Abstract
PURPOSE OF REVIEW: Radiotherapy (RT) is a mainstay treatment strategy for patients with radically treatable head and neck cancer (HNC). Efforts to personalize RT have spanned decades, yielding variable results across different treatment stages. The purpose of this review was to assess the potential of AI-driven models to bridge personalized radiotherapy strategies. RECENT FINDINGS: Radiomics, an emerging imaging analytics approach, provides significant quantitative features that can predict survival outcomes, treatment responses, and radiation-induced toxicity. Radiomics-based models in the studies demonstrate promising predictive efficacy with a high C-index or area under the curve (AUC) exceeding 0.70. AI-driven multimodal and multitemporal imaging models can stratify patients and guide risk-adapted radiotherapy. A four-step AI-driven RT framework may provide a template for future randomized controlled trials, supporting more trustworthy evidence.