Key points are not available for this paper at this time.
AbstractBackground Medical radiation science (MRS) research faces a growing asymmetry between a small body of high-rigour, statistically robust studies and an expanding volume of lower-rigour investigations that are under-powered, methodologically fragile or poorly reported. These trends, compounded by declining statistical oversight in peer review and persistent misinterpretation of statistical outputs, risk creating a research ecosystem that is data-rich but evidence-poor. These issues extend across nuclear medicine, radiography and radiation therapy contexts in which clinically valuable data are generated but are not consistently translated into methodologically robust and cumulative evidence. Purpose To discuss how generative artificial intelligence (GenAI) can function as a methodological support system to address structural, cultural and statistical weaknesses across the research lifecycle, with emphasis on planning, analysis, interpretation and reporting. Methods A conceptual framework is presented that maps GenAI capabilities onto stages of the statistical workflow. The distinction between GenAI as a reasoning scaffold and as an analysis engine is discussed, alongside broader issues of professional capability, adoption and governance. A narrative review approach was adopted as the most appropriate methodology for synthesizing a rapidly evolving, multi-disciplinary evidence base and integrating conceptual, technical and practice-oriented literature where formal systematic aggregation is neither feasible nor aligned with the interpretive aims of this work. Results GenAI is positioned as a distributed cognitive support system that can strengthen methodological reasoning, make statistical assumptions explicit, align analyses with research questions, and improve interpretive coherence. By functioning as a second reader and conceptual alignment tool, GenAI may help reduce common analytical mismatches, improve transparency, and support more responsible statistical practice. Within MRS, its potential value lies in both supporting individual projects and in strengthening research culture. Conclusion The principal contribution of GenAI in biomedical research lies in reinforcing statistical reasoning rather than automating judgment. When deployed within appropriate governance and human oversight frameworks, GenAI can act as a methodological scaffold that helps move research from fragmented data outputs toward more coherent, interpretable and reproducible evidence. Its promise is technical, professional and epistemic. Plain Language Summary Research in medical imaging can produce a lot of data, but not all studies use strong or clear methods. This paper looked at how generative artificial intelligence, which helps people organize and explain ideas, could support research at each step of a study. This study found that these tools can help improve how studies are planned, analyzed, and explained when guided by people. This matters because stronger research methods can lead to more reliable evidence and better care for people using medical imaging services.
Currie et al. (Sat,) studied this question.