AI Streamlines Radiotherapy in Low-Resource Settings

AI Streamlines Radiotherapy in LowResource Settings 00:00 00:00 03:58 GO LIVE Facebook Twitter/X Email Embed Speed Normal Autoplay undefined Speed Normal Autoplay 0 00:00 03:58 LIVE This transcript has been edited for clarity. Hello. I'm David Kerr, professor of cancer medicine from University of Oxford. Today, I'd like to talk a little about AI and how we apply it to modern personalized medicine. This is an enormous field and one that we'll come back to, I'm sure, with our Medscapers through these brief videos and lectures across a whole spectrum of medicine. As we read in the public press just now, there's some recent examples in which AI programs, because they're incentivized in a way to meet various targets, have escaped from a cage or a test room and have begun hacking into other AI programs to increase their own ability to meet whatever targets have been set by the programmers. To think of AI as a caged, ravening beast — a velociraptor from those Hollywood movies, caged but angry and desperate to get out — is the sort of sense that we've recently got from the public press. SUGGESTED FOR YOU Here is a much more benign example — a rather nice one from the American Society of Clinical Oncology's JCO Global Oncology, a journal of which I'm rather fond, which is currently edited very well by Katherine Van Loon. This is a study that came out recently from a number of colleagues who are collaborating to improve the delivery of radiotherapy in low-resource settings. We know that radiotherapy treatment planning is incredibly resource intensive. It's characterized by multiple manual steps that can contribute to treatment delays, reduce the quality of the delivered radiotherapy fields, and so on. This group has developed a radiation planning assistant, which is a web-based platform designed to deliver automated contouring and planning approaches, tailored particularly for use in poorer countries. What they've done more recently is that they have used and validated clinically end-to-end AI-driven workflows for prostate and cervical cancers, the two most common cancers of men and women in sub-Saharan Africa. It's a fantastic piece of work. They have validated it clinically in 50 test patients, 40 prostate and 10 cervical, and have shown that using this AI-driven, web-based approach, they can get the level of contouring that they require in terms of delivering treatment to exactly the areas of interest in 98% of cases. It's a fantastic step forward in those resource-limited settings where we don't have enough people and we don't have access to the physicists or the mathematicians required to contour and deliver modern radiotherapy. This web-based approach is a major step forward. One up for AI before it escapes from the cage, takes over the world. It's a rather lovely example of it, and I would like to congratulate the authors who put this very nice work together. Have a look at the JCO Global Oncology article, and have a look at this wonderful example of AI-driven technology improving the quality and safety of delivered radiotherapy in low-income countries. Medscapers, thanks for listening. I'm interested in your own ideas on AI. It's something that we'll return to. For the time being, thanks for listening. Over and out. Thank you.