[Detection Status and Cost Analysis of Community Pulmonary Nodule Screening with Mobile LDCT Combined with AI].

Abstract

Background: Lung cancer is the malignant tumor with the highest morbidity and mortality in China. Low-dose computed tomography (LDCT) is the core method for early lung cancer screening, but traditional fixed-institution screening has the problem of insufficient accessibility at the grassroots level. This study aims to evaluate the detection characteristics and influencing factors of pulmonary nodules by mobile LDCT combined with artificial intelligence (AI) in urban communities of Wuhan, and conduct a preliminary cost analysis from the health system perspective, so as to provide scientific basis for optimizing early screening strategies for pulmonary nodules at the primary care level.

Methods: A community‑based observational study between August 2024 and May 2025 were conducted. A total of 3436 residents in Wuhan communities completed questionnaire surveys and mobile LDCT scans, with a recruitment response rate of 26.7%. Pulmonary nodules were classified according to the Lung CT Screening Reporting and Data System (Lung-RADS). Multivariate Logistic regression model was used to analyze the independent influencing factors of clinically significant nodules (Lung-RADS category ≥3). Sensitivity analysis was performed by stepwise backward full-variable model, and multicollinearity was tested by variance inflation factor (VIF). Meanwhile, a descriptive cost analysis was conducted in accordance with the CHEERS 2022 statement.

Results: The median age of the screened population was 63 years, with 38.7% male and 67.3% aged ≥60 years. The overall detection rate of pulmonary nodules was 84.6%, with clinically significant nodules detected in 10.4% and high-risk nodules in 3.2%. Multivariate Logistic regression analysis showed that age 60-79 years (OR=2.771, 95%CI: 1.004-7.646, P=0.049), age ≥80 years (OR=6.224, 95%CI: 1.813-21.369, P=0.004), heavy smoking (OR=1.919, 95%CI: 1.352-2.724, P<0.001), secondhand smoke exposure (OR=1.762, 95%CI: 1.200-2.586, P=0.004), exposure to harmful chemicals (OR=2.972, 95%CI: 1.881-4.695, P<0.001), long-term kitchen oil fume exposure (OR=1.411, 95%CI: 1.102-1.807, P=0.006), and family history of lung cancer (OR=1.866, 95%CI: 1.173-2.967, P=0.008) were independent risk factors for clinically significant nodules. The total cost of the project was 666,580 CNY, with an average screening cost of 194 CNY per person. Under the whole population strategy, the costs for detecting one clinically significant nodule and one high-risk nodule were 1857 and 6060 CNY, respectively. Screening targeted at individuals aged ≥60 years could cover 87.3% of high-risk nodules, with the cost per high-risk nodule detected reduced to 4676 CNY.

Conclusions: Mobile LDCT combined with AI-assisted screening can effectively identify high-risk individuals of pulmonary nodules in community settings. The elderly and people with multiple exposure risks should be the priority population for screening. This study provides localized screening efficiency parameters and cost data of this model, which can provide practical basis for the optimization of primary pulmonary nodule early screening strategies and subsequent complete health economic evaluation.