@Article{or.2026.081398,
AUTHOR = {Farzana Siddique, Mohamed Shehata, Mohammed Ghazal, Guruprasad Giridharan, Sohail Contractor, Ayman El-Baz},
TITLE = {Comparative Evaluation of AI-Based and Non-AI Genomic Biomarkers for Predicting Lung Cancer Treatment Response: A Systematic Review},
JOURNAL = {Oncology Research},
VOLUME = {},
YEAR = {},
NUMBER = {},
PAGES = {{pages}},
URL = {http://www.techscience.com/or/online/detail/27878},
ISSN = {1555-3906},
ABSTRACT = {<b>Objectives:</b> Lung cancer remains the leading cause of cancer-related mortality worldwide, and treatment response varies substantially across patients. Conventional genomic biomarkers such as EGFR, PD-L1, and tumor mutational burden (TMB) are widely used in precision oncology but have important limitations. This systematic review aimed to evaluate and compare conventional genomic biomarkers and artificial intelligence (AI)-based genomic models for predicting treatment response in lung cancer. <b>Methods</b>: This review was conducted according to PRISMA 2020 guidelines. PubMed, Scopus, Google Scholar, and ResearchGate were searched for studies published between January 1, 2014, and March 15, 2026. Original studies evaluating genomic biomarkers or AI-based genomic prediction models in relation to lung cancer treatment outcomes were included. Due to substantial heterogeneity across studies, findings were synthesized narratively. <b>Results:</b> Fifty-four studies met the inclusion criteria. Established biomarkers including EGFR, ALK, PD-L1, and TMB remained strongly associated with treatment selection and clinical outcomes, particularly in targeted therapy and immunotherapy settings. AI-based models frequently demonstrated strong predictive performance, with reported AUC values ranging from 0.70 to 0.97. However, most AI studies were retrospective and had limited prospective or external validation. <b>Conclusion</b>: Conventional genomic biomarkers remain central to current lung cancer treatment decisions because of their established clinical validity. AI-based genomic models show promise for improving prediction and patient stratification, but their clinical utility remains uncertain without stronger prospective validation.},
DOI = {10.32604/or.2026.081398}
}