What excites me most are the advancements in cancer care research.
Our lab is very active in AI. We have developed multiple models to predict various outcomes, not just in oncology but also in other aspects of spine care, as well as other areas of orthopedics. For instance, we have developed models that predict which of our patients will be on opiates six months after surgery. We have also developed models that predict which patients are going to be going to a nursing home after surgery, versus going directly home. Those are driven by machine learning algorithms, which is a form of AI.
I think that the impact there is that we are able to predict things in a much, much higher level than we once were able to do. I think the models are there, and the use of AI is very much present. However, I think the single greatest problem with using AI, or machine learning algorithms, is implementation into workflow. Most of us use an electronic medical record and most of these machine learning algorithms are not integrated into our workflow. So if I want to use one of the algorithms that I’ve developed, I have to go out of the computer system that I’m using, and go to a website. Again, I go to the website we have created, but there are others, and you go to that website and then you have to manually put in parameters that are important for that particular algorithm. That works, but the problem is that it does not really fit into a workflow that well and it is not the most efficient. And for busy clinicians, they often won’t do that. They’re just too busy to do it.
Cancer Care Research and AI Integration
What excites me most are the advancements in cancer care research. We understand more and more about the immune system and the immune response to cancer. I think that there will be an integration with current cancer care research, such as immunotherapy, with AI research. As we gain more and a larger understanding of AI, you will see that that is implemented into basic science research and will hopefully improve the speed at which new medicines are developed.
EMRs and AI Integration
That is why I think that there’s going to be another phase of AI: the implementation into the electronic medical record and workflow. That’s really when AI will become part of what we do all the time.
It already is part of our practice. For instance, billing. Sixty percent (60%) of the bills generated in our hospital by pathology and radiology are done by AI. And people don’t realize that it’s AI working behind the scenes. That’s the other aspect of AI that I think will be become more and more common, that we won’t even notice that it’s there. Because, like when we use our phones, there’s a lot of AI driving our phones that we don’t realize, and eventually we won’t even talk about AI or machine learning because it’ll be like talking about computers. They’re just everywhere. So, it’s not something you will really comment about in the future.