The role of artificial intelligence in oncology and drug discovery

Artificial intelligence (AI) methods represent one of the most rapidly evolving domains in medical research and patient care. Large-scale datasets derived from modern imaging studies, omics analyses, and electronic health records enable the deployment of predictive and decision-support systems that may provide more accurate and consistent results than conventional statistical approaches. This review offers clinicians and researchers in the field of oncology a methodological overview of key AI-related areas, including radiological and pathological image analysis, biomarker-based risk assessment, practical applications of generative and large language models, and data-driven approaches for predicting therapeutic efficacy and side effects in drug discovery. We highlight validated tools and platforms (such as MONAI, Clara Train, and QuPath) that can reliably support diagnostics, interpretation, or specific stages of patient care. We also briefly discuss the limitations of predictive models, the role of interpretability, and practical risks – such as automation bias or the uncertain extrapolations of generative models – that require caution. Overall, the aim is to outline a methodological framework that assists clinicians in the evaluation and evidence-based implementation of AI-driven tools.