9
Final Remarks
Shaheen Dewji, assistant professor in the Nuclear and Radiological Engineering and Medical Physics Program at the Georgia Institute of Technology, offered a few concluding remarks in the symposium’s final session. “Throughout the past couple of days,” she said, “we’ve seen how AI [artificial intelligence] is reshaping medicine and public health, and we’ve also recognized the hurdles that still remain, whether [those are] in data governance, regulatory frameworks, or ethical considerations.” Several key themes emerged from the symposium, she said, and she listed five of them using the “Wouldn’t it be cool if . . . ” phrasing introduced earlier by Heidi Hanson during her presentation “Computational Approaches to Integrating Radiation Exposure Across the Lifecourse: Risks, Insights, and Innovations” (Chapter 3).
The first theme was the use of AI in medical imaging and diagnostics. Symposium speakers described how AI is pushing the limits of medical imaging, improving resolution, reducing noise, and driving new standards for accuracy. She asked, but “wouldn’t it be cool if we could fully automate imaging workflows without sacrificing clinician oversight or patient trust?”
The second theme was the use personalized medicine and digital twins. AI-powered models are paving the way for precision medicine, she said, and allowing clinicians to tailor treatment plans to individuals in ways never before possible. “So, wouldn’t it be cool,” she continued, “if digital twins could anticipate a patient’s response to treatment before it even begins, reducing trial and error in clinical decision making?”
The third theme was the use of AI in radiation therapy and oncology. AI is changing the way clinicians approach cancer treatment through an integration of radiomics, genomics, and clinical data to optimize outcomes, but, she asked, “wouldn’t it be cool if AI could dynamically adjust radiation doses in real time based on a tumor’s immediate response?”
The fourth theme was data governance and ethical considerations. The power of AI is only as strong as the data behind it, Dewji said. Currently, she continued, issues of bias, access, and standardization stand in the way of equitable AI deployment, “but wouldn’t it be cool if we could create fully interoperable bias-free datasets that fuel AI models without compromising privacy?”
And the fifth theme was the use of AI for risk management and public safety. From exposure modeling to emergency preparedness, she said, AI is proving to be an essential tool in radiation protection, but “wouldn’t it be cool if AI could predict not just the risks we know but also the ones we haven’t yet imagined?”
Looking ahead, she continued, the power of AI lies not just in what it can do today but also in how people choose to develop, refine, and deploy it responsibly for the future. First, “to make the most of these innovations,” she said, “we need to first ensure that there is trustworthy AI, ensuring the models are transparent, explainable,
and clinically validated.” This builds on earlier discussions where symposium participants highlighted the importance of integrating data labeling into workflows, using clinical settings as a key example. This approach could ensure that researchers and clinicians maintain human-driven feedback loops in their work to enhance the relevance, precision, and generalizability of AI models over time. Second, regulatory alignment —integrating AI into medicine and public health while maintaining rigorous safety standards. And third, multidisciplinary collaboration will be crucial to the ongoing health of the field, with people from diverse disciplines and sectors looking at fostering partnerships, including new partnerships across research, industry, and policy, to create AI that is safe, effective, and equitable. This connects to a key theme mentioned in past discussions: the importance of creating user-friendly, interactive tools that could help clinicians, practitioners, and new researchers better understand data labeling and collection for use in AI.
In closing, Dewji said that this symposium has not only sparked new ideas but also challenged us to think bigger about what is possible. Going forward, it will be important to keep in mind that the future of AI is not just about incremental progress but also about pushing boundaries, reimagining what is possible, and shaping a future where AI works for everyone.
