This study aims to construct an early diagnostic biomarker panel and risk stratification model for anti-tumor drug-related ILD through integrative analysis of multi-omics data including genomics, transcriptomics, proteomics, and metabolomics. Using baseline and post-treatment longitudinal samples collected from a multi-center prospective cohort, we will apply machine learning to screen for stable and reproducible feature sets and evaluate their sensitivity, specificity, and clinical applicability in an independent validation cohort. The goal is to achieve early identification and stratified management of ILD, optimize treatment decisions, reduce the incidence of severe adverse events, and improve patient survival and quality of life.