image features of CT scan(position,diameter,calcification,cavity,border,spiculation,lobulation,pleural retraction sign,vessel convergence sign).Another 84 SPN patients(49 males and 35 females)as group B with specific pathological diagnosis from January 2015 to December 2017 were used to verify the accuracy of the mathematical prediction model.After the receive operating characteristic(ROC)curve was drawn,the sensitivity and specificity of mathematical prediction model was tested by calculating the area under curve(AUC).Results: Through the Logistic regression analysis,age of patient,past medical history,a part of image features were independent predictors of malignancy in patients with SPN(P?0.05).The mathematical prediction model established by the retrospective study is: SPN malignancy predictive value = ex/(1+ex),x=-3.059 +(0.051×age)+(1.327×previous cancer history)+(1.065×pulmonary disease)+(0.627×diameter)+(1.129×spiculation)+(0.932×lobulation)+(-1.108×border)+(-1.982×calcification),"e" is natural logarithm.The sensitivity in group B was 89.8%,specificity was 80.0%,positive predictive value was 91.4%,and negative predictive value was 76.9%.The area under the ROC curve for the model was 0.855±0.027.Conclusion: Age of patient,previous cancer history,pulmonary disease,diameter,spiculation,lobulation,border,calcification were independent predictors of malignancy in SPN patients.Our mathematical prediction model is sufficient and accurate to pretest the malignancy of patients with SPN.