Innovative AI Approaches to Cancer Screening and Risk Assessment

Sanjay K. Juneja, MD, hematologist & medical oncologist at Mary Bird Perkins Cancer Center, discusses how artificial intelligence is transforming early cancer screening.

Transcription:

0:10 | But where I also get really excited is this multi-modality approach, where, instead of having this kind of hard set like male, female, this is your age, having AI’s understanding. Remember, it is just prediction, when you feed AI longitudinal data like these are all the variables. These are all the metrics. This is what ended up happening to people by the 1000s, like hundreds of 1000s, and then you can kind of ascertain what characteristics may be at higher risk. When you say multimodal, that means you consider their molecular stuff, if it is precancerous, or even if it is like germline, meaning, like the stuff you were inherited with. You might think, how much does metabolic syndrome matter? Any of these things that you could conceive they can help really delineate what is the risk factor for an individual, rather than a person of a certain age, and that's super exciting.

1:08 | Again, take a multimodality approach, but then also, you have all these kinds of tests coming out that are able to check for pre cancer mutations at a very what is called high sensitivity, meaning before you could even potentially see it on an image. You may be able to tell that there are some precancerous lesions that are classic to say, cervical cancer or colon cancer, or even early cancer itself.

1:34 | Now, a lot of people say, Well, how would you know where it is and what do you do with it? But that's when you can look at, you know, things called theranostics, to where you could actually inject things that say are metabolized, that are fuel sources for rapidly growing tumors. It is kind of like a PET scan, but you are checking based on metabolic activity of something else other than glucose. So, those things are getting much more defined, and earlier as well. Another one is also with endoscopy.

2:09 | So, if you think about it, you are putting a camera down to check. For example, when you have AI technology embedded, it can consider all the things that maybe you do not have as a gastroenterologist in your local area, like if you happen to have an Indian patient or a patient of Indian descent or Jamaican descent, and they have certain turmeric-based diets, then suddenly what you are looking for and the idiosyncrasies about that very early, you know, polyp lesion, you know, may not be top of mind. But now they have technologies where you input those variables, and it can kind of pull up all of the things that are more unique or classic about that person with a turmeric- or curry-based diet, or what medications they are on, or what—maybe you know—those idiosyncrasies for them specifically, and it can kind of just flag on the screen like, hey, I know we went by this, but there is something that is XYZ that is relevant to maybe having a little higher degree of concern. That is where you are able to detect polyps at a higher rate as well, and that can go both from upper and lower scopes.

3:14 | So these are all things that do not replace, you know, physicians by any means, or some of our diagnostic tests, but certainly leverage and kind of, you know, maybe one day add a log, like, actually, just logarithmically help the utility or use case for these imaging and molecular- or blood-based screenings that we do.