Construction and Validation of Immune prognosis model for lung adenocarcinoma based on machine learning

Jinyu ZhengXiaoyi XuMiao FuJie Yang^*

• Affiliated Jinhua Hospital, Zhejiang University School of Medicine, Jinhua, China

Lung adenocarcinoma is a common malignant tumor characterized by high rates of recurrence and metastasis, underscoring the urgent need for novel prognostic biomarkers. This study integrated transcriptomic and clinicopathological data to construct an effective immune-related prognostic model. A total of 55 tumor and 38 adjacent normal tissue samples from the TCGA database were used as the training set, with 30 additional samples for model validation. An independent external validation cohort consisting of 10 LUAD samples and their paired normal tissues was obtained from the First Affiliated Hospital of Wenzhou Medical University. Differential expression analysis and weighted gene co-expression network analysis identified key gene modules associated with LUAD progression. Functional enrichment analysis revealed that these genes are primarily involved in cell proliferation and immune-inflammatory responses. By intersecting immune-related genes from the IMMPORT database, 68 prognosis-associated genes were identified. Using Random Forest, LASSO regression, and SVM-RFE algorithms, four hub genes—CBLC, GDF10, LTBP4, and FABP4—were selected to construct the prognostic model. The model demonstrated strong predictive performance through multivariate Cox regression and ROC analysis, and its effectiveness was further validated using an artificial neural network. Immune infiltration analysis showed increased levels of CD4+ T cells, macrophages, and dendritic cells in LUAD, with significant differences in immune subtypes and survival outcomes. This study provides new insights into the immunobiology of LUAD and lays the foundation for the development of personalized therapeutic strategies.