Artificial intelligence now touches nearly every stage of cancer care, but its applications are not interchangeable. AI helps researchers find new drug targets, helps clinicians detect tumors in scans, and helps tailor chemotherapy dosing to individual patients—three distinct jobs that rely on related but different techniques. Understanding which application does what is the key to making sense of the field.
Introduction: The Transformative Role of AI in Oncology
Reviews of AI in oncology describe a shift from empirical, trial-and-error approaches toward data-driven ones. A 2025 review in Frontiers in Oncology by Caixia Fang and colleagues frames AI as a thread running through the entire oncology drug lifecycle: it sharpens the identification of drug targets, transforms drug discovery through virtual screening and molecular design, improves patient selection in clinical trials, and enables personalized treatment plans and dynamic dosing strategies tailored to each patient's profile.
That breadth is exactly why the field can feel confusing. "AI in cancer care" can mean a generative model proposing a new molecule, a deep learning algorithm flagging a lesion on a CT scan, or a model predicting whether a specific patient will suffer severe side effects from a chemotherapy regimen. A review in npj Precision Oncology, authored by Jun Li and colleagues, notes that diagnostics—particularly radiology and pathology—account for 80% of AI applications in oncology, with AI algorithms often faster and more accurate than human experts at interpreting complex imaging data from MRIs, CT scans, and X-rays. The remaining applications cluster around discovery, prediction, and treatment planning.
AI in Drug Discovery: Accelerating Innovation
In drug discovery, the goal is speed and precision at the earliest stage: identifying promising targets and designing molecules that act on them. A review of AI-driven innovations in oncology drug discovery describes how machine learning and deep learning are applied across target validation, drug repositioning, de novo drug design, and clinical trial optimization, enabling rapid identification of new compounds with optimized dosages and reduced toxicity.
The concrete techniques include computer-aided drug design, generative AI, and high-throughput screening, alongside molecular modeling and simulation. Researchers at the Kimmel Cancer Center, including Dr. Berger, describe AI analyzing large volumes of genetic data to identify which patients are most likely to benefit from a particular therapy—an approach they expect to yield faster approvals, lower costs, and treatments that reach patients sooner.
Dana-Farber Cancer Institute illustrates the mechanism with a vivid analogy: proteins are "locks" and drugs are "keys." For a drug to work, scientists must find a specific pocket on a protein's surface where a small molecule fits precisely. Dana-Farber reports on an AI tool, AF2BIND, built on AlphaFold2 protein structure prediction, that could speed the discovery of such binding sites—potentially identifying new treatment targets faster, especially for cancers with limited options.




