Identification of lactylation-related genes associated with heart failure via bioinformatics and machine learning approaches.

BACKGROUND: Heart failure (HF) is the terminal stage of cardiovascular disease with high global prevalence and poor prognosis. This study aimed to identify lactylation-related genes as novel diagnostic biomarkers for HF using bioinformatics and machine learning approaches. METHODS: Gene expression data of HF patients and healthy controls were obtained from the GEO database. Combined with lactylation-related genes (LRGs) from literature, differentially expressed genes and weighted gene co-expression network analysis (WGCNA) were used to identify HF-LRGs. A diagnostic model was constructed using machine learning and validated with external datasets and animal experiments. RESULTS: Four hub genes (HLTF, HMGN2, ARGLU1, and LSP1) were identified, all significantly upregulated in HF patients. Nine machine learning models achieved AUC values > 0.9, demonstrating high diagnostic accuracy. The expression trends were validated in the GSE84796 dataset and in TAC mouse hearts, where qRT-PCR confirmed upregulation of all four genes. GSEA linked hub genes to multiple metabolic and disease-related pathways. CONCLUSION: The identified lactylation-related genes, particularly HLTF, HMGN2, ARGLU1, and LSP1, serve as potential diagnostic biomarkers for heart failure. The machine learning-based diagnostic model shows high accuracy and clinical application potential, offering new perspectives for early diagnosis and treatment of HF.