This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Newsday: Prior Auth AI and Breaking Down Geographic Barriers with Colin Banas

GMT20240926-013826_Recording_avo_640x360: [00:00:00] This episode is brought to you by DrFirst. Their AI powered solutions improve the accuracy of medication history, enhancing patient safety and reducing clinician administrative burdens.

From e prescribing to medication adherence, Doctor First Solutions integrates seamlessly with your EHR, informing clinical decisions and improving care. Visit thisweekhealth. com slash doctorfirst today to transform your medication management with Doctor First's innovative tools.

Bill Russell: Today on Newsday.

Colin Banas: for every paper that was, AI is the second coming and the panacea for all that ails us, I could find an equivalent paper that was like, AI is not ready yet, or AI actually underperformed versus the clinician.

Bill Russell: My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of This Week Health. where we are dedicated to transforming healthcare, one connection at a time. Newstay discusses the breaking news in healthcare with industry experts

Now, let's jump right in.

(Main) [00:01:00] Alright, it's Newsday and today we are joined by an old friend. Haven't seen him in a while. Colin Banas, CMO for DrFirst Colin, welcome back to the show. Hey Bill, thanks for having me. I'm trying to figure out if you're wearing DrFirst Colors or University or Virginia colors.

Colin Banas: Luckily it's both.

You can, maybe you can't see it, but I am a UVA grad, but this is actually the DrFirst Orange. Wow. You've been in Orange your entire career then? Yeah I've also never lived anywhere other than the Commonwealth of Virginia, which is an interesting factoid.

Bill Russell: I'm looking at your background there , I see the football, the Virginia football, the Sports Illustrated, is that like Ralph Sampson or something?

Colin Banas: No. 2019 UVA won the title for basketball after being the first number one seed to lose to a 16 seed. The year before

Bill Russell: They did, they came back and won it. Well, we've got a lot to talk about. We've both been traveling we've done a bunch of 2 29 project meetings. You've just come back from Amdiss.

Give us the update. I mean, what did you hear at Amdiss while you were out there?

Colin Banas: Yeah, I mean, AMDISS is like CMIO, nerd [00:02:00] Bootcamp. It's all the usual suspects. It's been going on for, 25 plus years. And actually I have the privilege of giving the annual literature review. With a colleague of mine from university of Illinois Chicago, Bill Gallinger.

So I've been doing that for 15, 16 years which is pretty fun. It's, it is really well received. You can imagine. It was hard to not talk about ai. I. But what's interesting is you get the real world perspective of people actually what they've implemented, how they're governing it what their success stories and failures are.

The other really interesting thing is we had a fireside chat hosted by our friend Alistair with David Feinberg. So we got a little Oracle update. In terms of what they're planning, how their rollouts are going. We're hopeful to see another big announcement from Oracle maybe for their fall conference. because they've kind of fallen off the radar as you know.

Bill Russell: They've clearly been struggling. I mean, most of the articles I'm reading is migrations to Epic and it's felt like that for the last. [00:03:00] I don't know, five years. It'll be interesting. But the news probably this week, and I was gonna reference it here in a minute, which was essentially the, I don't even know what they're calling now.

The big beautiful bill passed, right? So a lot of stuff came through that. But one of the things I noticed, there was an article where the implementation for the va, they were asking actually for some more money for the EHR. Implementation. So it'll be interesting to see. We'll follow that. We're not gonna unpack the big beautiful bill in its entirety in this.

We are going to do it over a, it's too big. We're gonna unpack it over probably a couple episodes. We'll touch on it here. because we're gonna talk about telehealth. We're gonna talk about some of the things that the major insurance carriers are doing. We are gonna touch on ai. It's impossible not to talk about AI these days.

So we will get there. The first story, major US Health insurers say they will streamline controversial process for approving care. Six major insurers, including United Healthcare Aetna, [00:04:00] Cigna, are voluntarily committing to simplify and accelerate the prior authorization process across commercial Medicare and some Medicaid plans.

The goal to reduce delays in care and ease administrative burdens for providers with key targets set for 20 26, 20 27. Some of the key points in this 257 million Americans will be impacted across multiple payer types. Real-time electronic prior authorization process targeted 80% by 2027. Insurers will reduce the number of services requiring prior auth by 2026.

That's interesting. UHC expanding its gold card program, HHS leadership. Praise the move. Calling it historic and overdue. I would probably focus on the second, which is overdue, don't you think?

Colin Banas: What are we doing here? This is a pledge. We've actually seen similar pledges within the last five years.

I'm all for it. Don't get me wrong. Obviously, prior auth is the bane of most clinicians [00:05:00] existence. It's probably the number one thing targeted for burnout. Maybe close tie with the EHR, although I think a lot of the EHR woes are getting better. But what's interesting is a lot of this stuff was already mandated.

If you go back to CMS There's a lot of stuff in there about prior auth and FIHR APIs and a lot of that stuff being required by 2027. The other thing that's notably absent prescription prior auth is not part of that pledge again, so you're getting the medical prior auth and some of the new medications are indeed covered by medical benefit.

Dr First actually has solutions in that realm as well as you're talking about the high dollar infusions, et cetera. But why a pledge, why not a mandate where's the consequences if this doesn't actually happen? Well,

Bill Russell: a mandate would come from the government. A pledge would come from the other side.

And that's, I guess the point is like, is this just a pure move? I mean, ever since the, I guess [00:06:00] assassination would be the right word for it of the United Healthcare CEO. There's been a almost a revolt going on. I mean, some of the feedback after that was just like, Hey, there's so much angst out there that people had almost no mercy for this.

Almost feels to me like a pr move, if you will. Or maybe even a way to dodge what you just said, may a way to dodge the regulation that they saw coming at them pretty hard like somebody was gonna say, look, mandate must do that kind of stuff.

That's what it feels like to me.

Colin Banas: I don't wanna be too cynical obviously they're feeling the pressure and, when discomfort breeds innovation and results a lot of times. So hopefully this, they will remain true to their word.

The gold card thing is interesting, that's actually a process or a mechanism that has actually been shown to work. In terms of, the way the Gold Card works is I, I've proven to you, Mr. insure that I know what I'm doing. I always, do the right thing before I recommend this particular image or [00:07:00] intervention.

And that because I've proven to you that I know what I'm doing, you can stop prior authing me for these sets of scenarios. That's actually interesting, that's good. Let's expand that sort of thing.

Bill Russell: Well, let me ask you this. You we're gonna touch on AI here now, and that is AI powered prior auth.

we have this sort of gambit going where health systems are investing in systems that can automate this side of it. And then obviously the payers have more money and they've been investing in, aI On the other side, we have our AI system talking to their AI system. How realistic is it to eliminate the administrative pain points by just having AI talk to each other and follow the rules on either side, and then just do escalations when required.

Colin Banas: I, I think that's where we're headed. I think it's very realistic. DrFirst is using AI to automate portions of prior auth already. And the fear is you're getting into this [00:08:00] AI war of, they're using it for mass denials and we're just having to re-up on our end and it just sort of goes back and forth.

I'm not that cynical. I do think we'll actually get to a place where the automation. Can probably handle, the bulk of these things and that you really could use escalations for the one-offs or the very unique cases. That's the hope. AI is very good at this stuff. This is very efficient at mining, large records and pulling out key pieces of data and putting them in the right places as well as following rules.

I mean, AI is great at algorithmic logic.

Bill Russell: The main thing people want to talk about in the big beautiful bill is Medicaid. I'm gonna spare you, I'm gonna talk about other things in the bill. I'm gonna talk Medicaid with somebody else down the road.

There's a lot of angst about the Medicaid reductions. It is over 10 years. And it remains to be seen how much of that is going to impact health systems. But they are planning that. It's like the full brunt of all of that that the [00:09:00] CBO is putting out there. That's what health systems appear to be planning for.

I do wanna touch on something that I think has the potential to really change how we move forward with care. And that is Congress allows first dollar Telehealth coverage. For high deductible health plans I got this from Fierce Healthcare and a move tucked into the July 4th tax bill.

Congress made permanent a rule allowing telehealth to be covered before the deductible is met for HSA Linked hdps impacting 35 million Americans. This opens the door for employers. And digital health companies create low cost virtual care options without violating HSA rules. Here's some of the key points.

Coverage applies to plan years starting December 31st, 2024. Telehealth and remote monitoring can now be zero cost upfront. Employer flexibility expands copays can be waived. Telehealth advocates hail this as a workforce retention when permanence removes annual [00:10:00] renewal uncertainty that stalled.

Progress in telehealth. I'm trying to figure out, there's a lot of talk around what have we done to rural healthcare? Like rural healthcare's really gonna suffer and whatnot.

And also there was a move just recently, I just read an article move recently, like, Pennsylvania doctors can now practice in, I don't know, like 20 some odd states with the compacts that are going across. Are we gonna see more of this? And I'll give you one more data point.

The last data point is I just got my primary care doctor in southwest Florida. It is with Northwestern Medicine. Anyone who knows healthcare knows Northwestern Medicine is in Chicago, Illinois. But they have. Doctors down here, they have primary care doctors down here, and the move is one in which, hey, if I have something serious, I might fly to Chicago and get care

but for all other things, my primary care doctor down here is with Northwestern Medicine. Are we seeing the elimination of [00:11:00] geographic boundaries in healthcare? It used to be all healthcare is local.

Colin Banas: I'd like to think so. I'd like to think this is a thoughtful approach to, using technology as a force multiplier or being able to reach those underserved areas.

What you just referenced reminds me of Project Echo, which was many years ago, but it was using telehealth specialists. Rural hospitals could present their most complicated cases to these specialists who were, states and states away and everybody could join in to listen to the, how it was thought through and, what the ultimate outcome would be or what the recommendations would be.

And you were echoing specialty care throughout the country. Obviously, 10, 15 years later we've come a long way. The cynicism in me is that we saw this huge explosion in telehealth during the pandemic, and then it really shrunk.

Right. And it sort of leveled off at, I don't know, I'm gonna say 10% for Yeah. because you couldn't

Bill Russell: get reimbursed for it.

Colin Banas: Yeah. So, and they keep kicking that can down the curb again and I know it's coming up for [00:12:00] renewal very soon. And there are advocate groups that are looking to make it permanent in terms of how much.

Is it gonna be the same 9, 9, 2, 3, 3? Or is it gonna be, some sort of fraction of that. The other thing that you also touched on is this idea of reciprocity between states. It that is a probably even bigger problem worth tackling than what this one has stepped forward because the states are holding onto this stuff so tightly.

I'm actually interesting, in how Pennsylvania pulled this off to be able to get reciprocity with so many states. That is the problem worth tackling.

Bill Russell: As an employer and I have 15 people on my staff, if employers can now sub fully subsidized let's just say fully subsidized.

Telehealth before deductibles? I don't know. I'm gonna start looking at, are there companies out there that I could potentially partner with? One Medical Teladoc transparent, others like that to build out this capability for all my employees. The other thing is 15 employees, 10 [00:13:00] states.

I mean, it's hard to find a plan that covers them really well. And a lot of my staff is young. They're all for telehealth. I mean obviously there's a whole bunch of care that is still local. That's why I have a primary care doctor that's local. That's why everyone needs to have a primary care doctor that's local for those middle of the night and those kinds of things, I think this is a great benefit. I read the Providence, CEO, go to town on this and I was like, when you have multiple, it was just like, wah wah wah wah mean, when you have multiple billion dollars in investments, you make more money in investments than you do providing care.

I have little sympathy for you in terms of, oh my gosh, we're gonna lose some Medicaid money. I do have a lot of sympathy for just outside of our major urban areas. All those hospitals are just absolutely struggling and, to pull any money away from them is gonna exacerbate an already difficult challenge.

Colin Banas: Yeah, a critical access hospitals. And we serve a lot of those [00:14:00] hospitals at Doctor First. It's something to watch. Their margins are razor thin already.

Bill Russell: . I'm gonna dig into that one a little bit more. So, last story. Ochsner Health provides the AI support physicians are looking for title it's a MA Newswire July, 2025.

So, Ochsner Health. Is positioning itself as a national leader in clinically driven AI adoption with a strong governance model, training programs and measurable results like reduced inbox burden and pajama time. As AMA survey, data shows most doctors feel less in influential in tech decisions.

Ochsner is going the other way, putting doctors in charge of AI's. Why, how, and when. 700,000 messages avoided via EHR inbox AI plus ambient listening tools, 15 to 20% reduction. In after hours EHR, time for physicians, pajama time. Every AI deployment has a clinician champion AMA survey.

Only 70% of docs feel they [00:15:00] have most influence over tech choices. And finally, Ochsner requires training before usage and physician accountability for outcomes. I mean, you were just out at Amdiss Ochsner's a client as well for DrFirst, this is probably a proud moment for you, I would

Colin Banas: think.

Very proud moment. They've been using some of our AI solutions for three plus years now as it relates to medication reconciliation and prescription renewals. This is Informatics 1 0 1. They just know how to do it at Ochsner, they've got clinical involvement, clinical.

Engagement and clinical leadership for these AI tools. I think I actually listened to one of the news days a couple days ago driving home from July 4th, and how to govern AI was how we needed to govern, clinical decision support, which was how we needed, I know we,

Bill Russell: we should have learned this lesson, right?

Colin Banas: Yeah. EHR implementation, et cetera, et cetera, I think you even referenced blockchain, which really never happened. But yeah they do it right? And they should be proud of this. What I think is interesting and I told you, [00:16:00] I just came off the heels.

Of doing a literature review where pretty much every paper I read last year and then presented back to the group, had AI in it. And what's interesting is for every paper that was, AI is the second coming and the panacea for all that ails us, I could find an equivalent paper that was like, AI is not ready yet, or AI actually underperformed versus the clinician.

What is interesting is how. These institutions are finding novel ways to apply AI and at the pace that they're doing it. So I'm guessing when they say, clinical messages or inbox messages avoided, they're using AI to generate, or at least partially generate the replies back to the patient.

I did a paper where they were using AI to triage. The patient incoming requests to make sure that they weren't emergent and needed to be handled right away. Ambient listening, it's become table stakes, right? And it's only getting better and better. And what I'm really the most excited for is

aI being able to [00:17:00] summarize. So as a hospitalist, I used to get tons of patient transfers and they were coming to our hospital because we were the big dog in town. We had all the specialists, and they would come with reams and reams of paper or, hopefully nowadays it would be reams and reams of CCDs.

And you would spend hours trying to, decipher all of this. What was their last cardiac catheterization? Did they have an echo? What was their last, a NA panel for, lupus or whatever. And the amount of time you would spend doing that and maybe still not getting the answer was astonishing.

And then now you can just feed this in and get your answers or even interact with it. One more digression, if you'll indulge me, is patients are doing it too. So one of the papers I covered at Amdiss was how patients who are very frustrated, 10 years of not getting an answer would take the corpus of their medical data, dump it in the chat GPT, which we can talk about, probably not the best idea, but if you're desperate and get the answer.

And it was an [00:18:00] incredible at finding these nuances that are so hard to string together in years and years of EHR data. So I just think it's a really exciting time. I'm glad to see that Ochsner is not only doing it right, but willing to share the way they do it and how they get the engagement.

Bill Russell: I think people are missing the point. I hear all these things of, it's not up to the doctor and it's not up to the doctor's level. And I hear this it's not as smart as a human. It doesn't have to be. It's really good at these tasks of give it something and summarize, give it something and organize, give it something.

And, invalidate against its body of knowledge. I realize, it's, oh, I was talking to somebody, I wish I could remember who, and we were talking about it. I'm like, the thing is, it's my knowledge versus ai, and I'm not silly enough to think I know as much as AI because it's the collective knowledge of the internet.

Now I understand there's bias associated with that. There's bad information and they had to weed all that stuff out, and they're still weeding it out and [00:19:00] whatnot. But it's the collective knowledge of like all the medical journals. It's a collective knowledge of all the, I mean. It's, and it's good at that.

It just has that bank of knowledge now it needs to know when to use the right knowledge at the right time. And that's what it's not a human, I don't expect it to be a human. I expect it to make mistakes from time to time. And so I will have I've loops in my automations, I've loops and I'm sure Dr.

First does and whatnot, where I have antagonist ai, I have the first AI deliver that, and I have antagonist AI look at it and say. Are you sure this is right? I mean, because I'm looking at this and so I have AI checking AI and it's interesting how much higher of a in, of a rating it gets a confidence rating it gets when we get to the end of that process, because it's gone through multiple loops of AI checking ai.

again, not perfect. And when I did talk to Mike Peffer about stanford came out with chat. EHR. Yep. And I'm like, man, that was as obvious as can be. And his first comment is, yeah, [00:20:00] but a lot harder than you think it was. Yeah. Because we're talking about clinical information, we're talking about summarizing that clinical information, he goes, that took over a year of people really thinking through all the ramifications of that and putting the guardrails on it.

It's not just as simple as plug your model du jour into the EHR and go, Hey, we can interrogate the model. And that's my concern. And I think that's Judy's concern and everybody's concern with, just making clinical information available to health systems and saying, Hey, plug in your model.

It needs the right architecture, it needs the right validation, transparency, it needs all those things in order to be done right. And there's. A fair chance that somebody's gonna do it wrong.

Colin Banas: They need, proper prompt engineering and iterative prompt engineering so that, of course that's the skill that we probably need to be teaching at the ground level and persona. You need to give the EHRA persona because when you do, one of the papers was act as a [00:21:00] leading neurosurgery neurosurgical oncologist, and analyze this chart and.

The answers that it would give you were much more accurate because you were defining the persona that it needed to behave as. And I guess that's an offshoot of prompt engineering and iteration, but I found it fascinating. We're putting LLM into our solutions. We actually just announced with Elsevier that you can use clinical key ai.

Directly within our mobile prescription solution I prescribe. So literally when I'm, trying to prescribe you, for Lyme disease, at 10 o'clock at night, because you're on a camping trip, I can ask the LLM, Hey, wait a minute, what's the appropriate dosage for this particular adult weighing this much?

Things like that. That's the way to use it is in concert with me. Yeah.

Bill Russell: I'm looking at all these different models because we're doing some different things over here. Like we're putting every podcast transcript into essentially we're vectorizing it, putting it into a database that it could be referred to and you can interrogate it and ask questions.

So we will have eight. Not [00:22:00] almost nine years of things. If they want to know what Colin Banas said six years ago versus what he said this week, they can find that stuff out. One of my favorite prompts to put in there now is do not rely on your training, only rely on the information that is available to you via this thing.

So it uses its logic. But it only uses the data I give it. So it doesn't like, make up what Colin Banas said. Like, I don't want you relying on your training, because I don't know what you're trained on, but I love your logic. And so it's really, it's been interesting to play around with these things and I think we're finding more and more ways that it can be utilized.

It is not a human. I see. How many articles did you see? Where it beat doctors or it didn't beat doctors. That's what I'm

Colin Banas: saying. For every article where it kicked butt. You found another one where it didn't it didn't

Bill Russell: I think it misses the point. Like, I, we're not trying to replace doctors.

It's the age old thing. We're trying to get doctors to practice at the top of their license. So how much work are doctors still doing that is below their [00:23:00] license? A lot, fair amount. Let's elevate that whole thing.

I mean, I think that's the point. I think there's a lot of work to do to get there, but Colin I miss these conversations. We're gonna have to do this more often.

Colin Banas: Yeah. Well, I think I see you soon at a 229, so, looking forward to that. Oh, absolutely. All right. Take care. Thanks Bill.

Bill Russell: Thanks for listening to Newstay. There's a lot happening in our industry and while Newstay covers interesting stuff, another way to stay informed is by subscribing to our daily insights email, which delivers Expertly curated health IT news straight to your inbox. Sign up at thisweekealth. com slash news.

Thanks for listening. That's all for now