SHI Jiayun, LIU Xiaofeng, LIANG Linlong, XlE Qifang

( Department of Pediatrics, the Affliated Changsha Central Hospital, Hengyang Medical College.

University of South China, Changsha 410004, China)

A bstract0bjective 'To establish a diagnostie model for children with myeoplasma pneumoniae ( MP )eomplicatedwith lobar pneumonia by Logisties regression. Methods The clinical data of 239 children hospitalized withMycoplasma pneumoniae between June 2018 and June 2023 were retrospectively analyzed and divided into theconventional pneumonia group (78 cases) and the lobar pneumonia group ( 161 cases). Lasso regression model wasutilized to screen the relevant features of lobar pneumonia, and Logistics regression was applied to construct adiagnostic prediction model. Results Univariate analysis revealed that age, cough history, white blood cells count( WBC), neutrophil count ( NEU ), platelets ( PLT ), lactate dehydrogenase ( LDH) were associated with lobalpneumonia. The age, cough history, WBC , NEU, CRP, LDH, and bolus screened by lasso regression were stronglyassociated with lobar pneumonia. Logistics regression analysis showed that age, cough history, WBC, LDH, andbolus were independent diagnostic factors in patients with lobar pneumonia. 'The Nomogram model showed gooddiserimination and clinical utility in distinguishing pediatric MP from lobar pneumonia. differentiation and clinicalutility. Conclusion Lasso and Logistic regression are studied to construet a column-line graph model, and age, coughhistory, WBC, LDH, and talk embolism are found to be independent diagnostic factors for MP complicated withlobar pneumonia. And the model has a C-index of 0.803, which is highly accurate and helps clinical decision-making.

Keywords: mycoplasma pneumoniae; children ; pneumonia ; lobar pneumonia; diagnostic models

DOI:10.19296/j.cnki.1008-2409.2024-02-022

PDFdownload：click to download