Development And Validation A Clinical Prediction Model To Estimate For Pulmonary Nodules

Objective:By analyzing the clinical features,imaging features,laboratory test indexes and other data of patients,multivariate logistic regression analysis was used to construct the clinical diagnosis prediction model of benign and malignant probability of pulmonary nodules.Methods:Medical records from 240 Patients with a pathologic diagnosis of pulmonary nodules made between July 1,2017 and June 31,2020 were reviewed.Collect clinical characteristics(gender,age,cough,chest pain,hemoptysis,smoking history,time since quitting smoking,BMI,thoracic tumor history,history of COPD,history of tuberculosis,fungal infection history,family history of lung cancer),imaging features(nodule location,nodular nature,nodule sizey,border,Pleural retraction sign,Vascular convergence,surrounding cables,calcification,spiculation,lobulation,cavity,enhancement)and laboratory examination(Pro GRP,NSE,CEA,D-dimer,fibrinogen),pathology diagnosis,a total of 31 items,Univariate and multivariate Logistic regression analysis were performed to analyze independent predictors of malignancy in patients and establish a clinical diagnosis model for pulmonary nodules.Results:Monofactor analysis of gender,age,COPD,nodule size,pleural pull disease,vascular disease of goons,peripheral cords,edge,burr,blade,empty,gastric secrete element,CEA,fibrinogen and total protein statistically difference in benign and malignant of pulmonary nodules,multivariate Logistic regression analysis to filter out the gender,age,nodule size,edge,burr,fibrin originally independent predictors of malignancy,build a clinical prediction model:P=e ~x/(1+e ~x),X=-7.654+1.483×(Gender)+0.035×(age)+0.151×(Diameter)+1.588×(nature)+1.246×(fibrinogen)-2.708×(border)(Gender:female 1,male 0,border:blurred 1,clear 0,),including e as the natural logarithm,P=0.603 for a cutoff point,the sensitivity 86.05%,specificity of 72.22%,positive predictive value 88.10%,negative predictive value of 68.42%.The area under the curve is 0.892±0.05.Conclusions:Combined with the above analysis for pulmonary nodules,females,age,diameter,nature and fibrinogen are independent risk factors for malignancy,while blurred border is independent protective factor.The prediction model of benign and malignant pulmonary nodules was constructed by multivariate binary Logistic regression using the above indicators,with good accuracy and sensitivity.