Opinion
Video
Yair Lotan, MD, on potential future applications of AI in urology
Author(s):
“I think that there will probably be much more accessible tools within the next 5 years, and almost certainly within the next 10 years,” says Yair Lotan, MD.
In this video, Yair Lotan, MD, highlights potential future avenues for artificial intelligence (AI) to augment clinical practice in urologic oncology. Lotan is the chief of urologic oncology and a professor of urology at UT Southwestern Medical Center in Dallas, Texas.
Video Transcription:
I think there are a lot of potential applications. Some of it, of course, will require testing. It's hard to say there's going to be a commercial application unless there are companies willing to support this type of research and these applications. But there are a lot of areas in urology where there's still a lot of interobserver variability between pathologists, whether or not it's inside of pathology. So, cytology results, [where] there are 1000s of cells, [and] a pathologist may have a hard time looking at all of them, an AI program could quickly look through it and highlight the areas of interest. I think, looking for variant histology, which is sometimes difficult, [is another option,] if you could train a program for recognition. In China, they have a lot of facial recognition programs. You could have recognition programs for grade of cancer or for variant histology. These could be pretty low hanging fruits, I think, for training these AI programs, which really do a very good job with recognition, and probably do better than humans in most cases. In radiology studies as well, looking at subtle differences in the size of a lymph node, where you're asking a pathologist to differentiate a blood vessel from lymph node. I feel like those are probably going to be the applications that will come first.
I think, as I mentioned, early on there'll probably be ways to augment pathologists and radiologists. Eventually maybe in high volume areas or low access areas, they might be used more primarily, at least as initial screens. I do feel like the rate of adoption for this technology will be relatively quick, [or] at least quicker. It's in a lot of iPhones already and being adopted by large corporations to help with email and writing letters, documents, etc, that are [more] relatively wrote. I think that there will probably be much more accessible tools within the next 5 years, and almost certainly within the next 10 years.
This transcription has been edited for clarity.
