Artificial intelligence in non-small cell lung cancer: transforming diagnosis, treatment, and prognostic evaluation
Abstract
Non-small cell lung cancer (NSCLC) is characterized by profound biological heterogeneity, frequently leading to late-stage diagnosis, recurrence, or metastasis. Conventional clinical pathways heavily depend on subjective visual interpretations, invasive tissue biopsies, and experience-based medical models that constrain individualized precision medicine. Identifying noninvasive, cross-scale tools for whole-disease-course assessment is critical for optimizing clinical decision-making and survival outcomes. This review critically synthesizes recent research progress and core algorithmic applications of artificial intelligence (AI) across the diagnostic, therapeutic, and prognostic continuum of NSCLC. In diagnostics, AI models enable rapid morphological screening and tumor subtyping on digital pathology slides, whereas multimodal imaging radiomics extracts microscopic features from macroscopic CT, PET/CT, and MRI scans to noninvasively identify tumor heterogeneity and map metastatic risk. Therapeutically, machine learning and large language models predict treatment responses, characterize tumor microenvironment dynamics, optimize perioperative surgical strategies, and accelerate drug discovery by screening chemical databases and decoding candidate drug mechanisms. Prognostically, multiscale fusion models integrate liquid biopsies, pathomics, radiomics, and clinical records to provide precise risk stratification for recurrence and survival outcomes. While AI has demonstrated transformative potential across the entire clinical lifecycle of NSCLC, significant technical bottlenecks remain, including severe data heterogeneity across healthcare institutions, the lack of clinical transparency in end-to-end “black box” architectures, and low-throughput biological validation. Future research must focus on standardizing data protocols, improving explainable AI, and incorporating multidimensional time series data to realize true intelligent precision medicine.
