Importing treatment data in radiation oncology can require up to 17 manual steps. That single fact says a lot about why cancer patients sometimes wait weeks before treatment begins. GE HealthCare and Mass General Brigham are now working together to see whether generative AI can do something about it.
The collaboration, announced September 23, 2026, will explore building a generative AI tool designed to sit inside GE HealthCare’s Intelligent Radiation Therapy (iRT) platform. The goal is straightforward: give radiation oncology teams a way to query structured and unstructured clinical data, including notes, images, and treatment plans, from a single interface, rather than hunting across disconnected systems.
Why radiation oncology is a data problem as much as a clinical one
Nearly 60% of cancer patients receive radiation therapy at some point in their care. But the planning process behind that treatment is fragmented by design, or rather by historical accident. Multiple care teams, multiple systems, and a heavy reliance on unstructured information like free-text clinical notes mean that pulling together a coherent picture of a patient’s treatment journey takes real time and real effort.
One study of more than 500 radiation oncology cases found over 44 variations in treatment-planning workflows. That kind of variability isn’t just inefficient. It creates conditions where information gets missed and decisions get delayed. The proposed AI tool would let clinicians ask plain-language questions to surface relevant patient information quickly, with the system returning a summarized view to support more informed, individualized treatment decisions.
Built on a proven workflow foundation
This isn’t a cold start. The iRT platform itself grew out of earlier work between GE HealthCare and Massachusetts General Hospital, a founding member of the Mass General Brigham system. Software developed at MGH became a core component of iRT, and the results were significant. Analysis of more than 11,000 treatment plans at MGH showed the technology helped cut the time from patient intake to treatment start from as long as 30 days down to eight.
John Wolfgang, PhD, a clinical physicist in the Department of Radiation Oncology at Mass General Brigham and a principal investigator on the new research, described the ambition clearly: ‘We are excited to advance this work to explore how AI could further support radiation oncology care teams by making complex treatment information easier to access, understand, and act on.’
What the research will and won’t tell us
It’s worth being precise about what this announcement actually is. This is a research collaboration exploring whether a generative AI tool could work in this context. No clinical product has been announced. The findings are expected to inform future iRT capabilities across MR, CT, and theranostics workflows, but the timeline and regulatory pathway for any resulting tool remain open questions.
Still, the direction is clear, and it reflects a broader trend in health AI: moving from diagnostic support toward workflow intelligence. The hard part in radiation oncology has never been a shortage of data. It’s that the data lives in the wrong places, in the wrong formats, and takes too long to synthesize under clinical pressure.
The bigger picture for AI in cancer care delivery
Hospitals in multiple countries are already using iRT to manage radiation therapy and theranostics workflows. GE HealthCare’s Sam Kandala, General Manager of Therapy Guidance, framed the new work as a natural extension: ‘This new collaboration gives us an opportunity to build on that foundation and explore how AI could help clinicians not only work more efficiently but also make better use of the information available to them as they personalize treatment for each patient.’
If the research produces what the teams are hoping for, the clinical benefit isn’t just faster workflows. It’s clinicians spending less time searching for information and more time acting on it. For cancer patients already facing a long road, that shift in where attention goes could matter considerably.