Greg Irwin 0:18
Hi, everybody, thank you all for joining. We have, you know, big thanks to Neil, for pulling together a really fantastic group today. We are here working with Dell and TigerGraph, we're going to be talking about healthcare data analytics. The group is, as you can see, it's data strategy, data architects, data analysts, both from like just BI analytics to architecture. And so people are all coming in from different, different angles. But it's all healthcare, mostly payer providers. So the one thing I want to encourage you, if you've been to any of my sessions in the past, is I'm really big on this idea of leveraging the community. Which means I promise you the most interesting things that you could probably get from this call most likely, is going to be from somebody other than Dell, TigerGraph, and BWG. So look across the Hollywood Squares. If you hear something from somebody, you know, just do a casual outrage, or let Neil and us here, introduce people, you don't have to come through us. But if we could be helpful there, let's take advantage of it. We're going to follow up at the end of this with an email with everybody's name. So if you know, you don't have to scramble to figure out who's who. We are not going to be publishing people's contact info, as you might imagine, but certainly happy to help with any off any options, and mutual beneficiary beneficial connection. Another thing, let's have some fun, and we're encouraging everybody to turn on your video cameras. And one way I like to do that is let's do $100 gift card to whoever has the best zoom backdrop. Show us whatever you've got, the group will vote. And, you know, you'll, if you show us a good zoom backdrop, you get a $100 gift card to keep everybody engaged. To get started here, I'd like to bring in the team over at TigerGraph and the team at Dell. So let's get right into it. Maybe Michael Shaler you can get us rolling here. Real simple. Just give a little intro and just tell us no sales pitch but who's TigerGraph?

Michael Shaler 2:58
So I'll just answer that question. Real simply, we are a startup based in Silicon Valley. We are the leading graph database analytics platform. And we've been in business since about 2012. We were in stealth mode for five years, we just closed a $105 million seed round. So we're doing really well. It's really exciting to see the progress we're making.

Greg Irwin 3:22
Alright, and give us some of the glory statistics like, you know, do you have a big customer? Do you have how many customers the biggest, you know, give us something.

Michael Shaler 3:31
I can't wait to tell you. I can't wait to tell you all about what we're dealing with often. And that's a really exciting use case. And then they are the owner of so much data in the healthcare world. And so they're a great example. My college buddy, Dan McCreary, is doing amazing things with the technology group there.

Greg Irwin 3:51
Cool. All right, awesome. So Michael, thank you. Let's go over to Michael Krogh, as well. Michael, do us a favor and give a quick intro. Tell us about your focus at Dell and how you fit into this world of medical data analytics.

Michael Krogh 4:07
You bet, you bet. Hello, everybody. Thanks for joining in. So just to give a quick intro again, my name is Michael Krogh. I've been with Dell now for 10 years, and I fortunately been able to work in a role that I love in the data analytics and AI space for the entire 10 years I've been here so I've got to see some really cool tech get born like a dedupe and then I've seen others kind of regained their wings like AI. But ultimately I've carried several different roles as part of this team. And currently I'm responsible for developing strategy development for our AI and data analytics practice for Northern America here. I work with some very talented individuals that are on the team here and on this call a new person will chime in. I'm sure at some point Alex Rodriguez, one of our fellow leaders, our objective as a team is to bring forth very innovative solutions that matter for our customers. And we take a very unique role with the way we present solutions as we almost do it in a customer advocate way we cut, we try to turn the tables, and we don't want to look at this from just an ISV benefit or a a, you know, a seller's benefit, but more so what do we think our customers will really like from a solution standpoint. So that's why when Dell got involved with TigerGraph, we recognized that there was exceptional hardware here, or excuse me software here, with fantastic market separation. On top of that, coupled with it, we had an amazing TiER1 healthcare customer as a champion. And we did a you know, fantastic work with them in our labs, where TigerGraph not only outperformed with what, from just a performance standpoint, but also from a capability standpoint, and what was really stood out for me for Dell, as we started to, you know, put together the plan for developing a strategy was not only the de outperform the competition, they flat out destroyed the competition. So it really made sense for Dell to focus a lot of attention, develop a reference configuration, and then a solution so that I can then go out and train our sellers to position TigerGraph as a customer solution. So why would you want to, you know, look at Dell and TigerGraph combined. Again, as I mentioned, we've invested a lot of time in our labs to co develop a reference configuration where we've spent some time testing and vetting a new on the call here really helped spearhead that lab engagement. So really more than anything, what we just want to impress upon you is the level of effort that the solution has that we're presenting to you, and our level of excitement around TigerGraph. So, I'll pause there. I know that was a long introduction.

Greg Irwin 6:55
That was perfect. Perfect. All right, let's go one more. Anu, get in here to come off the beach. And tell us a little bit about what you do over at that. Well, are you new news with Dell also supporting TigerGraph? So tell us a little bit about your

Anurag Juneja 7:11
Hey guys, yeah, this is Anurag here. Sorry, my name was just showing Daniel Aten for some reason, I wanted to introduce myself. This is Anu and I've been part of a pre-sales organization within Dell. I've been with Dell for almost three years now. And my background is mostly consulting. I came from the consulting world; worked for Deloitte for some time before Dell. And my background is completely into analytics, which involves AI machine learning, deep learning extensively into the Hadoop world as well. Streaming ETL space, Informatica. Hadoop has been my bread and butter for almost seven years. And yep, that's my background.

Greg Irwin 7:54
Let's go. And the way I run these is I like to go to stories. So I'd like to hear a story. Tell us about one organization that has basically decided to deploy TigerGraph and why? Because there are a lot of databases that are agnostic databases, there are relational databases, there are time series databases, you know, why should somebody go and consider a graphic?

Anurag Juneja 8:27
Yeah, absolutely. So an organization you know, normally is comprised of a lot of data sources, and we all are living in the big data world right now especially, but talk about health care. Patients and patients are tied to multiple data columns and data records that are associated with each patient, the patient can have a condition, the patient can have zip code locations, and it can have social media data or data related to a family brand. And then how different medical conditions perform at what point a patient was in a different situation at a hospital. So there is a ton of data that is related to every patient and and similarly for any other use case elsewhere and organization having that kind of complex data where you have multiple tables associated with a single entity. That's where the graph becomes really challenging or associating the data different types of data tables for each entity and engaging all of those tables to come up with the end result becomes very challenging not just from complexity from building out filtering out and the data itself but also the complexity is also from hardware standpoint, where you have to end up joining multiple tables and then scaling at the same time and be able to within a predicate that is involved there. So that becomes very challenging because we have to combine and add multiple tables you have to join them Inner Joins, outer joins, and then finally, you're able to see a view, which might not be a good result or be a good result depending on how complex the data is or how much time you've spent into that. Coming back to the patient story, the hospital, Jose, is looking to do some analysis or running some reports on a particular patient and trying to understand different types of conditions, they might end up joining five different tables or 50 different tables depending on what kind of condition they're trying to analyze. And those tables can have their own foreign keys related to other tables as well. So those tables can have their nested joints included in that joining.

Greg Irwin 10:39
What's the difficulty of that? Is that latency, or is that just complexity in terms of structure. From a team leader perspective, if I want to go and get a patient 360 view, what was the problem of doing with Anu?

Anurag Juneja 10:57
The problem is multiple files from data structure standpoint, most of the data these days are nested nature, meaning they involve multiple layers of data within them, for example, we are just talking about customer 360, we might have just a session data from internet session that that customer has, and that those sessions might already have XML data and kind of complex data tied to each other. And those complex data had to be stored in a single column itself, because that's how the nature of the data is right now, which is called nested in nature. And that's like the technical complexity from data structures endpoint. And once you start playing with that nested data, you need a platform that has to be scalable, which is more distributed in nature, decentralized in nature, where we try to think about, hey, we need something like Adobe, and if something like NO SEQUEL platform or decentralized distributed platform that can handle those data structures, and be able to come back with some results. Now once you start doing that the problem becomes latency because you start triggering those complex queries in nature. And those complex query takes your entire platform just single query itself and come back after a certain hour to certain minutes. Now, again, you're stuck with—

Greg Irwin 12:08
Anu, I'm going to cut you off here because I appreciate the detail. And the detail is good. But I don't want to do it unprompted. Because frankly, it does sound like it comes off a bit like we're pitching TigerGraph. And I want to shift your focus to stories and priorities for this group. So let's pull it all the way back. And by the way, we can go deep on any topic here. But let's start at the high level. And I'm going to try and let's drive this with a little bit of the reactions from the group. Everybody here has zoom. My guess is you've all used this before there's a reaction button. Just if you scroll your mouse over the Hollywood Squares, and to vote, just do it with the yellow thumbs up symbol. I'm going to ask a question and let's see what the scores are: are you or your organization currently pursuing a customer or patient 360, consolidated view of the patient or customer? Let's see what we're getting here across 1,2,3,4. I have to score 6,7,8. Yeah. All right. That's what everybody's doing here is my guess. All right, perfect. Let's learn a little bit about that. What I'd like to do is ask you, in your chat window, let's start putting out some specific initiatives. Maybe it's, you know, patient data out of or out of epic, or tying patient data and prescription data, or other types of third party data. Or maybe there's a challenge you want to put out there, you put it in the chat. candidly, we have a phenomenal group on the line here. And let's see if other people are doing similar things. And try to drive a little bit of crosstalk here and there. Chris Otten, I love your backdrop. So I'm sorry if I'm mispronouncing it. Oh, you're not Chris, it's Charles. So maybe I got the last name.

Charles Otten 14:18
You got the last name right. Now you've got the first name right.

Greg Irwin 14:22
Nice to meet you. Do us a favor, give a real quick intro.

Charles Otten 14:26
For me, yeah, please. Sure. So I'm a senior manager for R and D for Zoetis as well as I'm in charge of solution partner work for our R and D genetics group. And my focus is entirely database, data centric and solutions and surrounding that and surrounding genetics data, which gets a little bit tricky and very, very wide. If you've ever worked with genetic data, you're talking like 56,000 columns. And it's and that's a small data set. And, and so that's my main focus. And I was actually about to put in the chat right here, data strategy, and Master Data Management is going to be my next focus. Because when you think R and D, I'm sure not surprising anybody here, r&d data is usually never really good master data.

Greg Irwin 15:18
Right, right.

Charles Otten 15:20
And so we're trying to clean that up, so we can better share it across the organization.

Greg Irwin 15:25
What's the main goal? So you're going to create, I mean, there are a lot of things you could do that might justify putting the team's effort into MDM. But why specifically, is this?

Charles Otten 15:38
So, yeah, so what it's gonna do is, and, yeah, it's gonna be an enabler for us. And the idea of, we have a lot of different shadow IT organizations so to speak, and they'll, they will just go off and do something. And all of a sudden, they're building some infrastructure that we already have. But it's not communicated. It's not centralized. So there's no person on the highest setting up that data strategy. So we're going to set that up, and at least have a starting point to say, you're gonna start with these tools. Now we can, you know, if you want to do something different we can, but we're at least all starting from the same place. And that should simplify the organization and should allow us to share data up, down left, right, so to speak.

Greg Irwin 16:21
Alright, perfect. Hey, Charles, thank you very much. Sure. I'm going to go around the group, I want to ask people about some stories. And my guess is we're gonna learn a thing or two, on Aaron Clark. Aaron, thanks for turning on your camera. And so it's really nice to meet you. Would you give a little intro?

Aaron Clark 16:40
Yeah. I, again, my name is Aaron Clark, and I'm a business analytics and data specialist for Quincy Medical Group, which is a large position though a multi specialty clinic. My background is unique. I'm actually an athletic trainer by trade, and I started off as a clinician, and I kind of moved around through the organization and worked and led quite a few different departments. So our data analytics is actually in its fledgling stages. And with my experience with leading multiple departments across the organization, you know, they asked me to kind of step in this role and see what we can do, we're a little bit unique. And like I said, we're a multi specialty clinic, and we do not have a hospital that we are technically affiliated with.

Greg Irwin 17:34
Okay

Aaron Clark 17:35
We admit to a hospital and do you know, all of our work in the hospital, but there's no affiliation there. We're on separate medical records. Our biggest challenge is taking our clinical data and meshing that with our patients coming from the hospital coming from other Medicare sources coming from, you know, other commercial payers and really trying to figure out the overall strategy, what's going on, again, without a unified record, and without all that lead, we struggle sometimes.

Greg Irwin 18:07
So Aaron, the idea for you is a single patient record.

Aaron Clark 18:15
Well, what does he know?

Greg Irwin 18:17
What's the net goal of putting your time, your team's time into, you know, into developing or hardening your analytics?

Aaron Clark 18:27
I mean the net goal is really to improve patient care and improve outcomes. You know, being able to pull that data together so that we have a background for things that we want things that we need, you know, especially when we try to go to the hospital, who's not only as our partner, but as our competitor, you know, to be able to leverage some of those things to say, you know, here's the science behind what we're doing.

Greg Irwin 18:53
Alright, Aaron, thank you very much. I want to keep doing this. I want to start at a very high level for the next 10 minutes or so, and hear the stories of where the initiatives are because they need to start with an end result. Voitek, it's great to see you. You and I are always on conference calls. So I never get to see you, face to face.

Voitek Gradziuk 19:14
Yes. Finally, we got to see you on Zoom.

Greg Irwin 19:18
I'm gonna ask you for this group to please introduce yourself.

Voitek Gradziuk 19:21
Absolutely. I'm about Voitek Gradziuk, and I've been in the pharma industry since 1994, where we were shipping at Merck research labs, the fax machines, two sides to SAP by fax the queries they have to on paper CRS, and we were, we were installing the having on the HP mini computers, the fax server so I have been in industry for a very long time. I have been in the CRO industry since 1997. It's been a few years at Queen's House, and for the last 14 years I have been with our seniors, which is the top three or top five, CRA, CRA. CRO, and my background is actually in computer science and math and data analysis. So I worked with up on large programs and large meanings with budgets, the clinical budgets over 100 and $50 million on one. And so there's a huge impact to the customer and to the zero on another hand, I work with a small biotechs to start the first phase three or phase two trial. So I have an I'm the technology user, I'm not a provider, what I do in my day job, I designed the randomization systems, and I you know, and supply algorithms and so on. But I talk to people in the industry, from the CEOs of, of the biotech companies, to principal investigators, clinical scientists, and I hear what you're saying, but it seems to me and I have a contrarian approach, because we've been, you know, we've been there before, you know, 20 years ago, we had a big discussion about the EMRs, you know, the electronic medical records. And here we are today. And, you know, for the last, for the last 30 years, or 40 years, we have a big announcement from the AI community that, you know, the artificial intelligence to help us to make sense of the data is around the corner, and we hear about it every 10 years. So I can tell you what I see: seniors were at any given point, we ran anywhere between 500 to 600. clinical trials.

Greg Irwin 21:50
Yes.

Voitek Gradziuk 21:50
Small, you know, with budgets, less than 5 million and large with budgets approaching, you know, 200 million. And I usually where I'm involved is with the data integrity issues, which result in, you know, precisely what you're trying to offer, which is that data analytics and how to prevent them, how to monitor the key performance indicators, and how to be proactive rather than reactive. Unfortunately, you know, my job is more on the reactive side, there's a big issue with data integrity, sites not following the procedures, sites not being adequately trained, patients are not adequately or not correctly, randomized, stratified adults. And there's a big issue for the other part of the backend of data analysis, which is by our statistician to create the data science, the atom datasets to present into the clinical study report data, which is analyzed by our biostatistician. So what I see here, you know, we met with different vendors to do the data analytics. And I think, you know, and that's what I hear also from the principal investigators, there's a lack of, there's a disconnect. There is a lack of understanding by technology, people about clinical and scientific requirements. And there is, of course, in the community, which provides the requirements, how relevant is each data point, how relevant is the value or the range? It's the lack of understanding by the technology people. So I'm sort of in the middle, I talked to both parties, I talked with IT department seniors with what is called now, business technology. So they are technology people like you, but they don't understand clinical requirements. So you may collect, you know, 1000s of data points and like 1000s of tables from different vendors and merge them, but how relevant they are to analyze a project.

Greg Irwin 24:00
I understand the challenges. Give us one initiative.

Voitek Gradziuk 24:04
So we had one issue—

Greg Irwin 24:06
Across all the studies, you know, in terms of a, you know from a data architecture data system perspective.

Voitek Gradziuk 24:18
Unfortunately, I cannot provide you with a good example. What we had, because, you know, part of the problem is that it's not widely deployed at seniors, I don't think we went beyond the pilot stage. This is part of the problem. So we are proposing the data analytic tools to potential sponsors, Big Pharma CROs, other CROs we collaborating

Greg Irwin 24:49
We have our show right here. We In fact, I think we have a couple of your sponsors on the line.

Voitek Gradziuk 24:56
Right, so there's, there's usually a sponsor section During the bid defiance, or the initial kickoff meetings, where some of the tools, what you're talking about are being proposed, but unfortunately, benefits are not clearly explained. So there may be an interest initially. And maybe that we could convince the sponsor to use it on a limited basis, usually cost is the issue. And you have to remember, this tool is one of maybe a dozen vendor tools, which is used in a typical study.

Greg Irwin 25:29
And let's solve it. Let's elevate it. I don't want to talk about products, I want to talk about solutions. And let's bring in Asha, Melanie Ball. I think we have a number of pharma professionals here, Asha, we get you into the mix. Tell us about some of the initiatives you've got on the dataset.

Asha Mahesh 25:50
Yeah, yeah, I worked for the Janssen R&D data sciences. So as part of that we have been very, I would say, really doing well, in terms of integrating data science into the multiple aspects of the clinical workflow, right, starting from drug discovery through the clinical development. So we've been doing it really well, for the last I mean, we also started slow, but how are we picking up steam and we are applying wherever it's applicable, but we're taking a very pragmatic approach and looking at where it actually makes a real impact for the patients. That's where we are applying these data sciences or AI. So one of the issues that I see in terms of linking that I put in the chat is we license tons and tons of external data sources, right, whether it's an EHR claims registry, whatnot, but most of these vendors, they don't allow us to link the data. So that's part of their contract. So that's Unfortunately, that's, that's, that's how it is. So the linking

Greg Irwin 26:58
IQ via script data cannot be joined to other databases.

Asha Mahesh 27:03
IQs script data is okay. But when it comes to patient data, right, electronic health records. So optim, for example, that we talked about optimum optimum does not allow us to link with, let's say market scan, or Yeah, so you take some other premier or something. So it doesn't yeah, that's what I'm talking about. The real world data patient data.

Greg Irwin 27:25
Melody, Melody, I'm gonna bring you in on that. How do you work around it?

Melanie Ball 27:29
You don't. I mean, you work around it because you don't really need them to be linked? And if you do, then, you know, there's definitely an opportunity if you were to if we're able to do that, but you could think from a patient standpoint, would you want that? And also, think about it from an EMR vendor's perspective, they don't want their data linked to claims and claims data providers do not want their data linked to EMR, they have zero incentive to do that.

Greg Irwin 27:59
Right.

Melanie Ball 28:03
So

Asha Mahesh 28:05
yeah, that we use tokenization. That's a permission, what we do is anonymous is okay. But however, it even gets tokenized. To a certain extent that a lot of things get masked. We don't get to see tons of data about patients, but at least we have some of the, I think, critical things that we need to see. So yeah, tokenization is one solution that's in place in

Melanie Ball 28:27
career. But if I'm correct Asha, like you can tokenize it, but you don't have to be able to abstract it more at a sort of scientific level or at a national level, because there's no real, there's no way for you to act on it at a local level. Like you can't go to a physician and say, I see your EMR patients, you can't go to a hospital and say, I see your claims and your EMR, you know, linked together, like no tokenisation will allow you to sort of identify it back to even at a zip code level. You'd have to do claims at a zip code and EMR at some other level of granularity.

Asha Mahesh 29:03
Yeah, that's correct. That's true with the big, big EMR vendors, but there are small-time vendors that are coming up, they have those options from healthcare providers, they are providing that kind of data nowadays. I mean, that's, it's like few vendors that we have seen in the marketplace.

Greg Irwin 29:19
And let's put this into the context of how people are navigating the contractual challenges to still, you know, run analytics that are valuable to the business. So that's what I'd like to hear about. Alright, let's, let's talk about some small wins. We don't have to reinvent, you know, reinvent everything, but maybe just small wins might be really instructive. So, I'm going to put into the group A star five or just jump right in, and I'll start if it's okay with Melanie and Asha. what's what's one small win in light of The the obvious contractual

Asha Mahesh 30:05
one small win I can talk about is on. I mean, I can't really go into details, but one of the vaccination programs that we worked on recently, so we did get approval to link the data because it was really important to monitor the patients. What happens, right? How many actually have to monitor them for the next who participated in the clinical trial. They have to be managed, monitored through the real world electronic health record data for the next two years. So that was something that we got approval for, we were able to work through the contractual limitations to link the data. Yeah.

Greg Irwin 30:48
Wonderful. Man, by the way, I want to raise that because there's a very obvious one, just look at the Johns Hopkins database. There's such a big spotlight now on public health, and tracking public health more so than there's ever been that it creates this. We've had a lot of conversations about general sentiment towards overall public health and monitoring public health and cooperation. And I'm wondering if it's filtering into a specific project activity, at least something over

Asha Mahesh 31:20
the interest of public LTL? Right, correct. That's the reason we were able to work through those contractual obligations. Yeah.

Melanie Ball 31:26
But I think that's separate from linking it to claims data, for sure. On its face, looking at electronic medical record data is great, but like linking it back to insurance claims, at least from the US perspective is probably less savory, but I think, yeah, I think the vaccine use case is definitely a good one. I mean, we obviously do synthetic control arms, for a lot of exceptions or more patient based endpoints that are relevant for filings and for generating evidence for access to therapies, which I obviously can't go into specifics there. But that's a primary use of the patient level. medical record data

Greg Irwin 32:08
monitor. Oh, Asha, Melanie, thank you. Let's keep stirring the pot. And I'm going to call out some folks. Chris Hart. Chris, let's get in here. Mike, shall we bring you back in? Um, boy, I've got 30 odd people here. I think we can learn a thing or two from each other. A mega, Rajon over at Novartis loves to get you involved today. raise hands, ask questions, and do me a favor. Everybody while you're listening, put in one major initiative, that is your organization's work. Maybe it's MDM, maybe it's patient. 360. Let's take it because I think it will help you, Chris. Charles, thank you very much. And with that, Chris Martin, let's get you in here. Do me a favor. Give a quick intro. Yeah,

Chris Hart 33:00
Chris Hart. HCA. So all things data, data warehousing, data integration, betas, all things we are talking about here. And that's interesting. From my side of the house, I'm not necessarily running the challenges of data patient integration, right? Because I own the market. And so I can get the data from wherever and link it across the payers across the EMR, and that actually solves the challenges that we work with and that we're dealing through in the next 24 months, as you know, as healthcare continues to evolve, and the way I access data continues to change. How do I continue doing those things? HCA Luckily, has a pretty mature data strategy and data models. But I think we all see that healthcare is evolving. And the way that we do things and manage current data has to evolve with it.

Greg Irwin 33:45
Interesting. All right. Tell us about a win. All right, we can talk about challenges. Let's look back over the last 18 months. Chris, what's one data when maybe, you know, you have a wide berth here in terms of the relation? You're talking to a lot of data data professionals? Yeah,

Michael Krogh 34:06
I mean, really easy. I mean, I'll say, when you know, we have a centralized data model. Because we operate in multiple states, and the requirements were loosey-goosey coming up quick and fast right across the board in the healthcare world. How do we meet the CMS requirements? How do we meet individual state requirements? And some cases? How do we meet individual county requirements, that data management so you know, having a standardized data model that was pretty mature already, that just needed a few tweaks to account for some of the COVID nuances was a huge win for us as an organization to be able to meet that many state level requirements and CMS level requirements under the constraints that we were all working under.

Greg Irwin 34:47
Thank you, Chris, you have a question. On us the data supports payer reimbursement. Yeah, we

Michael Krogh 34:55
do actually. That's, you know, a big thing last couple years and it's going to be continuing The big thing as we move forward is payer integration, and how do I get data to payers and how to payers get data back to me, so that they can close their curiosity, I can close my care gaps, and that we're not double booking services on patients that have already had a service performed in a different care setting.

Greg Irwin 35:15
But is there one thing I think about, is there a conflict of using that data for you know, the benefit of the payers through not necessarily in the benefit of the patient? Yeah,

Michael Krogh 35:30
Yeah, you know, I think there could be we're getting the data back from the payers is probably the long term goal. You know, I think, first from a reimbursement standpoint, the players really want the data from us. And that's where the reimbursement really comes from, is saying, Hey, you know, you, we need to, you've got the client data already, but this clinical data, so we can help close these care gaps on the payer side, I think there's definitely gonna be questions like that, once we get to the point of being able to ingest data back from payers, and then putting that in your clinical model and, and managing patient care from that point. That's probably another year or two out.

Greg Irwin 36:03
Okay.Hey, Michael. So Chris, thank you very much. And thank you everybody for jumping in. By the way, you can jump into the conversation, just, you know, audio on, unmute yourself and ask a question. If it gets unruly, that's my job. That's my responsibility. But if you have a question, don't be shy, please. I'm going to keep stirring the pot and go back to Michael Shayler. Michael, tell us a little bit about what optim is doing. Again, I want to avoid a sales pitch. So I want to talk so much about just Tiger graph, I want to take it up a level in terms of the challenge. And then the win.

Michael Shaler 36:46
Let me go ahead and share my screen real quick, there are some slip slides, this makes it easier to tell the story. In this use case, we're talking about a call center use case for 15 million Medicare members that are being managed by optim. And in this case, I'm actually a call center Rep. You know, responding to doors, you see or swim lanes on the left, you know, the issues that Asha and Melanie were talking about, you know, are a lot of different data sources, databases, apps and other starting points, that you know, we actually bring into the graph database. And and what what we generate from this is the struggling juicy of Doris his journey through through the system and through her wellness journey, we actually can then find people, you know, as a call center rep like and say, okay, there is you're missing a few visits, we need to make sure we're doing this right, I click on the Find similar members button. And that actually kicks off a graph algorithm that tells me a lot about how people like doors have had successful outcomes. And then I can actually then say, okay, as a call center rep, I can give Doris guidance, you know, go get a iPad with online yoga to help your mental health and in one capacity, in the worst year in the history of call centers, we've really made a big difference in terms of how call center reps are able to have all of that data, and again, with all the concerns and considerations that you've raised, but they're able to actually then identify, here's the outcome. And here's the recommendation. What that means is that we're able to link 12 million entities with 63 billion relationships. And that's access in real time by 24,000 production users. And this really makes a big difference for Doris and makes a big difference for optim. And it makes a big difference for me as a call center rep to see my net promoter

Greg Irwin 38:42
scores go up. So it's exciting, because of the potential for improved outcomes. Have they done any assessment? On your study on outcomes?

Michael Shaler 38:54
Yes, so obviously, there's the call center rep savings. So we were able to cut call times by 10% and deliver better outcomes for the members. That totals about $150 million in savings. But then it also means that we have measured the kind of outcomes for Doris and her cohorts, and we've seen a significant improvement in outcomes in a very challenging year.

Greg Irwin 39:23
Awesome. Okay, very cool. And how long didit take for this, for this program from ideation to implementation,

Michael Shaler 39:35
It took about six months. So they had already started down the path of looking at other technologies they went, they laid out a graph, and then they actually were able to implement it in the call center in very short order. One of the things that made it easy was we were able to provide them with a graph algorithm that we publish on GitHub that you can access that's open source, and that enables them to move very quickly to identify, here's the big win. And then here's how we can make our call center reps much, much more successful.

Greg Irwin 40:05
Cool. Congratulations. That's awesome.

Michael Shaler 40:07
Thank you. You know, it's, it's great because, you know, we're all talking about here. It is powerful because it changes how we think about health care and population health and all the other considerations go into this. And so we're really excited to see, you know, the big breakthroughs that have come through this. Very cool hey, by the way,

Greg Irwin 40:27
We are going to do $100 gift cards to great backdrops. So while you're listening in foreign around here, Charles is in the mix.

Michael Shaler 40:35
Robert, is the head of every good buddy.

Greg Irwin 40:39
But

Michael Shaler 40:39
you know what? I vote for Charles.

Greg Irwin 40:43
I am interested in going to Subic. Dos, and hear something over it. Senator. Do me a favor, please give a quick intro here soon.

Subik Das 40:56
Good morning. Great. Thank you. I'm sure we do. I chaired the multidisciplinary Data Science Center of Excellence. And I also lead the data science operations team that specializes in taking models from different teams and operationalizing. The Mongo Big Data Platform,

Greg Irwin 41:17
give us a little perspective, because I know Center is a big system. Is this across the entire population? Is this one therapy or one? One hospital? What's the purview?

Subik Das 41:31
Yeah. So sutra, as you might be aware, is a 28. hospital system with over 46,000 employees. So I sit in the corporate office, and whatever programs I develop go enterprise wide.

Greg Irwin 41:47
That's exciting. That's really exciting. All right, you know what, tell us a win show, show off the moment. Tell us about what you've been doing.

Subik Das 41:54
Yeah, actually, let me just quickly do one small example of the first point that I will turn over there, which is extracting the discrete data elements from clinical notes. So typically, if you look at heart failure patients, there's a way to figure out whether a person might get a high risk of heart failure. And too often, that information is buried deep inside a doctor's notes, or the echo reports. And it's very hard for clinicians to navigate within the ESR to find these reports, because they they are, they could have been done at external, external health systems, or even if they are done at a certain facility, that deep within the EHR, so too often, that piece of information is missed. So it's called the ejection fraction, which is essentially the volume of blood that the heart pumps out, versus that which comes in. And that piece of information, we are using a combination of NLP and particle clinical rules to pull out. It's a difficult problem, simply because of this ejection fraction number, it appears multiple times in these reports. And therefore, you have to be very careful as you pull that out, and figure out which one is the correct one. Now, this leads to identifying patients who would have been missed. But now we can identify these patients. And at this point, indeed, manage to identify 3700 patients just over the last year, who would have been missed. And we have now put on guidelines based therapies to make sure that they do not suffer. That has big financial consequences well, especially when negotiating with the IRS, because now we can accurately identify which patients are actually at high risk from the hospital.

Greg Irwin 44:13
That's fantastic. To ask the question, is there a regulatory approved methodology for extracting data elements from your clinical notes?

Subik Das 44:25
Yeah, so it depends. It depends on whether you're trying to use it to make clinical decisions, versus using it as an assistive technology to cut down on clinicians time or effort to find information. If you're trying to make clinical decisions based on it, then yes, there's a long FDA approval process for such things. But if you're trying to play in that space where it is simply a tool to Access to directions to get to the data. And then regulations are like.

Greg Irwin 45:09
Great. I am curious, why is it that? Is it lack of training or process? What is that ejection of the heart? The heart ejection fraction, if that's the metric is not being brought

Subik Das 45:24
up? The question is,

Greg Irwin 45:26
If this is an important indicator, why is it not being highlighted? More? More obviously?

Subik Das 45:34
Yeah, so typically, the EHR has to have a field to store this discrete data. But the way the process works out for clinicians, it becomes very difficult and cumbersome to record it simply because the CVS or the cardiovascular imaging systems, which originate the data, first need to be reviewed by the clinicians, and the clinicians can overwrite the values that these PVS systems generate. And then that value needs to be populated with the EHR. Now typically, the output of the CGS system, comes in as raw text, and there is no way to integrate that discrete field value transfer from the CDI system to the EHR system, especially when you have multiple CDI systems. Secondly, from a process perspective, clinicians, and you'll see this in any healthcare system that has gone through acquisitions or mergers, and therefore, processes of treatment and recording information has been historically very varied across different parts of the organization. And therefore, it's a challenge for you to have a uniform process across the system.

Greg Irwin 46:50
Got it. Well, again, congratulations. And now I'm curious. How many opportunities do you see you found? One power thing?Ah, I don't know, discovery within your data? How? How many others do you think there are that just haven't been exploited or discovered room for improvement?

Subik Das 47:18
Oh, I'm sure there are plenty more because this is the example that I gave up the ejection fraction. That's just one of the display data elements across the different disease lines or service lines. Yes, several such reports and doctor's notes, which have critical information. I mean, even in COPD, or potentially, disease of the lungs. When you do tests, like PSP function tests, or pulmonary function tests, the report status PDF scans or image files, and there are several valuable indicators. And that is in there, which we can pull out. Typically, as somebody else mentioned before, the vendors make it difficult to integrate policy because the proprietary nature is a threat to their business. They think because they're all selling analytics products on top of the software. So we have to sort of thread that needle and ensure that we get the disputed elements run and let it sit idly by.

Greg Irwin 48:23
h, excellent. Hey, I Pacific thing, thank you very much. Let's keep going around, I think we can get going on maybe two others if we push it and then we're gonna go back to Mike. And I knew and ramp up here. But I want to make sure like, honestly, my goal is for you all to feel like you want to learn something, make a connection and find value here. So if there's somebody you heard from that you want to hear from, drop it in the chat if you can direct me, all right. It's when it's when I don't have guidance that I'm left to my own devices, which can be dangerous. Okay, let's go on over to knock, knock, knock Rajon I've been meaning to, to invite you in. Please give a quick intro to your focus at Novartis and Alaska. What's one initiative around data analytics that your team is exploring? Yeah, sure.

Nagarajan Dharmarajan 49:21
Thank you. So I lead the analytics at Advanced accelerator applications, which is an award for the company. And it's really focused on radio again therapists our business. So one of the major things that we are trying to do is developing a deeper understanding of the journey of the patient. So essentially, what we are trying to do is that what we have observed is that when there are very complex therapies launched in the market and we are talking about gene therapies, radio ligand therapies, institutions actually have to invest in some facilities or something. They have to put some money upfront, right? So in order to build that facility to be able to treat the patient, this is often a very complex exercise, and because how do they know that the patients are going to be there, after they invest. So that's a challenge. So we'd like to try to understand, you know, longitudinally, where the patient starts, they usually start in a community or with their, with the physicians that they know. And eventually, as diseases progress, they end up at the treating sites. So it's a very long and complex journey. And it poses a lot of challenges to the way institutions are run, you know, you have to have a very holistic approach to interacting with institutions, convincing them of one the therapeutic value of the drug, you know, number two, then they have to invest in setting up the facility. And number three, the institution has to also set up processes end to end to be able to even deliver the drug for this, we need to understand, you know, how many, through how many physicians is a patient typically going through? Right from the time they are diagnosed? That's one of the things we're looking at.

Greg Irwin 51:15
What's the capex involved in standing up one of these facilities?

Nagarajan Dharmarajan 51:19
It really varies from Institute to institute. I mean, I wouldn't have a number off the top of my head, maybe it varies across the country action

Greg Irwin 51:30
So what inputs, how are you determining it now? What are the key inputs?

Nagarajan Dharmarajan 51:35
So we typically use as I think Melanie and Shawn were mentioning, we do look a lot at the claims data. And the other aspect that we're looking at is also the EHR data is something that, and actually, it was wonderful to face the same challenges that I just mean, we struggle to link these to these datasets, these reasons. So I think, if there could be a way to link these datasets properly, although there's no incentive, we have to use our judgment, right? I mean, we look at electronic health records and we try to figure out okay, at this stage, if the patient comes like they might not be ready for treatment, right, or he's past the stage where he will really realize the value of the treatment. Right? So that's, that's where I mean, it's a very relevant challenge that was already put forward by Melania. And dosh, actually,

Greg Irwin 52:28
Excellent, excellent. Let's, and let's try maybe one more. How about who are we missing here? Oh, Chris McGee,are you still with us? I saw you joined a little bit earlier over at Cigna. Now, maybe we lost crews. Alright. How about Deb? Deb. calves? Alberta? Alberta Health. Debbie with us?

Deb 52:56
Oh, good morning.

Greg Irwin 52:57
Good morning.

Deb 52:58
Oh, Hang on. Hang on. Hangon.

Greg Irwin 53:02
Turn your camera on. Man was one.

Deb 53:11
Morning.You hear me?

Greg Irwin 53:13
Yeah. crystal clear.

Deb 53:14
There's too many headsets, too many headsets? No, the one with the mic tends to run out of power. So I was just listening. Um, our experiences are somewhat different to our American colleagues. So I work with Alberta Health Services, and I work in the area of seniors health. And, and I was really gonna pronounce your name wrong. starts with a V. grads. Does it seem easier to pronounce your first name? How do you want to anyways, that I'm going to remember now because it runs a boy check, which is easy to remember. Anyway, some salon experience with Alberta Health Services is to be totally blessed. We have for over 4 million Albertans 4.3. I think our seniors are around 600,000. We have access, we got it all. We've got the hospital discharge abstract data, we've got the ambulatory data, we've got the physicians claims, we've got the community pharmacies, we've got the long term care ride data, and we've got the community ride data. And our challenges are just our biggest challenges actually are having enough well, or trying to have enough pans to answer the questions. But if I had to say what is our if somebody so deep, what's your number one question, what's your number one challenge every single day and that is his being the recipient of a good question, because I don't think people know what we're talking about being evidence informed. We're talking about being data driven. But the reality is that in the same way as we have literacy for four words, and we think of illiteracy in terms of not being able to read, we have numeral literacy in terms of being able to read or understand numbers and graph literacy. I don't know graph illiteracy, I don't know how to pronounce it. And most people can't read a graph. Most people are table challenged. And we use Tableau in our ecosystem. And we've got Oracle databases, and we're implementing epic for acute care. And it's just, yeah, it's like having a bowl of spaghetti. And trying to lay the noodles straight, then yeah.

Greg Irwin 55:28
In the realm of data and analytics, what's one initiative for your system for your health system?

Deb 55:35
Just one,

Greg Irwin 55:36
Yes, just one.

Deb 55:41
You know what I can't, I can't distill it down to that. We have a growing aging population. It's not a gray tsunami. Tsunami is a wave that crashes over your head and sweeps you out to sea and you don't come back. It's not any of those things. It's actually and this is not my words. These are, I think, Michael Radcliffe. But anyway, it's a great glacier. And the only way to get knocked down by a glacier is to stand still for a very long time. And I think our healthcare systems, which were an exception to the current COVID responses, are standing still when it comes to caring for a growing aging population. And our population is going to go from where it is right now. 15 to 16%, up to about 30, and what they call the dependency ratio, which is 15 to 64 year olds paying for everything under 15 and over 64 is shrinking. And this is an international problem. And I think the solution lies in really good use of data and analytics. But we have a long, long, long way to go. And Wojciech, I think, you know, your career is a testament to how far we've come. But to quote somebody that I heard from, I think, KPMG at a conference, Somebody said, Well, now that we're all working from home, are we digitized? No, no, we're just working from home. We aren't there yet. And we put colleagues on the phone, I mean, it's a wonderful, wonderful place to be working in the 21st century.

Greg Irwin 57:13
I agree. I agree that thank you very much. Our we're going to wrap up our session. Let's go to Mike, Mike and Mike, for the takeaways. Before we go to these gentlemen, we're gonna follow up with the emails, I promise. Obviously, Del and TigerGraph are pushing very hard in terms of innovation around, you know, pulling together extremely large, complex data sets. If it's relevant for you or relevant for a colleague, of course, they'd like to speak with you. And any of those intros or meetings would be very much appreciated. With that, let's go over to Mike crow first, and then Mike Shayler to wrap up our session, guys.

Michael Krogh 57:57
Okay, and I'll keep it tight. You know, the objective at Dell, obviously, is we want to speed the implementation of helping your organization get a technology like Tiger graph deployed, Michael is going to talk to you about speaking on speed to insight, which I think is hyper critical for the roles you're trying to accomplish. today. combined together, we bring together a solution to really accelerate that process across the board. So it will be respected and, and thanked by your IT organizations as well as your business organizations. That's what we bring to bear. Our team is ready and willing to share with you what we've done from our laboratory perspective for testing out TigerGraph on our hardware, a very prescriptive set of hardware. So please reach out. We're happy to chat with you. And I'll hand off quickly to Mr. Shaler. Thank you,

Michael Shaler 58:47
you so much, Greg. Neil, everyone, thank you so much for being such gracious hosts and participating here. Learned a lot today, we still have some fundamental problems. We all have solved it for this industry. And you know, privacy, plus, you know, kind of population health is an interesting tension and dynamic. We think that we can help with, you know, kind of graph analytics. But we also still need to do a lot of work around how we manage the security, how we manage the critical data elements, and how we really make sense of what is going to drive the kind of outcomes that we were looking for here. So you know, Tiger graph is excited about being able to help partners and customers. Dell, thank you so much for your time today. And thank you to everyone for your participation and a chance to learn more.

Greg Irwin 59:39
Awesome, and congrats here to Charles and Skyler. I saw you playing along and Christina, I saw you playing along. So congrats to the three of you. We look forward to following up. And everybody stay safe and have a great day. Thank you.