You knew it was coming, now how do we take advantage of it.
Today. And so it begins. AI assistants are going to be everywhere. My name is bill Russell. I'm a former CIO for a 16 hospital system. And Korean, or this week health instead of channels dedicated to keep the health it staff current. And engaged. We want to thank our show sponsors who were investing in developing the next generation of health leaders, short test and artists, two great companies. Check them out this week.
health.com/today. Having a child with cancer, one of the most painful and difficult situations a family can face in 2023 to celebrate five years of this week health, we are working to give back. We were partnering with Alex's lemonade stand all year long. We have a goal to raise $50,000 from our community. We're already up over 34.
Thousand. We asked you to join us at our website top banner. You'll see a logo for the lemonade. Stand, click on that to give today. We believe in the generosity of our community. And we thank you. In advance. All right. I pulled this article. From healthcare drive and it's Suki integrates generative AI assistant.
In epic EHR software. The reason I pulled this. Article is because I believe we are going to see this. Over and over and over again, not in the next five years, like in the next. Nine months. We are going to see. Generative AI be used specifically around, , listening to conversations, turning that, listening into a usable transcript and then taking that transcript and putting it into.
A format that can be utilized within healthcare. All right. So let me just give you the bullet points. One of the reasons I like healthcare. , Dive is that they give us the bullet points. Here we go. Dive brief voice artificial intelligence company, Suki. Has integrated its voice. Powered tools, Suki assistant into Epic's electronic health record software through the EHR vendors, ambient application programming interface.
The company announced Wednesday Suki incorporates generative AI to listen to. , to clinician patient encounters and auto-generate clinician notes. However clinicians maintain control by accepting, rejecting. Or editing the AI content, according to the company. Documenting a patient encounter with ambient. No creation can reduce clinician burnout.
With Suki reporting that this ambient note generation is capable of reducing documentation time per note, by as much as 72% in family medicine. All right. And so here's the, here's the thing. There's a bunch of open source software out there. Here's my, so what on this. There's a bunch of open source software out there. You can get whisper from open AI.
Which will listen and generate a really good transcript. And there's other models that are out there that are available to listen and generate a transcript. These used to be hard, used to be very difficult. You used to have to train the models on specific languages. And you'd have to train it on specific vocabularies. Like for example, a.
Primary care. Visit would have a different. Vocabulary than say an orthopedic visit. And so we spent a lot of time training it on these different vocabularies and bringing you up to speed. It turns out these large language models. , can move much quicker in this direction. And so we have the ability to generate the transcript. Once you generate that transcript, the challenge becomes.
How do you turn it into the perfect note? How do you turn it into the note that can be put into the EHR or broken down into its discrete data elements and moved around. And the answer is again. These digital assistance, these generative AI assistance. Have the capacity to do this. Now I played around with this a little bit. I took a.
Conversation from a physician. And patient encounter. And I tried to turn it into a soap note. On the, , I think it was Chatzky P yeah. I used chat sheet. GPT to do this. And it was okay. I mean, it wasn't perfect. I don't think a doctor would have been happy with it. It was a little bloated. In terms of, , what it was. I think what you're going to see is you're going to see these models get trained and be more specific and be able to, to strip out the extraneous input. The.
, just, just what matters into the note. And I think going beyond that, I think. You're going to be able to move. Different elements into different parts of the EHR and other multiple parts of this. Obviously you have the listening part. You have the, turning it into a functional unusable. , input into the EHR and then you have the EHR integration.
And the EHR integration just is that next level of automation. So instead of having to do, , you know, control C or command C to maybe on what machine you're using and copy it and then move it over into the appropriate space. If you have EHR integration, You can just click a button and it moves it into the note field or wherever it's going to move the data into.
Why am I talking about this? I'm talking about this because in the past month, I've seen this over and over again, Suki is just one of the latest examples. Of, , people really going after this space. I think the biggest challenge in this space is, , you have nuance, which has a significant amount of market share. But the challenge that nuance has is these models that come in, that don't utilize people and don't utilize scribes at all, which is what the nuance model has used for.
, quite a number of years until the, , the Dax express model that's being piloted this summer. , you had labor. And when you have labor, you have higher costs. And so the higher costs. R one of the things that is going to be looked at with more scrutiny as we move forward. If these tools prove out to be viable and scalable,
And can be put into these different specialties in areas. Then nuance is going to have a
Significant competitive challenge on their hands. Because the model of just using machines versus
where the model to use scribes actually use labor lose. Use people it's just going to be more expensive. And so you could see the cost of this. Come down. I don't know. I mean, significantly, really significantly. And so it'll be interesting to see Dax express once the pricing level gets out there, how people are thinking about it. Is it materially better than some of these other tools? Now? I don't know if Suki is a direct competitor.
To nuance and what they're doing, but some of these others are, , and not necessarily broadly applied to healthcare yet. You're seeing primary care. , practices, you're seeing it used in telehealth. And I think telehealth is one of those areas where it can really make a difference. Because the tele-health documentation is, has traditionally not been as robust as it could be.
If we're able to capture the entire conversation. In a ambient format. If we're just listening in capturing it, getting the transcript. And then turning it into a much better note. , you know, then we solve a significant portion of the challenge in terms of tele-health documentation. Then obviously you have the integration and the transport, the transportation of the data.
To the larger record that, and we're always after that loan, longitudinal patient record that we can utilize. So. I wanted to touch on this. I think you're going to see this over and over again. I wouldn't jump at the first one that you see. There's just going to be a host of these.
So how would I be approaching this? As a CIO, I'd probably be doing a pilot. If I were honest with you, I I'd look at one or two of these. I'd get five physicians, maybe primary care docs. I'd say let's, let's do a small pilot. Let's see if this works. And it would have to be somebody in a controlled environment that was, , a friendly that was willing to try some stuff out.
And the reason you're trying it out is does it represent a different cost model moving forward? , what are you sacrificing? Are you sacrificing quality? Are you sacrificing speed? , integration into the HR. What are you sacrificing?
Pilots are about learning. And so what are we trying to learn? Well, the model work. Is, does it represent a significant bend in the cost curve? , does it integrate well with the EHR? Those are probably some of the things I would want to learn and what are going to be testing for and looking for in these models. As we move forward, if I expect a significant amount of them over the next year, I expect this space to change pretty dramatically over the next year.
I want to see what's out there. I want to create a mechanism for doing multiple pilots so that we can learn quickly. About these models and what they bring to bear. So, Hey, that's all for today. If you know someone that might benefit from our channel, please forward them a note. Think of scrubbing our website. This, we call it.com or wherever you listen to podcasts.
This really helps us. Appreciate it, if you would do that. We want to thank our channel sponsors who are investing in our mission to develop the next generation of health leaders. Short test an 📍 artist's site. Check them out at this week. health.com/today. Thanks for listening. That's all for now.