Revolutionizing Healthcare with AI and Data Analytics: A Conversation with RJ Kedziora

[00:00:02] [SPEAKER_00]: I like use cases. It's a good way to give us a sense of what's going on in a particular

[00:00:06] [SPEAKER_00]: space and get good ideas about things we might be able to work with our customers. So I had

[00:00:11] [SPEAKER_00]: the opportunity to talk to RJ Kedziora, who's the co-founder of Astenda Solutions, and they

[00:00:17] [SPEAKER_00]: focus on software and data analytics for the healthcare market. He joined me on this bonus

[00:00:22] [SPEAKER_00]: episode of the Business of Tech.

[00:00:27] [SPEAKER_00]: Thread has declared death to the ticket. Did you know tickets date back to the 1800s? And

[00:00:33] [SPEAKER_00]: customers hate tickets. Thread uses an approach for connecting, communicating, and ultimately,

[00:00:39] [SPEAKER_00]: collaborating. Thread allows people to come together around a topic. They can discuss

[00:00:45] [SPEAKER_00]: and make decisions. They can share and invite the right people. The future of service is

[00:00:50] [SPEAKER_00]: collaborative.

[00:00:51] [SPEAKER_00]: Supercharge your service experience by seamlessly integrating AI and automation to meet your

[00:00:56] [SPEAKER_00]: customers where they are – in Teams, Slack, and desktops, not in tickets. With instant

[00:01:02] [SPEAKER_00]: updates to and from ConnectWise, Autotask, and Halo PSA, keep all communication in one

[00:01:08] [SPEAKER_00]: place where it should be – with people. Decrease time to resolution 30% with chat-based

[00:01:15] [SPEAKER_00]: support by visiting getthread.com slash MSPRadio to declare death to the ticket.

[00:01:24] [SPEAKER_00]: Well, RJ, thanks for joining me today.

[00:01:28] [SPEAKER_00]: Thank you for having me. Glad to be here.

[00:01:30] [SPEAKER_00]: I'm going to start with the easy one. Tell me a little bit about what your firm does

[00:01:35] [SPEAKER_00]: so that we can get a sense of where you fit in the ecosystem.

[00:01:39] [SPEAKER_01]: Yeah. Astenda Solutions is a company I co-founded, been around for 20 years, which is always

[00:01:46] [SPEAKER_01]: crazy to talk about. We are the custom software data analytics and now AI company that helps

[00:01:54] [SPEAKER_01]: other entities create products that they're taking to market. Very much focused on the

[00:02:01] [SPEAKER_01]: innovative side in the world of digital health. We've focused on digital health before it

[00:02:06] [SPEAKER_01]: was called digital health, which has been quite a fascinating journey over the last

[00:02:11] [SPEAKER_01]: 20 years. But mix of customers from Fortune 50, large medical device manufacturers, and

[00:02:20] [SPEAKER_01]: also startups working with them on developing those solutions. Always much focused around

[00:02:26] [SPEAKER_01]: the idea of improving health and wellness of our fellow humans.

[00:02:31] [SPEAKER_00]: I'm sure you've got one or two use cases that are always your great examples of work you've

[00:02:35] [SPEAKER_00]: done. Give me a sense of what those are.

[00:02:37] [SPEAKER_01]: Yeah. Two quick projects come to mind. One, we think of it as a flagship software project

[00:02:43] [SPEAKER_01]: that we've been part of for the 20 years now. It's through a partnership with the Jolson

[00:02:48] [SPEAKER_01]: Diabetes Center, working with the Indian Health Services here in America, which is

[00:02:53] [SPEAKER_01]: responsible for the American Indian, Alaskan Native population. It's a diabetic retinopathy

[00:03:00] [SPEAKER_01]: surveillance program. So as you have diabetes, which is a chronic condition, it can impact

[00:03:07] [SPEAKER_01]: your vision. And so diabetic retinopathy is the leading cause of preventable blindness.

[00:03:12] [SPEAKER_01]: And we're part of a network of cameras, about 100 deployed throughout the US now,

[00:03:18] [SPEAKER_01]: where we can take images of the back of your eye, the retina of your eye for detection

[00:03:22] [SPEAKER_01]: of diabetic retinopathy. And through that program, we integrate the medical record systems and do

[00:03:28] [SPEAKER_01]: extensive data analysis and research as part of that. And in terms of AI, retinopathy solutions

[00:03:37] [SPEAKER_01]: and looking at those images as some of the earliest approved by the FDA systems. So that's

[00:03:43] [SPEAKER_01]: one of the large scale programs that we work on. But then if we look at the startup realm,

[00:03:49] [SPEAKER_01]: we've helped some companies create digital weight loss solutions, looking at data and

[00:03:55] [SPEAKER_01]: bringing together that wearable information, making sense of your daily situation, what's

[00:04:01] [SPEAKER_01]: going on with your life and bringing together that medical information to create a comprehensive

[00:04:08] [SPEAKER_01]: picture of the patient. And then more recently with the advent of, I'll call it the advent of AI,

[00:04:15] [SPEAKER_01]: but everybody sort of forgets AI has been around for quite a while now, decades. It's not something

[00:04:21] [SPEAKER_01]: that just jumped on the scene. We've been involved in various AI systems over the years, developing

[00:04:28] [SPEAKER_01]: medication recommendation systems that were expert system based back in the early 2000s.

[00:04:34] [SPEAKER_01]: And today now implementing gen AI types of solutions to address the challenges in healthcare.

[00:04:41] [SPEAKER_00]: Now, I've got a couple of questions you led me right to the first one. So I'm with you. Like

[00:04:47] [SPEAKER_00]: we've been talking about AI, it was there's machine learning before that there's been data

[00:04:50] [SPEAKER_00]: analytics. We all remember big data. There's been lots of trends where there's kind of this

[00:04:55] [SPEAKER_00]: thread through this, but something is different about this kind of moment in time. And it could be

[00:05:02] [SPEAKER_00]: a way it could be an awareness piece. It could be the breakthrough in what generative AI can do.

[00:05:10] [SPEAKER_00]: You know, the models have reached a level of maturity. It could just be funding. Like,

[00:05:14] [SPEAKER_00]: I want to ask you, like, what do you think is different about this moment than the previous

[00:05:22] [SPEAKER_00]: work that you've been doing over the past 20 years before that?

[00:05:27] [SPEAKER_01]: Yeah, it's the ability for the average person, the non technical person to take advantage of

[00:05:36] [SPEAKER_01]: this technology and really make an impact. So going back, I referenced the medication

[00:05:42] [SPEAKER_01]: recommendation system we did and we're ingesting large amounts of data and developed an extensive

[00:05:48] [SPEAKER_01]: expert system behind the scenes. And then the presentation for that in the user interface was,

[00:05:55] [SPEAKER_01]: you know, very much of that style of the early 2000s. And fast forward today to create that same

[00:06:03] [SPEAKER_01]: system, we can now use extensive APIs and the power of gen AI to craft those rules and figure

[00:06:11] [SPEAKER_01]: out what we want to be able to do with them and present it to the user in a very useful,

[00:06:18] [SPEAKER_01]: interactive fashion. But a lot of what we do in those machine learning algorithms and gen AI today,

[00:06:24] [SPEAKER_01]: it's all about the data. And one of the biggest shifts is, you know, I learned SQL, which is a

[00:06:32] [SPEAKER_01]: programming language to be able to interact with data. And it's relatively easy to learn in the

[00:06:39] [SPEAKER_01]: grand scheme of things, but it's something you have to learn and understand and how to use it

[00:06:44] [SPEAKER_01]: effectively. Where now we're developing systems that put the gen AI on top of that data. So the

[00:06:52] [SPEAKER_01]: average person doesn't have to learn or understand tools and technology and can start querying and

[00:06:57] [SPEAKER_01]: looking at that data. You don't even, you know, people generally start with like Excel, but you

[00:07:02] [SPEAKER_01]: don't even have to start with Excel and learn how to use Excel. You can start querying the data and

[00:07:07] [SPEAKER_01]: okay, what does my patient population look like today? How many people do I have that are over

[00:07:13] [SPEAKER_01]: 60 with diabetes and start looking at all sorts of different situations without that deeper

[00:07:20] [SPEAKER_01]: understanding of technology. And I think that's the power of what's the inflection point that

[00:07:25] [SPEAKER_00]: we're seeing today. It's interesting because I want to ask you then about as somebody who's

[00:07:29] [SPEAKER_00]: been doing a bunch of product development, so I've served with some product organizations for a period

[00:07:33] [SPEAKER_00]: of time and I always felt like the report builder was the bane of our existence, right? The fact

[00:07:40] [SPEAKER_00]: that you end up in a situation where you're trying to give customers data and you're trying to give

[00:07:45] [SPEAKER_00]: them access to it. And either they have, you either build a bunch of pre-canned reports and then they

[00:07:50] [SPEAKER_00]: always tell you they can't get enough data out of it or you build a very intricate report builder

[00:07:54] [SPEAKER_00]: and no one ever uses it. Are we at the moment now where like that is the problem that we're able to

[00:08:00] [SPEAKER_00]: solve in a smarter way now? And what does that look like? Is it a chatbot? Is it voice?

[00:08:06] [SPEAKER_00]: Is it a drawing tool? What are you seeing in this report space that's so different?

[00:08:13] [SPEAKER_01]: That is one of the challenges. So I think Gen I has sort of been searching for the solution and

[00:08:20] [SPEAKER_01]: where does it add value and make sense? Everybody, you look at the startup world now and everyone's

[00:08:25] [SPEAKER_01]: like, we use AI and okay, do you really? But in that report space particularly in getting data

[00:08:31] [SPEAKER_01]: out of a system is huge. So in the past we would have created and still do have these projects

[00:08:38] [SPEAKER_01]: where we're creating the very static controlled reporting solutions that they typically go to

[00:08:45] [SPEAKER_01]: management still, the monthly reporting, things that are very iterative in nature,

[00:08:49] [SPEAKER_01]: that still very much exists. But as we work on different projects and make the data available

[00:08:56] [SPEAKER_01]: to people to look at and understand in search, again, you don't want to learn that skill of

[00:09:03] [SPEAKER_01]: creating the report or go through the exercise, product development exercise and figure, okay,

[00:09:10] [SPEAKER_01]: what are the parameters we want to display? And okay, do we use a pie chart or do we use a

[00:09:15] [SPEAKER_01]: bar chart? What makes sense to display this data accurately? Just let people have at it.

[00:09:21] [SPEAKER_01]: And it's through that natural language interface where you're just typing,

[00:09:25] [SPEAKER_01]: find me all the patients that have diabetes or find all the patients that have congestive

[00:09:31] [SPEAKER_01]: heart disease and what are their comorbid conditions is a very easy thing to stay

[00:09:38] [SPEAKER_01]: in natural language, speaking like we are today. It's easy to do versus writing a SQL query, which

[00:09:46] [SPEAKER_01]: we won't go into, but it does provide, you need that level of expertise and data analysts and

[00:09:54] [SPEAKER_01]: data analysts aren't going away anytime soon. They're definitely a necessary group of people

[00:09:59] [SPEAKER_01]: that we need in the world to figure that data out. But more people can get to this data now,

[00:10:05] [SPEAKER_01]: whether it's just through typing or even voice mode that some of these systems are creating.

[00:10:11] [SPEAKER_00]: Now, what do you think are the prerequisites for customers to be ready to take advantage of this?

[00:10:17] [SPEAKER_00]: So we, it used to be, of course, everything had to be incredibly structured, but I'm not buying

[00:10:22] [SPEAKER_00]: into the fact that everything can just be chaos and then it will all get sorted out by AI. Give

[00:10:27] [SPEAKER_00]: me a little bit of a sense of what's necessary and where the opportunity is to sort of help

[00:10:32] [SPEAKER_01]: customers get their data ready. Yeah, it's a great point because one of the starting conversations

[00:10:38] [SPEAKER_01]: I have with a lot of people, they come to us and they're like, we need to create an AI strategy.

[00:10:43] [SPEAKER_01]: Okay, great. Let's start with a data strategy because it is, you can't just have a massive

[00:10:49] [SPEAKER_01]: information that is then understood by an AI. You need to provide that structure or methodology

[00:10:58] [SPEAKER_01]: to be able to put that data in a format that is usable by the AI, whether it is unstructured data

[00:11:05] [SPEAKER_01]: out of a PDF document or your electronic medical record system. Now, it has to have an understanding

[00:11:12] [SPEAKER_01]: of what that information is and available. And then from that data perspective, all of the

[00:11:19] [SPEAKER_01]: cybersecurity, privacy guidelines all apply very much to this data. As you are delving into

[00:11:29] [SPEAKER_01]: healthcare information and HIPAA guidelines come to play very quickly in terms of how you access

[00:11:36] [SPEAKER_01]: that data and make it available. So AI is important, but it's really the underlying data

[00:11:42] [SPEAKER_00]: which is driving those conversations. And are there particular frameworks that you're embracing for

[00:11:48] [SPEAKER_01]: that, that have accelerated the ability to help customers? Yeah. So Estenda as an organization is

[00:11:56] [SPEAKER_01]: ISO 1345 certified, which basically says we have a good well-documented software development process,

[00:12:03] [SPEAKER_01]: but it's really driven by checklists. It's, okay, let's make sure we are thinking about X, Y,

[00:12:10] [SPEAKER_01]: and Z, thinking about privacy, thinking about permissions. We typically start with what's

[00:12:17] [SPEAKER_01]: called a zero trust architecture. It's like, as you get logged into the system, you have no access

[00:12:21] [SPEAKER_01]: to anything. And then I have to apply those principles to it. So as we think about these

[00:12:27] [SPEAKER_01]: projects and develop them from the ground up, typically it's applying those checklists.

[00:12:33] [SPEAKER_01]: And okay, we might not implement feature A or X today, but at least we've thought about it.

[00:12:40] [SPEAKER_01]: And then that can be applied to a later stage. But there are a number of ISO standards out there,

[00:12:46] [SPEAKER_01]: which are incredible in terms of framing these conversations, in terms of quality,

[00:12:52] [SPEAKER_01]: in terms of security, of how you don't have to reinvent the wheel every time. In terms of

[00:12:59] [SPEAKER_01]: cybersecurity, the government has NIST guidelines that are available. OWASP out there has extensive

[00:13:07] [SPEAKER_01]: guidelines. HIPAA has guidelines, it's less prescriptive, but definitely applies. And the

[00:13:14] [SPEAKER_01]: thing to keep in mind with all of this is technology is important, but it's very much

[00:13:20] [SPEAKER_01]: about people and processes as well. When you are implementing a cybersecurity system or protecting

[00:13:26] [SPEAKER_01]: your data, looking at privacy, you can't forget the people and the processes in place to control

[00:13:32] [SPEAKER_00]: that data. So give me a sense of how you're using AI for the development process. Is that now

[00:13:39] [SPEAKER_00]: something that is infused into your developers? Are you writing code with it? What are you finding

[00:13:44] [SPEAKER_00]: that the practical applications in your own organization are? It's interesting because

[00:13:51] [SPEAKER_01]: as time goes on, you're not going to be able to avoid it as a developer. It is just there.

[00:13:58] [SPEAKER_01]: It's being embedded into the tools and processes that we use all of the time. So you're not

[00:14:03] [SPEAKER_01]: avoiding it. It's then how do you take more distinct advantage of those tools and technology?

[00:14:09] [SPEAKER_01]: I went through an exercise very recently. I was talking about enabling a system to do

[00:14:17] [SPEAKER_01]: querying of data through natural language, and I used a new API to make. I came up through the

[00:14:23] [SPEAKER_01]: programming ranks, but I don't program anymore. But very quickly using Chats GPT, I was able to get

[00:14:29] [SPEAKER_01]: Python running on my system, figure out how to use the API and create a solution. It's a demo

[00:14:35] [SPEAKER_01]: solution. It's not production worthy by any means. So you do have to make sure that you just don't

[00:14:43] [SPEAKER_01]: take what is being offered to you and plug it in there. You really do have to make sure you

[00:14:49] [SPEAKER_01]: understand that. And it doesn't create the perfect solution every time, not because it can't, but

[00:14:56] [SPEAKER_01]: because it doesn't have enough information in the moment. So if you ask Chats GPT, write me an

[00:15:02] [SPEAKER_01]: algorithm to do X. Okay, it can do that. It's not going to get it right all of the time, but it's

[00:15:09] [SPEAKER_01]: going to give you that jumpstart to understanding it. And then you have to understand the framework,

[00:15:13] [SPEAKER_01]: what else is going on within your production application to make sure it's sense. There's

[00:15:20] [SPEAKER_01]: an aspect of copying and paste. Just don't copy and paste it. Make sure you understand it.

[00:15:26] [SPEAKER_01]: We use it there from a software development perspective, but also from that higher level

[00:15:31] [SPEAKER_01]: of product development perspective of requirements analysis and figuring out what needs to be done.

[00:15:37] [SPEAKER_01]: I love just asking it, what have I forgotten? So I talked about these checklists. It's the same

[00:15:43] [SPEAKER_01]: type of idea. It's like, okay, here's what we're doing with the system. Here's what I'm thinking

[00:15:47] [SPEAKER_01]: about. What else do I need to consider to be able to help jumpstart those thought processes?

[00:15:53] [SPEAKER_01]: And then on the back end, from a quality perspective, I put together a webinar a

[00:15:58] [SPEAKER_01]: couple of months ago of just how to use it for generating data in medical scenarios. So you can

[00:16:04] [SPEAKER_01]: sit there and be like, I have a patient with diabetes, generate me a blood glucose profile

[00:16:10] [SPEAKER_01]: for this patient where they eat meals three times a day. And it generated that information for me.

[00:16:17] [SPEAKER_01]: And then I was able to change it and tweak it very quickly. So instead of like, okay,

[00:16:23] [SPEAKER_01]: well, they had a bigger meal at lunch. So what is the impact on the blood glucose?

[00:16:28] [SPEAKER_01]: And typically, it goes higher in that situation if you have diabetes, and it was able to generate

[00:16:33] [SPEAKER_01]: that data. So yes, I could have done that myself, but it's about efficiency. Here's a great example.

[00:16:44] [SPEAKER_01]: I've been doing user interviews for decades now. It's very easy to put those together,

[00:16:51] [SPEAKER_01]: those questions, but it still takes some time. I had to email a PhD about olfactory testing,

[00:16:58] [SPEAKER_01]: testing the loss of sense of smell. Plugged it in, plugged the individual's profile.

[00:17:05] [SPEAKER_01]: And it was like, generate me 10 questions to ask as the interview as part of this application

[00:17:11] [SPEAKER_01]: we're developing. Boom, boom, boom, boom. I changed them and tweaked them so it better fit

[00:17:15] [SPEAKER_01]: my style, but it saved me a couple hours of time of putting that together, that list.

[00:17:22] [SPEAKER_00]: Definitely advantages in terms of efficiency. So final question, what are customers asking

[00:17:28] [SPEAKER_00]: about? What is the conversation you're having with them right now about what they're asking for?

[00:17:35] [SPEAKER_01]: It's a lot of how do I get started? Where do I get started? How do I take advantage of this?

[00:17:40] [SPEAKER_01]: Where do I take advantage of it? And that's where we then very quickly shift to the data

[00:17:45] [SPEAKER_01]: conversation. So yes, AI policy process, play with it, very important, but it's then let's shift.

[00:17:53] [SPEAKER_01]: Okay, you really need to understand the data. And that's where a lot of those conversations drive.

[00:17:58] [SPEAKER_01]: Healthcare is the last industry to really embrace the idea of data. They've totally embraced

[00:18:04] [SPEAKER_01]: technology. You look at your MRIs and all sorts of lab testing that's available and

[00:18:11] [SPEAKER_01]: genetics, cancer research, so very much embrace technology, but it's new to the idea of data.

[00:18:19] [SPEAKER_01]: So still lots of challenges there. And that's where those questions quickly go.

[00:18:23] [SPEAKER_00]: RJ Kedziura is the founder of Estenda Solutions, which specializes in software and data analysis

[00:18:29] [SPEAKER_00]: for healthcare and medical companies. He's an expert in software development, data analytics,

[00:18:34] [SPEAKER_00]: and a focus on the healthcare market. RJ, thanks for joining me today.

[00:18:38] [SPEAKER_00]: Thanks for having me. Great conversation. Looking to reach an audience of thousands

[00:18:43] [SPEAKER_00]: of MSPs and IT service providers? Put your ad right here on the Business of Tech and be on

[00:18:50] [SPEAKER_00]: the show that 64% of MSPs report having listened to a recurring top 50 tech news podcast. There are

[00:18:58] [SPEAKER_00]: affordable options for you to reach our audience and we can support any budget. Podcast listeners

[00:19:04] [SPEAKER_00]: are more engaged, have a higher level of brand retention, and are more willing to listen to ads

[00:19:10] [SPEAKER_00]: here than any other avenues. Want to know more? There's information at mspradio.com slash engage,

[00:19:18] [SPEAKER_00]: including a button to book a time to talk. I'm looking forward to that discussion.

[00:19:25] [SPEAKER_00]: The Business of Tech is written and produced by me, Dave Sobel, under ethics guidelines posted

[00:19:31] [SPEAKER_00]: at businessof.tech. If you like the content, please make sure to hit that like button and follow or

[00:19:38] [SPEAKER_00]: subscribe. It's free and easy and the best way to support the show and help us grow. You can also

[00:19:44] [SPEAKER_00]: check out our Patreon where you can join the Business of Tech community at patreon.com slash

[00:19:50] [SPEAKER_00]: MSP radio or buy our Why Do We Care merch at businessof.tech. Finally, if you're interested

[00:19:58] [SPEAKER_00]: in advertising on the show, visit mspradio.com slash engage. Once again, thanks for listening

[00:20:05] [SPEAKER_00]: to me. I will talk to you again on our next episode of the Business of Tech.

[00:20:12] [SPEAKER_01]: Part of the MSP radio network.