Timestamps:
Summary:
In this episode, Mallory interviews Jeff Jones, a seasoned technologist in the healthcare industry, who shares insights into how his organization is leveraging AI to improve patient care and outcomes. He discusses a sepsis detection model that has been impactful in saving lives, as well as AI projects aimed at increasing productivity through automated clinical note abstraction. Jeff also touches on the challenges of using AI with sensitive healthcare data while navigating regulatory environments. Additionally, he reflects on his previous experience working with associations and offers advice for healthcare associations looking to leverage data and AI to better serve their members.
Let us know what you think about the podcast! Drop your questions or comments in the Sidecar community.
This episode is brought to you by Sidecar's AI Learning Hub. The AI Learning Hub blends self-paced learning with live expert interaction. It's designed for the busy association or nonprofit professional.
Follow Sidecar on LinkedIn
Other Resources from Sidecar:
Tools mentioned:
Other Resources Mentioned:
More about Your Host:
Mallory Mejias is the Manager at Sidecar, and she's passionate about creating opportunities for association professionals to learn, grow, and better serve their members using artificial intelligence. She enjoys blending creativity and innovation to produce fresh, meaningful content for the association space. Follow Mallory on Linkedin.
Disclaimer: This transcript was generated by artificial intelligence using Descript. It may contain errors or inaccuracies.
Mallory Mejias: [00:00:00] Jeff, I want to thank you so much for joining us on the Sidecar Sync podcast today. How are you?
Jeff Jones: Good. Happy to be here, Mallory.
Mallory Mejias: Absolutely. I'm hoping you can share a little bit about your background with our listeners before we dive into our conversation today.
Jeff Jones: Sure, yeah, I am, uh, the system vice president for quality reporting and analytics for a large healthcare organization.
I've been with the company, oh, about 18 years now. Um, I started out more on the finance side and over the last 8 years I've moved over really more on the clinical side. So working more on the side with the clinical data.
Mallory Mejias: Absolutely. So your role is Systems Vice President of Quality Analytics Applications Reporting.
Is that right? I had to write it down. What exactly do you do in your typical day? Day to day work, maybe some projects that you're currently working on?
Jeff Jones: Yeah, I think the, kind of our whole department's mandate, really we focus on setting quality goals at a national level for all of our clinics and hospitals.
Um, [00:01:00] at the company, and you know, every health care organization really wants to focus on their quality, which mostly involves making sure we don't harm the patient anymore, right, coming in the door. So making sure that the quality of care we provide is evidence based so that, uh, um, When we're providing care to the patient, that it's all based on evidence based practices.
We kind of reduce standards and variation in the care that we provide them. So at a national level, the quality department that I'm part of, we set these national goals. So clinical leaders will work every year. They will, uh, get together and work on what are the national trends, like this patient The latest ones now is really mental health.
So we're really focusing a lot on a national level. How do we make sure we screen all of our patients that should be screened at a national level? So clinical leaders will look at all of these trends. They'll look at where we're not doing as well nationally on certain quality trends. And then my team will set these [00:02:00] goals.
We'll pull data from all these different EHR systems, electronic health systems, and then we'll pull them together, synthesize the data, and then we have to set a baseline. So we know where all of our clinics or all of our hospitals are performing at a baseline level and then we set targets. So based on nationally where we are, maybe we want to get to the 70th percentile, maybe we want to get to the 60th.
And then we will set targets, quality targets, for all of these hospitals and clinics to try to achieve. And my department comes into play because we set these targets, uh, we communicate them out, we get feedback. And then we take the data from all the systems, um, synthesize it, automate it. And then publish it out to a data warehouse where we have dashboards and analytics that are pushed out at a national level to really provide monitoring of the data, uh, to the executives, but even all the way down to the bedside, providing patient level data so that it allows each [00:03:00] clinic or hospital to focus on those areas, maybe those care units or those hospitals that aren't doing as well, that have the high opportunity to really give them the data that they need to say, here's where you need to focus your time. And then hopefully by the end of the year, these hospitals that were not performing well, they all hit their targets. And that's generally, I think, been the most rewarding part for me is coming from a finance background where we're just dealing with the finances of the company to now really impacting patient care.
So when we look and see that all of our hospitals start trending up in quality, we become maybe at the top quartile nationally on a certain goal that we set out, that my team and the data we provide to the executives to help influence support for the measure, to help But also down to the bedside where we're providing nurses, doctors with the information they need to help improve the quality of care.
You can actually see that being impacted through the data we're providing.
Mallory Mejias: Well, that, that sounds incredible, Jeff. Thank you [00:04:00] for sharing all of that. I think it helps set the tone for what exactly you do. And I love hearing about the impact you have kind of all the way down to the patient. It seems like everything you just mentioned is a playground ripe for artificial intelligence in terms of synthesizing data, um, sharing it, setting goals, benchmarks, what kind have AI, what kind of AI projects have you been working on recently?
Jeff Jones: Yeah, we work with another team under the clinical side. It's, it's the data science team. So when we set these goals and we publish our dashboards, we work with the data science teams. There's several data scientists. They're clinical. Some of them are doctors. where we say, what do we need to move the needle to improve performance?
And I think one really good example is, you know, sepsis mortality has, is, um, it's always been a goal of ours to improve quality on for years. And if you're not familiar with sepsis or the audience, is it, it's really just a, it's an infection of the bloodstream. And the danger of it is, is that it [00:05:00] can look like flu like symptoms.
So if you come into the ER, you may just have flu like symptoms. And if it's something that's sepsis, septic, it can be caught and resolved very quickly. However, if it goes untreated, it can become deadly. Um, the part that kills the patient is the body's response to the infection, and it's that body's escalated response that can really cause a mortality.
And so the idea with sepsis is, and our quality goals is, what are the evidence based practices to make sure we detect it soon enough and treat it soon enough? So that really becomes a non event for the patient and we can, you know, improve their care and move on. So every patient that comes in gets a risk adjusted score based on all their comorbidities, if it's been a patient of ours, all of the symptoms that they've presented when they come in, and there is something called a sepsis app that's pushed out to some of our facilities that as soon as a patient comes in the door, it analyzes all the data about the [00:06:00] patient and And if they are potentially septic, it will shoot an alert to a doctor or nurse to say, Hey, this patient might be septic.
You need to go and treat them. And so it was very high impact. And it actually took a few years to really tune this because I think initially it was giving false positives, you know, and then doctors and nurses get alert fatigue. And so they have trouble. Okay, well, I'm not going to believe it anymore. And so through tuning.
Uh, of the model over and over again, we were able to really tune it to a point where now it's actually very accurate and, you know, saving lots of lives.
Mallory Mejias: Wow. And so you trained this model in house, I'm assuming, based on all the previous patient data that you've had?
Jeff Jones: Yep. Yeah. Yeah, we trained it in house based on the data that we had.
And like I said, it took, it took several years to get it right. Uh, cause there's just so many factors that go into it.
Mallory Mejias: And when did you start working on this project? I'm curious. I'm curious.
Jeff Jones: Oh, well, that was before, um, it was probably about six years ago, I think, when we started [00:07:00] working on it. So it's been in production for a while.
It's still limited to only certain hospitals because it currently only works with a certain set of EHR systems, where it's sort of plugged in and the alerts can be fired automatically.
Mallory Mejias: Well, the conversations we know we talk about on this podcast around AI have been going on for A lot longer than it seems at least with the press that we AI has gotten in the past few years So hearing that you all have been working on this for six years is kind of surprising even to me given that In our worlds, right?
We're talking 2020 2021 is when we saw kind of this huge boom in AI. Have you found sense? Everything's blown up in the past few years in terms of artificial intelligence. You've been able to make bigger strides with the model
Jeff Jones: Oh yeah, definitely. And I think, I think sometimes it's actually kind of, uh, the technology that it's sitting on is actually fairly old.
You know, you think seven, eight years ago, that's ancient. I think in today's technology world, um, and it, it does need to be updated. And I think with the new invents of technologies coming out with generative AI, I think there's bigger [00:08:00] opportunities to certainly to improve it.
Mallory Mejias: Absolutely. So I imagine this is more on the, the predictive side, right?
Looking at historical patient data and predicting that this patient might be likely to be in sepsis, if that's the correct term,
Jeff Jones: um,
Mallory Mejias: septic, but there is not a generative AI feature in the sense that you can't kind of. Interact with the model and ask questions. Is it more of just like a yes, this is possible or no, it's not
Jeff Jones: correct Yeah, and a rating, you know, this patient might be in an 80 percent chance.
They're septic or 20 percent chance. They might be septic Yeah Yeah, so it's a almost a binary response
Mallory Mejias: This is just, it's so much to take in. I think a lot of what we speak about on this podcast is around the, the impact that AI will have in the world as a whole. But also we talk a lot about marketing use cases and generative AI, you know, repurposing your old content with AI.
I mean, this is actually impacting people's lives, which is, is pretty incredible. [00:09:00] Um, do you have any other AI projects on the horizon that you could share?
Jeff Jones: Yeah, and I, I can tell you just the patient care and how to, that's what really drives me, I think, in being in the clinical world now in healthcare, that is just, uh, what drives me and what drives the team that we're actually impacting patient care.
We're almost on a pause right now, I think, with new AI projects. I think we're trying to figure out, um, with the new generative AI models coming out, we're getting ready to move on to a different cloud platform for our data warehouse. And so we've been consumed by doing that migration first. I think some of the challenges that we're dealing with, with things like generative AI, is, you know, health care is heavily regulated on.
The problem is, is you don't want to do too much. That's a I generated or has a I inserted into it that touches regulation. I mean, the sepsis one was pretty rare. Um, so I think that's a pretty big barrier. At least for us deploying more AI type [00:10:00] work at the clinic setting, just because we have to make sure that there's, uh, the regulatory environment is not, we're not introducing more risk into our regulatory environment, if that makes sense.
Mallory Mejias: No, it definitely does. I think healthcare particularly is one of those industries where you just, you can't, 98 percent accuracy is not good enough. You know what I mean? When you're dealing with people's lives, the accuracy has to be basically at 100 percent or as close to it as possible. We talk on this podcast about the idea of having chatbots.
For associations that are trained up on all of their, you know, treasure trove of content that they've created throughout the past so many years. Um, and what's interesting is when we're talking about scientific associations or healthcare associations that are interested in implementing a chatbot like that, they need high, high accuracy.
Uh, it's just a whole different ballgame when you're talking about saving people's lives or talking to doctors, physician assistants, nurses, et cetera. Um, my big question with. The model that you just mentioned, and [00:11:00] in general, is that considering hospitals handle highly sensitive patient data, what are the major challenges you see in leveraging AI effectively?
Jeff Jones: Yeah, from a security perspective, that is something that no health care organization. You know cuts costs on You know, we go through a very very rigorous process. We call it patient health information or phi That's the acronym that's thrown around we make sure that all the data that we're using and how we're using it Um is protected so that protected health information is is key And it is somewhat of a bear.
It's somewhat of a barrier, but I think it's one that we can get around Um, you know, again, I think the bigger barriers is the concern around the regulatory environment. And even sometimes it's really the, the level of data maturity in the organization. I think healthcare tends to lag a little bit more in the maturity of using data or being data driven as it would, as people will say.
Um, I think [00:12:00] that maturity level needs to catch up, I think, to other industries. Um, but that being said, you know, we still do try to use AI. I think you had asked, you know, what other AI projects we're working on. One that I do know is more back office related, right? Because it's less risky. Uh, and one of them is around scribing.
So, uh, one thing you, you may not be aware of is, a lot of times when doctors meet with a patient, they have an electronic chart. There's things in notes, there's sometimes discrete fields that they select from, yes, no, or a pick list, but a lot of information goes into a note. And there's a lot of regulatory reporting that we have to do.
So say for example, uh, we have to report to the CDC, you know, how many catheter associated urinary tract infections a patient may have. And you can't just do a query on our EHR system to see that. You need somebody that has a clinical background. to read through the chart and to determine was this caused by [00:13:00] our hospital or Or was this something that the patient came up with?
So you have to have somebody objective, obviously independent, clinical knowledge to be able to go through these charts and then code it to say, okay, this was something that's what they call a CAUTI, a catheter associated urinary tract infection, and that's called abstraction. So there's, it's very costly.
Hospital systems have to usually employ, many times they're contracted, abstractors. And abstractors will go in through the charts and they will, uh, look through the charts and interpret them and then record it, recode it for regulatory purposes or other purposes. And so there are some, uh, companies out there that we're looking at now that actually, uh, interpret through AI, do some of that abstraction work for us.
It's not going to replace the abstractors, but it will probably increase the expectation that instead of doing, you know, five in a day, now maybe they can do 20 or 30 in a day because these [00:14:00] artificial intelligence tools are going to scan the charts and make a recommendation that we think that this is what the diagnosis is, or this was a CAUTI or not, and then the abstractors are just doing more review.
rather than from scratch going through all the charts themselves. And I think that's a huge opportunity. And now that the health care industry is really struggling with cost, keep cost under control, especially in this environment, having a I implement the back office. I think it helps. Uh, it helps the data maturity because you kind of get used to see what these things can do.
Um, and it will help reduce costs. And then eventually, I think when regulation catches up, with using artificial intelligence in a clinic setting, then there can be more AI used, uh, for direct patient care.
Mallory Mejias: That is interesting. As you were describing the abstractor process, immediately my mind went to, well, surely an AI model could do that.
I would say that we have all the tools and resources available for that. You know, give, but bar the [00:15:00] regulations that we currently face and the sensitive data that you have to be incredibly careful with. Um, but I love what you said about it being the AI plus the human. And it allows them to do more quality, efficient work as opposed to, you know, Replacing them now.
I think I've mentioned this on the podcast before but my fiance is in physician assistant school He graduates next Month, that'll be well. I'm not sure when this episode will be published, but he graduates may of 2024 So we're very excited about that and he is very interested in ai as it relates to to healthcare, but has struggled to kind of find a lot of resources around there.
Now you mentioned scribing, and I know he has to do notes as part of his curriculum, and will continue to have to do notes, uh, when he has his job, but do you see generative AI playing a role in that process? Um, in, in terms of like speaking to a patient and perhaps having a, a large language model take that information in, summarize it, is that on the horizon?
Are we there yet?
Jeff Jones: I think it's coming. I [00:16:00] think one of the, you're really hitting on something that's very important. There's a lot of frustrations, I think, with physicians and nurses, because we're trying to collect so much information on the electronic health system, they're almost spending, they're not spending enough time with the patient versus just typing and coding things into the EHR system to collect the data.
And so when we set up these goals, I had, as I had mentioned at the beginning of our conversation, there's A lot of times, what we're trying to measure isn't being captured yet, at least not in a discrete field. And so we'll have to spend time, um, saying, going working with the EHRs and saying, we're going to implement this new field in here, so this is what you have to fill out, this is the way you have to do it.
You keep adding new fields and new interfaces for the doctors or the nurses to fill out, that's more time they're spending on data input than they're spending with the patient. Uh, so there is a huge opportunity as well with what you're talking about. Uh, I think with that scribing. to really help improve physician satisfaction and also patient satisfaction because the doctor's not Worried about how do I have to put this [00:17:00] in to our electronic health system to make sure it's recorded properly?
Rather, there's some way that the input can be done through voice and properly, you know, dictate it into text and properly translate it into clinical knowledge like the abstractor would do in real time. So if we could find something or some, some technology with AI where a doctor can speak to the EHR system, it can go to text and then do the abstraction in real time.
I, that, that would be amazing. But that is a pain point where I know with doctors and, and nurses today is just the amount of time and the frustration with the electronic health systems are changing all the time because of, you know, folks like us at the corporate level trying to measure all these activities.
If we can, automate that through AI, the input process from the doctor. They're spending more time with the patient and we're actually getting richer information much faster too.
Mallory Mejias: Oh yeah. I like to call myself an AI [00:18:00] optimist on this podcast. I certainly have concerns. There are things that are really scary to me in the world of artificial intelligence, but overall I'm optimistic and I think this is the perfect example of leaning into the things that we do best as humans and allowing a doctor or a nurse to spend more time speaking with a patient, explaining things to a patient.
That just seems so important. So powerful and and it's to me at least i'm not a medical provider, of course But it seems like a much better use of time than sitting there having to type out all these things Um and make sure that they're you know coded exactly right when it seems like we have the technology that could assist with this for sure
Jeff Jones: Yep.
Yeah, I think it's definitely there
Mallory Mejias: Very exciting on the healthcare front now, I want to rewind a little bit jeff I think Ameth mentioned to me that you have some experience working with associations in the past, is that right?
Jeff Jones: Yeah, oh yeah, he and I worked together years ago, working with associations, yep.
Mallory Mejias: Yeah, can you, can you talk a little bit about that?
Jeff Jones: Yeah, I think, uh, when I first started with, [00:19:00] uh, Ameth, it was probably in the early 2000s, um, back when his company was called CGI, Component Graphics. Inc. He'll probably kill me because I don't remember the acronym. But, um, yeah, so I started working with him.
I had, my background's originally a certified public accountant. So, um, I'm an accountant by trade. That's how I started. I was doing financial audits for a public accounting firm for several years. And then, um, technology came in the late nineties with, uh, with computer systems and I just fell in love with it, got a master's in information systems.
And I think I always had an entrepreneurial spirit. And so a friend of mine connected me with Amit and his company and I was just really jazzed to come and work for Amit because he's obviously very much an entrepreneur as well. So much energy, very bright, um, and I just loved working with him for the years that we had together.
And so I worked, um, on his application. I was initially, I started out as a developer doing, um, [00:20:00] customizations to the software. And then moved up to become his vice president over consulting, working with him. And so I spent a lot of time with the associations and the clients that he worked with, really developed some great relationships with a lot of associations.
And I think the one that stood out to the most to me was ASAE. Um, they were such a great association to work with. And I, For a while there, I used to remember some of the names of the folk that I worked with there before, but hopefully they're retired by now. I'm sure they might be. But, um, it's just a fantastic group to work with.
Uh, I think the association market was just, you could tell the folks that were in that market and all those associations really cared about their members, you know, that was their focus. And so I think the software that Ameth was providing to them, and I really believed it too, was providing them extremely high value because they were able to customize or configure, there was a difference, there was configuration over customization, and keep things within the product realm, they were able to configure their products to [00:21:00] really meet their business needs, uh, almost on the fly.
Uh, so it really provided a huge, A gap that wasn't being filled by any other software product, uh, at that level. And so having been the, the vice president over the configuration and the customization of the software was just rewarding for me because I was able to show all the customers how you can use and configure the software to really meet your business needs, uh, very, very quickly.
And it didn't require a lot of developers in the back end trying to configure their software for, uh, for months before they could use it. It was something that they could do on the fly. And, you know, at the time, what Amit had done with his product was really, revolutionary. Um, I think over the years I saw how other software products were starting to follow sort of his model and do what he did in other industries, but it was just great experience.
You know, I just have to thank Amit for the experience that he gave me and the opportunities to work with some of those different associations out [00:22:00] there. And, um, I really love Washington DC. So I had plenty of opportunity to, to travel out there. That's where most of our clients were, like Washington DC and Chicago.
Mallory Mejias: It's funny that you say that. We actually have two, uh, kind of Blue Cypress Wide events coming up. One next week and one, um, Well, they're both in May, actually. And one's in Chicago, one's in D. C. I feel like I'm going to D. C., you know, every other month these days, because that is kind of association land, as I like to call it.
Jeff, well, it seems like you've kind of done a little bit of everything. CPA, developer, um, I'm curious, is there anything that you haven't done?
Jeff Jones: I haven't skied with a meath yet. Oh. I think that's what we're trying to do next year.
Mallory Mejias: Alright, that sounds like this is a big bucket list item. Yeah. Well, Fur, it sounds like you have, you know, some good deep experience actually with associations and the healthcare organization you work for now is a non profit.
Is that right?
Jeff Jones: Yeah, yeah, it's a large not for profit.
Mallory Mejias: Got it. So how would you contextualize, knowing that most [00:23:00] of our listeners of the Sidecar Sync podcast are, do work at associations or non profits in some capacity, how could you contextualize the work that you are doing in a way that might be helpful or insightful for those listeners?
Jeff Jones: Um, I think, you know, for us, it's, it's a lot of, it's about controlling costs and I think getting higher productivity, you know, with any organization. If you're a not for profit, you're really trying to control your costs. And, you know, being a CPA, I used to audit not for profits and the focus there was always, you know, making sure that the money that was coming in was truly going to the benefit of the purpose of the organization.
So whether it be an association or a healthcare organization, you want to make sure that the money that's coming in is really mostly going towards the purpose. And I remember auditing, uh, not for profits. There was always a statement that would go in the back of the financial statements. We call this statement of functional expenses.
And that would break down how much went to administrative, how much went to [00:24:00] back office and how much went to the true purpose of the organization. And so I think AI and what I'm doing today. really helps drive, lowers cost and increases productivity at the same time. So I know there's folks even on my team that are afraid of, uh, I know Microsoft's co pilots coming out with the ability to just generate dashboards on the fly.
And I'm constantly trying to ease them saying it's, I don't think it's going to eliminate your job. It's just going to mean that the expectation is you're going to be able to do more with the same people that you have. So I think productivity is going to go through the roof and some of the Some of the routine and mundane stuff you deal with is going to be gone.
So I think it will allow us to be more creative. So I think our creative sides will be able to come out because a lot of this routine stuff that we have to do will be gone. I think AI will be able to do a lot of the mundane routine stuff for us and we can focus on being more creative. How can we take the data that we're getting and really focus it [00:25:00] so that our customers, our users, can get the data they need quickly and get out.
I always try to, you Tell my team, people that use our dashboards and analytics and reports, we want to get them in and out as quickly as possible, right? So if they're in and out quickly, hopefully that means that they were able to identify the gap, the opportunity that they need to solve to improve performance on their end quickly.
And they're not spending time sifting through the visuals or the tables to figure out where the opportunities are, that if they pull up a dashboard, instantly they know where the opportunity is and they're out. And we tell that story. And part of it's an art and some of it is a science.
Mallory Mejias: Well, I love what you said there too.
I think we'll see an incredible boost in productivity. And perhaps, you know, in some industries we will see job loss as well. But I think the idea is being able to do more in the same amount of time. And for associations particularly and not for profits, it's using that productivity to make more of an impact [00:26:00] on the purpose of your work or the mission of your work.
Which, I mean, I I don't think anyone could disagree with it. Sounds pretty good. At least on paper. Yep. Do you all work with any industry associations, um, at your health care organization?
Jeff Jones: Oh, gosh. Yeah, there's quite a few. Um, I think most of them are government related though. Uh, I think the advisory board is one of them.
Um, AHRQ is probably another one. There's, there's a bunch related to quality that we work with, but I think a lot of them are probably more government agencies.
Mallory Mejias: I gotcha.
Jeff Jones: And
Mallory Mejias: for any of our listeners who might work at associations that impact, you know, physicians, physician assistants, nurses, any kind of anything that impacts the healthcare industry, what advice would you have for them as someone who works for a healthcare organization in terms of educating their members, things they need to be on the lookout for, anything like that?
Jeff Jones: You know, I think number one, obviously, is the patient care, right? And just making sure that, you know, the regulatory environment is, is [00:27:00] kept up, that they're not doing anything that's going to, um, increase their risk in a regulatory environment. But, um, I think any associations that may work in the healthcare sector, it's, a lot of it's really about data now and how can we get the data, um, either for regulatory purposes or for decision making as quickly as possible.
And I think in, in terms of associations trying to service. Uh, the healthcare sector, I think it's keeping that focus, you know, maybe helping, uh, I'd mentioned data, data maturity, maybe helping that healthcare organization move up on that data maturity and using data more towards a, an advantage and making it more a strategic asset, um, than just something that you consume and have to report on a regulatory basis because you have to do it.
Um, I think the sooner that healthcare can become more data driven. Uh, in a smart way, uh, I think that will, will certainly help improve that productivity and really control costs.
Mallory Mejias: I guess in the greater landscape, [00:28:00] are you all concerned about AI replacing, um, medical providers? Or do you feel like those jobs are fairly secure?
Jeff Jones: I think they're fairly secure, um, I think the only one I know that we've discussed that could be a risk might be, uh, imaging. You know, I think imaging, and again, I think, but it still might be, uh, somebody that's looking at imaging or doing imaging may have to do a lot more than they do today. Um, that one's mostly at risk.
I have a, my son is becoming an anesthesiologist. He's finishing up his, uh, bachelor's degree at Duke, so he'll go on to med school. And my daughter, his younger sister wants to be a nurse, so she's going to be going down to San Diego State. and going in the nursing program there. So we've talked about this with them and I don't see, um, nursing or doctors getting replaced, not anytime soon by AI.
Um, I think especially the nursing side, I think you really need that patient care. I think you need somebody that can be compassionate with the [00:29:00] patient. I think that's so important in that environment when you're dealing with somebody's health. And that not yet anyway, it can be replaced by, uh, by AI.
Okay. But I think it'll be a long time off before doctors or nurses will become replaced. Again, I think it might help their productivity. Um, I think they'll be able to handle and see more patients than they do today.
Mallory Mejias: And then hopefully we'll see better patient outcomes because of that.
Jeff Jones: Exactly. Number one.
Mallory Mejias: Within the next few years, it's kind of a two part question. What would you say are the biggest opportunities you see for AI in healthcare? Like the, maybe the top one or two. And then kind of the follow up is what do you see as being the one or two biggest challenges for AI in healthcare within the next few years?
Jeff Jones: Yeah, I think, well, I think the number one right now is to really control costs. I think I had mentioned, um, I think controlling costs, maybe improving some of the scribing, the abstraction. Uh, I don't think the direct patient care is something that, uh, we'll see in the next couple of years just [00:30:00] because of the regulatory environment.
Um, and then I think the second one to improve is, is probably going to be, uh, that's a tough one. Um, probably more on the being more data driven, I think the more insightful information we can get, because part of the problem I think with a lot of healthcare organizations and especially ours is we have so many disparate systems, um, and finding that single what we call longitudinal patient record is very, very difficult.
So if you want to look at a patient. And their whole continuum of care, you would think that that would be easy to find, but even within our organization, it's very difficult, if not impossible, to stitch that information together because of all the different systems. The clinic ambulatory side doesn't talk to the acute side.
And then even if a patient may go to a radiology clinic, they may go to one that's not part of our system. And so we don't even have the data. So I think. And it's almost less of an AI question is more. It is a consolidation is the more we can try to [00:31:00] bring all that data together so that we can see that full continuum or longitudinal record for a patient.
I think the more we can improve the care. Um, and then the challenge to that is the opposite side is privacy, right? Getting that longitudinal record. You know, there's going to be a lot of concern in privacy, especially in the environment that we're in today, of somebody being able to have access and see that full longitudinal record of a patient.
And that would be a really big challenge, I think, is privacy.
Mallory Mejias: That is just very interesting that you're talking about because I'm realizing as you're explaining this issue of having data kind of consolidated in different places. We've talked about this exact thing for associations, although very different, of course, but kind of having your data in your AMS.
And your LMS and maybe having data in your CRM and none of that data talking to each other. And so something that, uh, actually a project that Ameth and the Blue Cypress family of companies has worked on is a common data platform, or a CDP, where you can basically link [00:32:00] all the systems together, have all your data in a, you know, a read only place where you can see it all at once.
Does that concept of a common data platform, CDP, exist right now for large healthcare organizations?
Jeff Jones: No, and I know Amit shared that with me and it was just, it was just an amazing concept because I can tell you usually in any type of AI project or when we're pulling data in, you know, 80 percent of the effort is just getting the data in one place and synthesize into one place.
I mean, that's, we spend so much money and so much time just getting data from disparate systems and then, you know, synthesizing it so it makes sense because the way you record something in one source system, electronic health system, it may be the same exact thing. Like a blood pressure may be recorded completely different in a different system.
So even though you pull them together, it's not ready to go. You may have to do different types of transforms on that data from these two systems. Even once you get the data, to put it together so it's meaningful, um, for one [00:33:00] simple application. And I know what Amit has shared with me, what he's doing with his common data model, and being able to use AI to apply it, you know, that's going to reduce a huge cost in just being able to compile data for associations across various different systems.
It's just a massive cost. And I can tell you from our standpoint, um, we spend a lot of effort, a lot of time, On just massaging data pulling it together just to get it to the point where we're presenting it to get insights And generate, you know ai insights as well
Mallory Mejias: Wow, this is blowing my mind I just would have assumed that a common data platform of course existed for large healthcare organizations if we're you know Talking talking about it in the association context.
Are there any? Do you have any vendors in the space that are working on that, as far as you know, for healthcare organizations?
Jeff Jones: We actually looked at a few, um, there are some vendors out there that do it, but I think they do it for a different purpose, I mean, there's one out there that pulls data together more for billing.
Mallory Mejias: Okay.
Jeff Jones: So, they have to do it for billing because, you know, sometimes, um, [00:34:00] healthcare providers have what they call bundled payment systems where, um, the way that hospitals will get paid or providers get paid is there's a single amount of money that they'll get for a bundle of services, right? Right. But those bundle of services may be across different, it may be one healthcare system and then a different doctor group or another clinic and they're just different companies and they have to split the data up.
Well, there's, there are companies out there that pull the data from these different systems. They have it in a cloud environment. Um, and they massage it together so that then they could do the billing, uh, but it's, they spend and they're very expensive because we actually looked at several of them potentially for our, uh, internal data warehouse that we're building today and they were very expensive and it's that effort of pulling all that data together and synthesizing it into one cohesive data set that's very challenging.
Mallory Mejias: It's very challenging. It does also seem like the key though. Once we get that, and we layer in some, some AI insights on top of it, we will be a [00:35:00] force to be reckoned with, it seems.
Jeff Jones: Yeah, and unlike, uh, some of these other vendors we looked at, we looked behind the scenes, we thought maybe they were automating some of it, but they're still doing a lot of manual work, that abstraction work on the back end, whereas I know where Amit is going, he's using a lot of generative AI to automate it.
So I think that would help him keep the overall, you know, cost of the effort down to a point where, um, it's very affordable. You know, unlike I had mentioned some of these other organizations we were looking at that do this, they have a big staff of, of people behind the scenes normalizing the data, uh, more of a manual basis just because of the difficulty.
But I think the AI models that are coming out today really provide a huge opportunity to automate that work.
Mallory Mejias: Yep, and for everyone listening, if you aren't familiar, uh, the company I'm talking about within the Blue Cypress family is called Member Junction. And they are creating, or they're, it's open source actually, they're allowing you to create a common data platform or CDP for your association.
So we'll link that in the show notes. Jeff, one of the last questions we like to ask some of our guests on the [00:36:00] podcast is, you know, we've talked about how you're leveraging AI in your work within, uh, the healthcare organization, but how and if, how do you use AI in your personal life, if at all, or kind of in your daily workflows?
Jeff Jones: Oh, I've been using it quite a bit, especially Copilot lately. I use it now to, if I'm writing code for something, uh, definitely that's a first one I go to. But even now, we were, I went to visit my son at Duke for his spring break and he wanted to go to the Outer Banks to, uh, for a day because he's never been there.
And I just said, plan a day trip to the Outer Banks. And it planned everything out. It told us what restaurants we should go to based on the timing of when we would be there. Um, it's fantastic. And now we're thinking about moving to, uh, Southern California. So we're trying to find something down there that's, um, affordable.
So we're using AI to try to find locations down there, looking at the timing of the markets when we think maybe costs may go down or go up when interest rates are going to change. So I use it quite a bit. [00:37:00] I mean, it's almost become my go to now rather than a, um, just a regular search in a browser. I'm now going into co pilot using generative AI to do a lot of these things today.
Mallory Mejias: That's really cool to hear. I think there are so many use cases that are honestly still undiscovered at this point. Um, you know, I'm an avid chat GPT user myself, Amit uses chat GPT all the time, and he's really a big fan of using Otter as well or audio transcription. I don't know if that's something you played around with, but if you find you're going to it more and more as a tool for search or just kind of recording your thoughts, I will say that is a useful way to go about it.
Jeff Jones: I do like to use it. I always explore things and read, um, on my own to see how that can improve what I do at my job. So I've been spending a lot of time, um, learning about, uh, chat GPT and some of these large language models, really trying to understand at a fundamental level how they work. They are very complicated, but at least having that baseline understanding of how they work I think will help me then [00:38:00] learn how I can apply them at work. And you know, there's, it's just amazing. I think I saw one of Amit's presentations where he had said that, um, what was it? Moore's law was doubling capacity every, what, 12 to 18 months. And now the AI, I think it every, every six months. And that just blew my mind.
Mallory Mejias: Yep. We did a Fundamentals of AI episode on this podcast.
Several months ago at this point if you're listening, you can find it It's fundamentals of ai part one and part two and you know, I talk about ai every week pretty much with ameth I always like to think this is like ai school for me where I go and learn tons of new things with ameth But we also talk about ai Every day, all the time, a sidecar, but every time I do any sort of fundamentals learning, I learn something new, which I love.
I think there's, in a way, I think AI is kind of a black box, right? That's what people refer to it as often. But there's also, there's a big opportunity to learn, even just piece by piece, kind of how things work, how neural networks work. And to me it's [00:39:00] fascinating, and I think the more you know about what's under the hood, at least a little bit, I think you can actually use these models better and prompt them in a more efficient way.
Jeff Jones: I agree. I think it takes some of the scary part of it away.
Mallory Mejias: Indeed. Indeed. Well, Jeff, I want to thank you so much for joining us on the podcast today. Do you have anything else that you want to share with listeners? Is there any place that they can find you or follow you to see what you're working on?
Jeff Jones: Yeah, well, no, I really appreciate you having me on Mallory. Um, I'm definitely on LinkedIn. I'm sure you can find me there. Um, if you have any questions about anything I spoke of today, uh, feel free to reach out and talk to me there. But, you know, I always enjoy connecting with you and Ameth. Um, it's been a while since we've connected, but, um, it's always great to hear from him.
Mallory Mejias: Absolutely. Well, we're thankful to have you.
Jeff Jones: All right. Thank you.