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

 Today, on this week in health it, this prescriptive analytics is gonna kill us. Those that prescribe do not wanna be prescribed too. Give us information that helps us make a better decision. That's all we're asking for. Don't say you've gotta do this, you gotta do that. Just, Hey, here's what's been going on with that patient.

Here's what's going on now. Here's what's likely to occur.

Thanks for joining us on this week in Health IT Influence. My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of this week in Health. IT a channel dedicated to keeping health IT staff current. I. And engaged. Special thanks to our influence show sponsors Sirius Healthcare and Health lyrics for choosing to invest in our mission to develop the next generation of health IT leaders.

If you wanna be a part of our mission, you can become a show sponsor as well. The first step. It's to send an email to partner at this week in health it.com. I wanna take a quick minute to remind everyone of our social media presence. We have a lot of stuff going on. You can follow me personally, bill j Russell, on LinkedIn.

I engage almost every day in a conversation with the community around some health IT topic. You can also follow the show at this week in health IT on LinkedIn. You can follow us on Twitter, bill Russell, HIT. You can follow the show. . This week in HIT on Twitter as well. Each one of those channels has different content that's coming out through it.

We don't do the same thing across all of our channels. We don't blanket posts. We're actually pretty active in trying to really take a conversation in a direction that's appropriate for those specific channels. We really want to engage with you guys through this. We are trying to build a more broad community, so invite your friends to follow us as well.

We want to make this a dynamic. Conversation between us so that we can move and advance healthcare forward. Today we are joined by Charles Boise, the CTO for Clear Sense, and actually one of my favorite CTOs, and I love having these conversations. Charles, welcome back to the show. I. Hey. Hey Bill. Good to be here.

Excellent. Actually, you know, sorry I didn't dress up for you, but we've rescheduled this three times this week, and you happen to get me right after my workouts. I'm not advertising box, they're not a sponsor, but somewhere along the line, I got a box shirt. So, hey, you caught me the only day this week with a collar, so what the heck?

So you're, you're dressed up. We're not playing the right parts here. I'm supposed to play I And you're supposed to play the CTO, aren't you? That's, yeah. No, that's okay. This is. CTO on dress up day. And this is CIO on dress down day, on dress down day. The last time you were on the show, we did a back and forth with you as the CTO of a health system and me as the CIO.

And since you're the best CTO I've run across in healthcare, and it's been a while since we talked, I'd, I'd like to continue the conversation if that's all right with you. No, that sounds great. And as usual, I have no idea what you're gonna ask me, which is typically the case in A-C-I-O-C-T-O. I'm gonna call it a confrontation.

And or collaboration. Come on, Charles, this. Oh, that's right. Healthcare, we're collaborating. Absolutely. Absolutely right. It is, and I have not given you any indication, and what I'm pulling from is just conversations I've had with clients, conversations with other guests, and. Just things that are going on in the industry, things I'm reading about and whatnot.

By the way, the last episode that you and I did as C-I-O-C-T-O was really well received. It was one of the most listened to shows of that timeframe, so I guess people appreciate this sort of back and forth and I, I'm looking forward to doing it. Let's start with call centers. Okay. You have a lot of background with call centers and in healthcare we have a lot of them, and I think that the need for 'em is only growing.

We need call centers to support many parts of the conversation with the patient. Our remote patient monitoring group would like another call center and . Let's approach this, I, I think in two directions, but I'm gonna start here. Our call centers that we have in healthcare today are not optimal. Give us what is possible, what are the best practices and what's possible based on what you've seen out there versus maybe what you're seeing in healthcare today.

Yeah, you're, you're right in the fact that. Optimal, I would call it. We've got all the components in place, we've got the best intentions in place and, and as usual, it's best for us to look out to see what others are doing. So there are all sorts of BPOs out there. And what are some of the technologies that they've brought in?

That potentially could be beneficial for healthcare. So let's talk about a few of 'em. Ones that I've been, you know, working with that I think are, you know, exceptional, that can really, really help us. And kind of Bill, what, what always happens when we go down this route is we pick and choose a vendor for this channel, a vendor for this channel, a vendor for this channel.

And channel can be a voice, it can be text, it can be some type of automated response, and. Trying to find a vendor that satisfies all of is difficult. What I usually start with is, what are we really trying to accomplish, not just now, but what do we need to do in six months, 18 months, a few years out with this, and I wouldn't call it a call center per se, but really it is part of that.

They're using that term, a glass door, a healthcare glass door and whatnot. And this is absolutely a component. Part of that, believe it or not, there are technologies out there that. In real time record and actually do not just NLP, but NLU and they can actually bring information back to the agent in, in real time.

So if somebody's trying to, you know, schedule and the agent's got their location's, got their provider, they don't necessarily have to go to a scheduling system. This can actually be done from an automated perspective. So they can actually. In their ear, hear them with the best slot, and then as part of the conversation, those slots actually get, get filled and you can put a, um, some intelligence behind it to optimally fill.

We're trying to fill the closest appointments and so forth. And then from a quality perspective. We can do an analytics on that whole interaction with the patient and give a score back to the, uh, the agent and whatnot. That's just, you know, one real simple example you've given of all. NLP, natural Language Processing, NLU.

Natural Language Understanding. Correct. Wow. So it's actually listening in real time and picking up on cues. And based on those cues, it's able to trigger workflow or process of some kind in the background? That's correct. And that's just kind of. Next ladder of maturity, if you will. Some of the work that I'm doing with digital humans is, is very fascinating.

Where the, um, patient has a complete interaction with a, a digital human. That can be either in the form of a bot or it can actually be in the form of a conversational agent, where you can actually walk 'em through, uh, a number of things. Triage, you can walk 'em through scheduling, you can walk.

Informational components of your organization, you can even fill prescriptions. And that can be done with a voice agent and or the proper bot. And I'm not talking about bots or conversational agents where you pre, you know, canned responses. I'm talking about technology based on graph where. The whole organization has been scraped.

Their whole informational assets have been scraped, put in a graph. So when that question comes across that comes, that response comes from that graph. So there's a lot of really interesting technologies that really support this activity and it's really satisfying for the the client. And what's really interesting is.

Especially with these conversational agents, nine times outta 10, they don't even know they're speaking to a artificial human. Now, we're not dispensing clinical information this way at this point. We're just doing administrative stuff, right? Yes. But if you think about it as part of that conversation, if it gets to the point where somebody needs to either dial 9 1 1 or talk to somebody, that call can then be transferred to the proper

A triage nurse and so forth, but from initial intake and directional, and I'm not talking IVR type stuff, press one and all that stuff. I'm talking about a real conversation back and forth. And an understanding of why that person's engaged that call center to properly the call. And it could be a direction to another conversational agent, or it could be a direct to a human being.

Alright, so you're hitting on a couple different areas. One is when we're looking at a call center, we're looking at being the most efficient organization we can be. That's one aspect of it. And the other aspect is providing the best experience that we possibly can. And then the third aspect is being as effective as we can.

With the resources on the desk, we have this NLU. As the machines are listening to the conversations, there's something they can really do. In all those things you mentioned that they can trigger workflows and identify the, the most optimum appointment that's happening. That makes our system efficient.

When we think about the patient experience and the effectiveness of our call center staff, you can also put a score on the conversation, right? I mean, there's that kind of, 'cause you and I have talked about this before, that it listens and it listens for cues and those kind of things, and it can actually grade.

The conversation and say, look, you responded in this way and it would be better to have responded in this way, or those kind of things so you can actually grade. How are we doing for the consumer, for the patient, and how are we in providing education back to the staff, the support staff that call centered staff to help them get better as they go forward with the next call.

And it builds. This measures tonality too. So if you're not showing empathy, those can be absolutely measured in feedback, be brought back to the, the particular agent and whatnot. An actual scorecard. You're absolutely right. Wow. So call centers, remote, patient monitor. We're seeing them expand and there's a need for this.

You talked about the approach that healthcare typically takes. We go out and we find a vendor. We pop some technologies in there, and then we try to weave 'em all together. And whatnot. If we were greenfielding this thing, if, if we were starting from scratch, greenfield starting from scratch, and somebody said, let's put in a new call center, how would you approach it?

Sure. I would approach it in a couple of respects. One, I. I'd start with the concept of infrastructure as a certain infrastructure that's built not specifically for this purpose, but has this purpose right now in mind as well as what, um, we're gonna be doing in the next 18 months. You know, two years out.

And people say, what are you talking about? And it really is about choosing the, the right infrastructure and the right place for this, whether it be a cloud, multi-cloud on premise, whatever that might be, but really understanding that. Our technologies are going to be exponentially, um, changed over the course of time.

And make sure you do not put yourself in a box and make sure that these contracts that you're making with these various vendors and whatnot have out clauses. 'cause you know how awful it would it be to put yourself into a five-year box and in two years your vendor didn't quite keep up with it and your competition is killing you and you're stuck with this thing.

So. Your assets. That's true of every cloud contract or as a service type contract. But it's harder. You and I talked cloud the last time you were on Yeah, it's harder. It's harder said than done to just, okay, so the contract didn't work out. I'm gonna move from this cloud provider to this cloud provider.

Those are serious projects to move from one to the other. Yeah, and I'm not talking about from a cloud provider's perspective, but where you decide to put this. 'cause I'm gonna tell you, it really doesn't matter. A lot of people will argue with me on that, but more importantly, what you're bringing into that environment, make sure that you can get out from under it if you need to get out from under it.

So if NLU exponentially improves. You're stuck with, you know, somebody that's not, you know, paying attention. You need to be able to modularize that NLU component, bring in what you need now, and then go forward with it. You gotta not think of the technology from end to end. You gotta understand the component parts of that technology and how they play the role so that if there's a, a change or a need to change it out, you're able to change it out.

One of the things we talked about in the past was the use of analytics in call centers and how much more sophisticated they're getting and how much more real time they're getting, and how much more effective that can make an organization. Talk about the maturity of analytics on the backend and how that can really set the table for a much more effective and efficient way of working the queue.

Sure. So if you think about, if you have this tech in place and you actually have the data producing the, the proper real time dashboards, if you will, not only can you solve the quality issues that we talked about earlier, but you know, why do you arbitrarily start at 7:00 AM. You end at five, you understand the volumes, you understand the most commonly asked questions, or the most commonly, frequently tasks that these folks are doing and whatnot, and you really have a very, very good handle on a day-to-day and how you can switch, switch things even more important.

It helps you get a handle on from a scheduling perspective, providers, how they're scheduled. Are there hours of operations in line with the folks that are requesting them? So you get a lot of really good operational stuff. I'm gonna term operational research. It actually is a thing, and with the right individual or team paying attention to the data that's produced in these call centers, you can actually optimize them a lot quicker than you thought you might just by doing that research and so forth, especially on.

If you think about future placement of clinics or even moving staff around and whatnot, you get a pretty good idea of the population, their needs and so forth. And when you try new things out, you get the feedback, is it working or is it not working? I'm gonna have the agents say this. In this situation, yes, it was effective.

No it wasn't, and so forth. Bill, one thing I didn't kind of talk about earlier that I wanna bring up that kind of hit me is we can actually by voice authenticate who's on the other side. It's really quick to walk you through your initial authentication, and then from that point forward, as soon as your voice is heard.

We know who you are and we can, you know, produce to the agent a little bit of information about you to get them, you know, acclimated to you as you go forward with a conversation. Alright, Charles, I'm, I'm just gonna move on. So development, we talked build versus buy. The last time you were on the show we're, we're finally going to have to build some applications or at least heavily modify some tools that we get from various partners.

In healthcare, and it is been a while since we've done app development, and as this is starting to emerge, I think CIOs and organizations are, again, they're exercising a new muscle. We used to all do development back in the day and then we all stopped doing development and said, now we're gonna, we're gonna go more in the build route.

But now we're standing it up again. Where do we start? Standing this up, we don't even know where to start. So is it processes, procedures, is it finding the right partners? I mean, where do we start with regard to standing up an effective development process? Sure. And we're talking about development from the aspect of, Hey, I'm a healthcare organization.

I wanna build out my own products. Correct? Yeah. Let's just, we'll make it easy 'cause we'll just pinpoint it. Sure. It's, uh, digital front door. Okay, so favorite buzzword of of the day. So we're gonna take our epic assets and we're gonna start building our own tool, and it's gonna be a mobile tool as well as a web tool.

And we were a little unhappy with the fact that we couldn't customize it during CVID, and we needed to customize some of the things, and it was just not as effective as we wanted to. Coming out of . The pandemic. We're saying, Hey, we wanna be more effective. We wanna be able to customize the tool, maybe drop some new tools in there, drop a bot in there, drop some other things in there to be more responsive to our community through our digital front door.

So that's, that's the background. We're standing up this capability. Where do we go? Okay, bill, let's do it one, let's practice number one or rule number one. Let's not make this a religious effort. Okay? Let's make this extremely practical. Let's take a minimalist approach. Let's only put in place what we need to get this done, but also to build off of, because we're gonna get done with this.

We're gonna do enhancements, and we're gonna do such a wonderful job that we're gonna have a lot of other products to build. I'm one for adopting a product development, you know, methodology, if you will, or infrastructure. So . Let's go out and find ourselves a, a really good product, somebody that really understands what we're doing, and if we use an agile process or whatever the processes we may use, let's only take out what we need to make this thing work and that let's not make it a religious effort where we're working the agile process as opposed to building out what we need to build out and let's spend.

Uh, considerable amount of time in the planning phase. Are we gonna do this on-Prem? Are we gonna do this in Azure, AWS, Google? And let's look a little bit more broadly. Do we wanna be a multi-cloud? We've got a hundred hospitals Bill. What's wrong with a multi-cloud approach where we combine the best of the best in any environment as well as our on-prem environment, depending on what the workload eventually be.

Let's not be really narrow focused on this, and let's spend a lot of time on the planning side of it as we're planning and we understand the different, you know, components. Let's sure we make, let's make sure that we have the proper . Architect on board. Yes, we can absolutely do this from a consulting PERS initially, but let's use that time to find somebody and Bill, if we're not familiar with hiring these types of people, let's find some folks that can help us hire the right people.

Because in healthcare we always get ourselves into trouble. We think we know we're the best at hiring. We find out that we've hired a team. That's not quite what we expected, and then we've lost six months and whatnot. So let's talk about where we can get in trouble. Where do these development projects go off the rails?

They go off the rails initially when whatever we've conceived isn't in alignment with the board. It's not in alignment with our mission, vision, values, and goals, and it's not in alignment with our C-suite. It's when we've conjured up some kind of idea that we think is gonna be the blockbuster idea, and then we start going down that route without bringing in the proper stakeholders and whatnot.

So I think that. There's a couple things that we do sometimes, right? We come up with this stuff on Oursel on our own, as well as we're asked to build this out or be participative. So it's really important that we bring everybody on. That's the first thing that sometimes gets us in trouble. And second is not bringing on the the right team or that kind of thinking, Hey, you know what?

I think I got the right resources here. Well, I mean, you're gonna have to ask yourself, do you really. That's a lot of where we get unhinged. I. Agile has something that I really like, and that's an iterative approach to these things where we come up with an MVP that actually works and has some function and we build on it from there.

And that allows us to pivot because although we did the best planning, I. We did the best architecture and we've done the best reach out to the stakeholders and we've done the best mockups and whatnot. Things change as we're going along and how can we pivot and make these applications pivot as we go forward?

Yeah, those are the ones that I kind of think get us in trouble and what we don't get feedback. Look at the EMRs. They're still look at the way, you know. They look like the same thing that I used when I was using a paper chart. That hasn't changed. It was a replication of the chart. What are we trying to do?

What are we trying to accomplish? How do we fit into the workflow? I agree with all those things. I would say that there's a handful of other things. One is. We don't begin with the end in mind, so we sit there and go, all right, we're going to develop this app, and we've included the stakeholders and we're getting input from the clinicians.

We're getting input from the patients, and we're building out what they really want. But we get into it and we don't recognize, Hey, we're gonna roll out a mobile app and a web app. That's all well and good, but what's gonna be our update procedure? What's gonna be our. Keeping it current procedure. 'cause every time Apple comes out with a new iOS version, there's a whole bunch of stuff that gets deprecated and you have to come out with a new version.

So you're gonna be recoding. You're never gonna stop coding this application or it's gonna die. That plus Android is a lot harder if you think that's arbitrary and hard, that every time iOS comes out, they deprecate something and, and you have to rewrite. Android has like six different flavors you have to write to.

So if you're gonna write a mobile app, it's not like, Hey, I wrote a mobile app, and away we go. And you could write a. A responsive app and just put that on your iOS device or your Android device. But at the end of the day, responsive apps have some limitations that you're going to wanna consider as well.

So we need to start with the end in mind. You're gonna constantly be updating this app. You're gonna be adding new things, so architecture's gonna matter. So you can remain agile. You can continue to add things to this all along the way. The other mistake I see people making, I see CIOs making is I'm gonna see that code in a month.

You should see code all the time. New code should be coming across your desk weekly that you're looking at. And it could be a simple function like, let me see the two factor authentication, and Oh, that's how it works. Yeah, that workflow will work fine. They could do that in a box and it's modular, and then they could do the initial screen and they could do whatever.

You should be seeing code all the time. I think one of the places we get in trouble, I. If you haven't done development projects, agile is neat and nifty, but if you're doing Agile, you should be seeing code early and often all along the way. Everyone who has done Agile well will tell you you're looking at code all the time and you're looking at progress all the time.

Any vendor who's sitting there saying, Hey, we'll have something to you and you'll see something next month. And if you're not actually working with real live code in a couple of months. Then something's broken. You shouldn't be looking at PowerPoints two months in. So those are some of the areas that I found that we just get in trouble.

And Bill, let's not forget security and bring in security in early on to ensure that there's no vulnerabilities built in and folks that are using low code, no code environments. Make sure that you bring in security early on so that . You don't, no code low, code yourself into a vulnerable application. All right.

So you're gonna have to define low code, no code. What does that look like? Yeah, so basically you or I, because we haven't done any development in a long, long time, basically we have a menu and we kind of drag and drop different components of functionality in and app. And at the end of the day, we have a functional app, but we just gotta make sure that we can do the SonarCube tests and make sure that there's no vulnerabilities in it and we're not.

Putting a field in there that somebody can drop malicious code in and really, um, causes a problem later on. So, Charles, one of the things I've heard is people saying, I'm not gonna build out my own team. I'm gonna partner with somebody. So they find a local partner who. Has some shops of some kind. They've done some things in the industry.

Can I take advantage of a partner or how do I manage that outsource development partner once I do identify the right one? Yeah. If I was gonna do outsourcing and I do, if it's totally offshore, that's not one for the timid, if you will. It takes a little bit of time to get the experience. Alright, so let's stay onshore, but I, I will come back to the offshore.

Let's, so I hired somebody in Baltimore to do this. Hey you, you got peers. You ask around. Really ask around, uh, because right now it doesn't matter whether they're in Baltimore or California or you know, Oregon for crying out loud. Just make sure that they have, you know, some background in healthcare.

They've actually worked with healthcare organizations and they got a portfolio. They got something they can show you. They got some wins and whatnot. It's probably not uncommon to say I hired somebody from Baltimore, but they're actually using overseas resources. That could very well be true. Absolutely.

So you, you've gotta be careful on that. Oversee resources. What's the upside? Downside? Sure. Upside. Because that I've been doing this for, you know, 15 plus years, uh, and I picked the, the best of the best outta the best university. So from a skill perspective, phenomenal. From a cost perspective, there's a savings there, but that's not necessarily, you know, my thing, if you're not familiar with the culture, the time difference, the business practices, it can be a little bit tough.

I would definitely recommend going through a local agency for those types of. Resources unless you've done it before. All right, let me switch the gear. I, I always wanna talk to you about data 'cause you've architected a great solution over a clear sense. And anytime I get a chance to talk to you about data, I'd like to do that information blocking rule is coming down the path a little quicker than the last time we spoke and we could be facing penalties if we don't share the data through APIs with personal health apps.

In addition, the, the proposed HIPAA rules are now calling for moving the. Sharing of information with the patient from 30 days down to 15. It's 15 business days. So what is that? That's 20 some odd days. I guess as ACIO, if we're just still roleplaying, I would look at you and say, how are we gonna do this?

We're gonna share information with personal health apps that we cannot really verify. Much, and we're being asked to move a little quicker than we have in the past with getting this information into the hands of the patients from 30 business days down to 15 business days. How, how should I be thinking about this?

How are we gonna approach this? How are we going to comply with this rule? It's good. And Bill, you're absolutely right. It absolutely getting around. We probably can't hang out and wait for our EMR vendor partners to set the stage, although that's kind of what we're doing. Right. Would you say that's a fair statement,

Unfortunately, yes. But if I'm of a certain size health system, I almost have to rely pretty heavily on my EMR partner. Yeah, and you're gonna have to totally, totally understand that interim. From a education perspective, you know, understand that backwards and forwards as ACIO, you absolutely have to. And then from a technology perspective, what are some of the things we've gotta get ourselves, you know, prepared for?

We're gonna be charged with the responsible and ethical transformation or transmission of that data and . We're gonna be on the hook. One, make sure that it gets where it needs to get, and two, that it's done in a secure environment and we're sending what we're supposed to send to the the right party.

It's a tough one. Do I have like a answer right now? Absolutely not. All right. So are there some things that we're looking at? Absolutely. So let me ask you, I'm gonna ask you more specific questions. A lot of times we hear APIs from some sides of the world, and they talk about it like it's magic, like, oh, we're just gonna share the information through APIs.

But it's not magic, right? First of all, you have to write those APIs and it's just like any other code. It has to be secure, has to be. Stable has to be all those, all those wonderful things. So the APIs have to be written, you have to get the data into the form that can be delivered through the APIs is another thing.

We're gonna rely on our EMR to provide those APIs as a lot of people are. They're coming down with the fire APIs and whatnot. Okay, that's all well and good. We're now gonna start having to share that with personal health apps. And that could be you and I in a garage. We just wrote an Android app and we're gonna start collecting that information.

So let's assume we are devious actors. We're not. But if we were devious actors, we would write in our Ts and CSS multiple ways for us to make money. One is, hey, as the patient, we're gonna collect your information, move it into a personal health record, and we're gonna bring your, your exercise data into a personal health record, and we're gonna bring some other data into a personal health record.

And you know what we're gonna do for you. We're gonna make your life easier. We're gonna provide you an exercise regimen, we're gonna remind you to take your medication. We're gonna, all those things, we're gonna do this amazing stuff through this app, but because we're entrepreneurs, we say, okay. How much is the consumer really gonna pay us?

Well, they're gonna pay us $5 a month, $60. Well, we're gonna have to get a lot of patients unless we figure out another way to make money. Another way to make money is there's value in that data. And so in our TSS and Cs, we say, Hey, we're gonna help you with your health, but we also retain the right to use your data and sell your data and do whatever the heck we want to with your data.

And that's buried in a. 14 page Ts and CS document that no one reads, they just click on because they look at what the benefit is and they go, we want that. Alright. As a health system, is there anything we can do to, and, and I'll just tell you, I, I talked about this today, on today in health it, and what I'm proposing is that.

We should have a clearinghouse of some kind that reads the T's and C's at a minimum reads the T's and C's identifies maybe some outliers of what people are doing. And we create notices for our patients. So they sign up for this app and they, they request the information, and as they're requesting the information, we just pop up a little message that says, Hey, just wanna make you aware of the fact that the Ts and Cs from this application will share your information with this, this, and this.

If you still approve, go ahead and click okay and we will share your information. Is that kind of framework possible? We do that in research. If you look at the state of New York with their HIE, it's completely opt in. So what you're doing is you're putting together, and I agree with it, where the patient has the right to opt into those types of ventures or not.

Totally agree with that. And Bill, here's kinda where. The work that I'm doing currently with the ethical and responsible use of this type of data, once you've anonymized it and you've made it so that there's no token attached to it, if I'm a researcher, if I'm pharma or doing genomic research and I find something that previously was undetected, I now have no way to get it back to the patient.

So I'd like to just put it out there. Yes, we're doing a lot of stuff with monetization of data. Yes. Patients are . Approving of that, but are we putting the place guards in place that if we do make a discovery, that we can get that back to the provider and the patient? I know that's a little bit off topic, but folks aren't thinking that way and they absolutely should, but going back to

Your example? Yes. Yes. We absolutely have to do that. And we're expected to, you know, if you look at our consents, there's a line in there that your data and or your specimens may be used for, you know, research purposes. And you can either opt in or opt out of that. So yeah, we have to give them total control.

And Bill, I'll even take it a little bit further that we needed to think about is, um, allowing. Those patients to keep resident on the phone information that does not go any farther than their environment. So yes, they got the, the app builder that's putting all this together and whatnot. But hey, I don't mind your exercise stuff coming on the phone, but my

Personal health record information's gonna stay on here. And guess what? It's not going back to you or anybody else for that matter. Interesting. Charles, I'm gonna give you the opportunity to pick the next topic. I've done call centers development. We've done some data and information blocking. Is there any topic that you've been running across that you think it would be interesting to chat about?

Yeah, I'd love to talk real quick if it's cool. And Bill, if it's just between you and I and it never makes it to the airways , that's, that's totally cool too. And it has to do with, and one of my passions is, is data science. I. And I believe in explainable data science. And I believe that data science products that we're building these algorithms, these models and whatnot, absolutely have to be explainable.

And the participants, those folks that are using them, have to be participating in the build of those. And what I mean by that is they have to understand the data sets. They have to understand the features. Even if it's a neural network, they have to have understanding and they have to be there. When those results are known and the training is done and the confusion metric is produced, and they have to be there to make those decisions on, Hey, clinically I'd rather have a few false positives than false negatives, how can we move this model to, they need to be there for that, these black box type.

Stop. It's hurting right now, bill, and we've gotta all be okay with putting the math out there. It's all math for crying out loud. It's not proprietary. It's, it's not secret. We need it to get out there. And then Bill, I think the last thing I'll say is there are applications now out there that allow a clinician like myself or others.

To build out and do that. I called Citizen Data Science to be able to build out those models and actually deploy those models and they understand the data from the beginning all the way to the end. I have a hypothesis I can now prove out, you know, my hypothesis and then throw it off to the, the, the data science team.

I'm worried, bill, that too much black boxing is gonna hurt healthcare. So, Charles, if, if I'm an entrepreneur trying to. Money in the data science world or the AI world, it's based on the algorithms, right? That's my distinction. It is the intellectual capital, I guess, that we've brought to the table, the people we've brought to the table.

How do we do that? I mean, how do we do that without jeopardizing our revenue model? You can do that. They don't need to see the math, but they need to see the process as you, you walked it through. And more importantly, bill, they need to understand that this model that we built for your town, Baltimore, he said before is gonna have to be tuned for Southern California or Sarasota, Florida.

It's not one size fits all. We're not telling anybody what to do. Right? Right. We're giving 'em information based on these models that help them. Come to a conclusion or cause a cognitive trigger, whatever that might be. This stuff should be considered intelligent assist as opposed to, and, and this prescriptive analytics is gonna kill us.

And I'll tell you why. Those that prescribe do not wanna be prescribed too. Give us information that helps us make a better decision. That's all we're asking for. Don't say you've gotta do this, you gotta do that, whatever. Just hey. Here's what's been going on with that patient. Here's what's going on now.

Here's what's likely to occur. Interesting. So, so that's still happening. Essentially, we still have black box models that are coming out there. Yeah, we still got black box stuff. Synthetic data. Let's get off the synthetic data thing. Let's use real data. Okay. Let's use the real deal. And if you're a brand new startup, ai, there are places you can get de-identified.

Real patient data. If I'm building models out to help clinicians work with and treat patients that are in advanced heart failure, whether it's category one through four, whether they're on an LVAD or some other type of device, and they're on 15 drugs, synthetic data is not gonna help me build that out. I need the real deal.

So how's it going? Teaching the next generation of data scientists? Are you still teaching at Stony Brook? Oh, you better believe it. Again, my first lecture talks about it's 80% subject matter expertise, 20% programming and 20% stats. And I'll tell you, bill Meharry Medical College in Nashville, they actually have a data science program as part of their medical, uh, program.

So they are graduating medical students right now with certificates in data science, but they're well on their way to master's and PhD. I'd like to see that in all disciplines, because to me, uh, physician data scientist is, you know, really something to, to, to strive for and bringing data science in clinically, whether it be physicians, nursing, pharmacy.

I think this is something that, this is my passion, so I'll take this all the way into my eighties teaching at that, at level. So yes, explainable AI is what I teach. Is that still the best way to get educated for the clinicians and people who sort of. Back into data science is still to find a program like that?

Or is there a way to learn from within the industry itself? Sure, bill, we can learn from within the industry itself. Also, we can take advantage of Coursera and others. So let's say I've been in healthcare for 20 years. I was a phenomenal at Stats. I need some programming skills 'cause we got . Figure out how to work with our Python and some other programming languages, but just the fact that I have 80% subject matter expertise.

I know healthcare backwards and forwards, and I have a passion for it. Yes, you can get to that point. And I'd say, how do you do that? Get on the data science team with your organization and just start work helping 'em, and that data scientist that, you know, PhD or or Master's guy is gonna, you know, point you in the right direction.

Fantastic. Charles, always a pleasure to have you on the show. Always fun. Yeah, I hope it wasn't too whack Bill, but I enjoyed it. No, it's always fun to have the CTO in the office. I, I keep a little special section over here where I write down questions. We definitely have more to talk about. We'll have you back on.

It's been a challenge for a lot of people. It's been interesting to say the least. Yep. Well, Charles, thanks again for your time. Cool. Really appreciate it. What a great discussion. If you know of someone that might benefit from our channel, from these kinds of discussions, please forward them a note.

Perhaps your team, your staff. I know if I were ACIO today, I would have every one of my team members listening to this show. It's it's conference level value every week. They can subscribe on our website this week, health.com, or they can go wherever you listen to podcasts. Apple, Google. . Overcast, which is what I use, uh, Spotify, Stitcher, you name it.

We're out there. They can find us. Go ahead, subscribe today. Send a note to someone and have them subscribe as well. We want to thank our channel sponsors who are investing in our mission to develop the next generation of health IT leaders. Those are VMware, Hillrom, Starbridge advisors, Aruba and McAfee.

Thanks for listening. That's all for now.