Though we’ve talked about it quite extensively, artificial intelligence (AI) still sounds like something from a science fiction novel. This idea of futuristic and lifelike computers is only encouraged by the way that a lot of what we talk about in AI are possibilities. Ideal scenarios of how much our lives could be improved with AI. However, not all of AI is futuristic sci-fi. In fact, there is a lot of AI that is already being used in healthcare. To help keep us grounded in what is currently happening, at the CHIME 2023 Conference we talked to our wonderful Healthcare IT Today Community to hear their insights on some of the uses of AI being used today. The video below is a compilation of their answers.
Lena Kannappan, Co-Founder at Healthcare Triangle – Primarily the AI for documentation automation where you take out these medical notes and fax documents coming into the hospital facilities. There is a huge need for improving operational efficiencies, reducing costs, as well as improving clinical decisions. By automating these documents, digitizing them, and classifying extracting information out of these documents. There’s a whole lot of great information about patient data, whether it’s lab records, information, medications, referrals, and all of that which doesn’t necessarily exist instantly in the EHR systems. So converting these and leveraging AI from the NLP in terms of classifying extraction, I think those are the types of use cases I see more.
Chris Sullivan, Global Strategy Lead at Zebra Technologies – I’m a big fan of AI applications in the realm of operational efficiencies. So for example, being around patient throughput and patient flow, how can hospitals better utilize the assets and investments they’ve made in their facilities? I think of it like an airplane onboarding and leaving, or a hotel room turning the rooms over. How do you get those procedures quicker, pay people in the ED faster, and get people discharged faster – AI can play a very big role in that. AI can look at a disparate and diverse set of data points and be able to have anticipatory actions that can move on the realm of simple and small, such as surgeries 10 minutes early, nudge the environmental services team to be ready when the procedure ends earlier, or complex looking at weather patterns, looking at prior year procedural and admission volumes, looking at the full moon on a Friday night, looking at all kinds of things that can anticipate longer term, more complex modeling so you can load balance resources. You don’t want to have too many resources that are under-occupied, you don’t want to have not enough where people aren’t being cared for.
Inderpal Kohli, CIO at Englewood Health – We have seen areas where they create some level of efficiencies for clinicians, some level of efficiencies for patients themselves too in terms of cell models. So some examples include a virtual agent assisting a patient with how-to information. A step further than that is a self-serve model which is tightly integrated with EHR, then we take it over to the clinician side where charting is coming into play which is a huge game changer because it’s real-time charting without a clinician needing to be in front of the screen. Then there’s a use case being utilized in many organizations and we are exploring is about generative draft response for inbox because that’s another physician burden, which we’re trying to solve. We’ve been using AI for a very long time but mostly on the infrastructure and cyber security – it’s really the generative AI that has all the excitement built around it now.
Erik Pupo, Director of Commercial IT at Guidehouse – So a lot of it’s rev cycle. We do a lot of revenue cycle automation work and what you can do with both mid-late revenue cycle process and workflow is embed LLMs. You train them, they’re very purpose-built for areas like claims denials, you can do it for prior auths and have these models that operate and work with co-pilots or with chatbots where it helps the staff, in terms of a lot of the denials administrative tasks that they do. They’ll actually go through those, the co-pilot itself, and cut times down 30-40% pretty easily once you get an implementation in place. So those LLMs are purpose-built, they’re running off existing data. Now what we see a lot as a challenge with that too is that you’ve got to have good data governance, good data quality, and a good overall data technology stack to be able to run LLMs, and often times organizations need to work on that first before they get to AI, but you do see a lot with revenue cycle. That’s an area where organizations have a good amount of data they can work with, they’ve got defined workflows, and we can come in and start to implement that right away.
Huge thanks to Lena Kannappan, Co-Founder at Healthcare Triangle, Chris Sullivan, Global Strategy Lead at Zebra Technologies, Inderpal Kohli, CIO at Englewood Health, and Erik Pupo, Director of Commercial IT at Guidehouse! And thank you to all of you for taking the time to read this article and watch this video! We could not do this without all of your support. What are some of the uses of AI already being used in healthcare that you are interested in? Let us know either in the comments down below or over on social media. We would love to hear from all of you!
