Role of Artificial Intelligence in Radiation Oncology: A Missed Opportunity for LMICs?

Artificial Intelligence (AI) and Machine Learning (ML) are driving a major shift in modern medicine, particularly within radiation oncology. AI in radiation oncology offers low- and middle-income countries (LMICs) a way to offset critical shortages of oncologists and physicists by automating segmentation, planning, and quality checks, improving efficiency, and expanding equitable access to care. Its actual value depends on validation against LMICs' infrastructure and patient populations, an area where current evidence remains limited. This review aims to comprehensively summarize the current evidence on AI and ML applications across key domains of radiotherapy, while highlighting global disparities in adoption, particularly in LMICs.

This critical review conducted a comprehensive literature search across PubMed, Embase, and the Cochrane Library for studies published between 2010 and April 2026. Only relevant studies emphasizing the efficiency of AI in radiation workflow and the existing gaps in its adoption within LMICs were included in this review.

Eleven studies published between 2019 and 2025 were included. Study designs comprised original research (n=5), reviews (n=4), a survey (n=1), and a study protocol (n=1). Studies were conducted across multiple LMICs in Africa, Asia, and Latin America, predominantly evaluating AI-based contouring and treatment-planning tools, with Radiation Planning Assistant (RPA) being the most frequently reported platform. AI tools demonstrated comparable performance in LMICs relative to High-Income Countries (HICs), particularly with regard to time efficiency and the quality of target volume contouring. Yet the majority of LMICs still don't have AI tools in their radiation centers to work efficiently.

The current literature reveals a gap between the need for AI in radiation oncology in LMICs and the available evidence supporting its implementation. LMICs could benefit substantially from AI-enabled improvements in radiation oncology. Several conclusions about the applicability of these findings in LMIC settings were drawn by extrapolation from studies conducted in HIC contexts, which substantially limits their direct generalizability to lower-resource environments. Cost-effective and accessible AI models hold promise to address workforce shortages, improve efficiency, and expand equitable access to quality radiotherapy services.