Background: Lung cancer remains the leading cause of cancer death worldwide, with lung adenocarcinoma (LUAD) being the most prevalent subtype of lung cancer. This study aimed to identify a lncRNA-mRNA regulatory module related to the prognosis, classification, and potential treatment of LUAD.Methods: Publicly available gene expression data of three cohorts were downloaded from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO) databases. Differential expression analysis between LUAD and normal samples, as well as the survival analysis, was performed. Protein-protein interaction (PPI) network and co-expression analyses were conducted to further identify key genes. A least absolute shrinkage and selection operator (LASSO) Cox regression model was developed to predict overall survival. Five machine learning models, including logistic regression, K-nearest neighbor (KNN), support vector machine (SVM), random forest, and extremely gradient boosting (XGBOOST), were trained to distinguish early-stage or epidermal growth factor receptor (EGFR)-mutation LUAD from others. Furthermore, connectivity map (CMap) and molecular docking analyses were performed to identify compounds with the ability to reverse the expression profiles of the key genes.Results: A cohort comprised of 535 LUAD and 59 normal samples in TCGA was used as the training set, while the GSE31210 and GSE30219 datasets were used as validation sets. 189 mRNAs and 11 lncRNAs were differentially expressed and associated with the overall survival. 43 hub mRNAs were further identified from the PPI network, and 3 lncRNAs were significantly correlated to the expression of hub mRNAs. Six genes with nonzero coefficients were selected by using the LASSO COX regression analysis, and the corresponding risk score was derived. The time-dependent ROC and Kaplan Meier analysis demonstrated that the risk score accurately discriminates the patients with a high or low risk. KNN and XGBOOST were the best models to recognize early-stage and EGFR-mutation LUAD, respectively. Purvalanol-a and Etoposide obtained a score of -99.98 in CMap and were successfully docked with three key genes.Conclusions: A lncRNA-mRNA regulatory module, including 4 mRNAs and 2 lncRNAs, was identified, and this module can facilitate the exploration of pathogenesis of LUAD and speed up the development of new treatments for LUAD patients.