Aligning your Data Strategy with your Business Strategy

This session explores how data leaders can connect digital transformation initiatives to organizational needs, how to make the business case for using new technology and what technical approaches provide WCG with the data it needs to support business initiatives.


Art Morales, 

Vice President, Data & Technology Solutions at WCG


Speaker 1: DataMasters Summit, 2021, presented to you by Tamr.

Andy Palmer: Hi everybody. I’m Andy Palmer, chairman, CEO, and co-founder at Tamr. I’m really thrilled today to have my great friend Art Morales with me. Art is the VP and head of all data and technology solutions at WCG and WCG is a key provider of solutions that improve the quality and efficiency of clinical trials research. Art is going to talk about all kinds of things, but really specifically how we connect data projects to ROI and drive business objectives, it’s something that we’ve been doing together on and off for more than a decade. So, Art, thanks for being with us today. Great to have you here.

Art Morales: Happy to be here.

Andy Palmer: To start, give us an overview of WCG and what are some of the key facts that people should know about the company? And what makes it unique in terms of your data?

Art Morales: Sure. WCG is a clinical services organization. So the idea is that we try to fill the niches where clinical research companies don’t have that, or we can have a unique expertise, trying to be science-based, in helping companies run the clinical trials. We are a conglomerate of about 34, 35 companies that have been growing by acquisition for over many years. Starting with IRBs, so we own many of the IRBs in the United States, all the way to patient engagement, safety, marketing insights, all pieces of a clinical trial where we can help. That obviously has a number of challenges when you think about bringing so many companies together that they’ve evolved independently, usually small to medium-sized companies, and how to say… The companies come into the WCG, they continue to work as independent companies with their brand and their offerings, and then the idea is, how do we make it better? How do we make the WCG level and the individual company level?

Andy Palmer: So by design, you have a lot of data silos.

Art Morales: Absolutely.

Andy Palmer: Wow, that’s incredible.

Art Morales: They’re still running, right? They’re still developing, and they all have different ways how they evolved, different levels of architecture. Some of them were built on a shoestring, some had more resources and are maybe on version three or four of their tools and some were just started 10 years ago and they’ve been very slowly building without the ability to say, “Let’s rethink the architecture.” “Let’s rethink how we do data.”

Andy Palmer: Amazing. So you’ve introduced a bunch of new technologies at WCG, particularly focused on making better use of all the data at the company as an asset. What advice do you have to offer on making the business case for people bringing new technologies in and making use of their data?

Art Morales: One of the challenges in a company, if you’re a company that’s has a P&L it’s harder. If you’re just the L, you’re just doing research, then is, just like say, “Let’s discuss this research, this technology, it looks really interesting. Let’s bring it in.”

Andy Palmer: Yeah, we used to do that a lot, right?

Art Morales: Yeah, exactly. But now when you’re trying to say, “Well, what is the business case of building something?” So we can’t just bring a new technology because it’s interesting. It has to be tied to saying, “Well, if we bring it in, what do we expect that improvement to be?” And therefore things just because they’re interesting, we can’t just bring it in. We have to say, “Well, here’s the value that we can say.” So quite often thinking about, can we show that there’s an improvement or a proof of concept study? Or is there something that we can say, “Here, look, this is our biggest problem that we have today, if we solve it, this is what’s going to happen and this is a tool that we need to try to solve it.” And, of course, it’s still an experiment, but the experiment has to have a bigger ROI.

Andy Palmer: Wow. And so, when you do this… As we’ve experienced before, it’s changed for people, and getting people to change and do things differently is hard, what are the experience has been at WCG? And what’s worked and hasn’t, in terms of getting people to change?

Art Morales: Well, one of the challenges is that all these data tools are very sticky. Once you bring a data tool in trying to bring another one in, or even if it sounds the same, you’re going to have the challenge like, “Well, you have that, use that.” Or, “Why don’t you use the thing we have?” And the challenge is you can see many of the tools out there, they use the same words to describe the solution.

Andy Palmer: All the marketing sounds the same.

Art Morales: The marketing sounds the same. And we run into this problem with some of our tools that we developed for helping clinical trials, it’s how do you differentiate that? And it becomes an issue of saying, “Well, that’s what we have. In parallel, let’s build something, to say, to compare it.” You can’t stop and say, “We’re going to just migrate it.” You have to say, “Well, we’re going to continue as you are and then I’m going to show you that it works. If I can’t show you, that’s fine, but then I’m not signing for a long-term contract, we just try something, it worked, it didn’t work.” But then at some point, it’s really the comparison to show that. There’s only so much I can convince people without showing them the-

Andy Palmer: Yeah, yeah. Well, those people probably they’ve invested in those other tools, so as bringing in new tech, you have to sort of take this compliment approach-

Art Morales: Exactly.

Andy Palmer: … or compliment them in the short term. Wow, that’s fascinating. So you’ve played around with almost every tool on the planet, probably yourself in some way, shape, or form. Talk to me about like best of breed versus single vendor, single platform kinds of approaches.

Art Morales: Yeah, so that’s an interesting problem because quite often every tool is good at something, but then they convince themselves that they’re good at everything. They’re like, “Oh, because we have the platform, we have the users, marketing says, or BD says, “Hey, I can expand this into all these other pieces and now that’s my new marketing.” But imagine trying to replace a big company that spent millions and millions of dollars on building a platform, trying to say, “Hey, I have a new tool that’s going to replace everything.” Nobody’s going to do that ever. I can replace a piece and continue working or complement, maybe it replaces it later. But, when you go into the big tools, many of these tools are… they become good at everything, but not great at everything, and sometimes they will end up… Because you have a one platform thing, maybe that one piece that it really isn’t that good at, you’re stuck with it because you can’t bring the better tool because everybody says, “Well, it does everything.”

Andy Palmer: And God forbid, that one thing is the most important thing.

Art Morales: Yes, that happens often. Yes.

Andy Palmer: So what are some of the technologies that you’re having the most success with? And what are the things that you like about those technologies or the companies?

Art Morales: So, well, when we think about technologies, one of the things that I’m looking at is data integration and the duplication. Finding tools… That finding what my record is on one company to the next company, to the next company, internally, that’s a big challenge because as we talked about data quality and architecture is different, but also matching it to external sources. So obviously we’re here at Tamr. One of the big companies that I’m very happy with what we’re doing so far is it’s really Tamr on helping us saying that that how do we match the same entity across tools… across databases, I’m sorry. And not only that, but also how do I track down long-term? How do I track the data provenance of that? So one of my challenges is, today I can have data that, this is my present today, it changes two months for now, but I may have made decisions based on this data, being able to track where all that data came from, it’s something that is really important to us.

Andy Palmer: So you need the provenance-

Art Morales: Yeah.

Andy Palmer: … of the data, in order to at least be able to understand what the state of the data was at the time you were making that decision, even if it’s changed dramatically?

Art Morales: Exactly. And not only that, but with GDPR and more data privacy coming into play, when I have a question, that’s a GDPR related question, I need to be able to track it down to the source.

Andy Palmer: Incredible.

Art Morales: And it’s not just as simple as saying, “Do we have it?” Yes, we have it, probably. Where is it?

Andy Palmer: Where did it originally come from?

Art Morales: Where did it came from? Did it come internally? Was it external? Because if it’s external, it’s a different answer than if it’s an internal source.

Andy Palmer: So you’ve got this aggregation of all these different companies. I would imagine that each one of those companies or the people kind of feel like their data sort of belongs in their own little fiefdom. How do you tackle the data ownership questions?

Art Morales: That’s a great problem because everybody has that. You spend all your time working on this and we buy another company, it looks like it, like, “Well, it’s my data.” [inaudible] this stuff. But one of the challenges is this, whenever we do data integration tools, data integration projects, most people that are running data integration projects, think of them as selfish projects. I described them as selfish project, which basically means, “Give me all your data, I’m going to put it together. I’m going to do stuff with it.”

Andy Palmer: Got it.

Art Morales: And the value back to the data producer is not as evident.

Andy Palmer: Interesting.

Art Morales: So one of the things that we’re changing is, we’re saying, “Hey, give me all your data, I still want it. But here’s what I can give you back here. Now, I can not only help you clean your own data, which is important, but I can build services that we can put into your data systems that now make sure that your data forthcoming it’s going to be more clean.”

Andy Palmer: Wow.

Art Morales: So that, if you’re entering a new entity, “Let’s look it up before you enter it. But not look it up on your system, look it up on the clean system that I just spent all this time and effort doing. So that now your data quality goes up and I can give you more data and I can supplement your data.”

Andy Palmer: It’s an amazing dynamic that give and get, right? Like where you really can’t expect people to change and to embrace these new techniques without giving them something back, very tactically, right? Like it’s something really quick.

Art Morales: Absolutely. The question is, “I want this data.” “But, why?” I can go with a hammer, it’s like, “Well, I own you.” But that’s not really the right thing. The right answer is, “I want to get this data to create more value and you’re going to benefit from this value.”

Andy Palmer: Wow, that’s really cool.

Art Morales: I mean, that’s the important part.

Andy Palmer: This idea of controlling data quality at the point of data creation, I think is really powerful one. We’ve seen a bunch of our customers doing, like, they take the master data set, and then they bring it into like an auto suggest, an auto-fill service in their data entry screens and they find just doing that basic thing, which we all expect from modern internet applications, but doing them internally inside of companies it’s a powerful capability.

Art Morales: Yes. For me waiting until the point of data integration to clean the data, it’s an exercise in futility because you’re going to continue doing it all the time, and you’re never going to clean the source data. So you have to go back to the early stage and not only clean it there, but also help the user. So if a user, for example, is saying, “I have a new site, I’m going to enter that information.” If I know that in my integrated layer, I have a lot more data, I can help the user by pre-populating. I don’t need to ask for all that information again. Maybe I want to confirm, but then I can make that experience at the user level, at the business level, just be a lot easier and I also can avoid duplications for many reasons.

Andy Palmer: What’s amazing, we spend a lot of time working with our customers, talking about their data consumers and what they want and need, and not only do we have people doing all kinds of cool analytic things, but also this source remediation and improving data quality in the source is kind of a key thing.

Art Morales: Absolutely. I think getting data from a bunch of different data sources and putting them in a bucket and then searching through it and just find treasure that’s one approach. But if you spend the time and clean the data, you also can provide back to the source, a value and in addition, you can search through. So it’s not just about the search. The search is easy, like to a certain extent that’s just matching and… Well, the user can absolutely say, “Well, these three are the same.” But why not fix it at the source? So that now when you do analytics, your analytics are dealing with cleaner data, that’s easier to maintain clean.

Andy Palmer: It makes it easier, yeah. All these techniques are kind of necessary and important, but none of them alone are sufficient-

Art Morales: Exactly.

Andy Palmer: … in order to solve the whole problem. So let’s switch and talk a little bit more about the consumers of data. So all this clinical data that you guys maintain, I mean, it’s an incredible data asset. Who consumes the data primarily? And what are the kinds of use cases that get you most excited when you see the integrated data?

Art Morales: So we think of data that we have access to us different. So we have the commercial data, which is operational data, and there’s a clinical data. So clinical data it gets tricky because it’s patient data, so we stay away from that. But the aggregate data of how sites are working, how investigators perform, how trials perform, how indications perform, so those kinds of insights are really useful. Because we see so many trials… The majority of trials in the United States at one point or the other, and during the trial, the history of the trial, we can go back and say, “Well, in this indications, these trials, they get delayed more often.” “These trials are more complex.” “These investigators do better with these trials.” These sites may need more help.”
Then once you have those insights, you can use that insight to go to the client and sponsors and CRS and say, “Well, we noticed this. By the way, we have a service that can help you do this.” And that’s where you start to get the ROIs, not just the insight. The insight alone is, “Yeah, that’s interesting.” But when you action on the insight and say, “Well, because we know this, we also have a solution to help you address that.” So it’s all about risk management. “We have a risk, we’ve seen that risk, we can show you what the risk historically is, and by the way, we may be able to help you with that.”

Andy Palmer: So you’re taking these analytic insights and turning them into prescriptive suggestions for the design of trials, which sites you might prefer, which PIs you might prefer, that kind of a thing?

Art Morales: Yes. And sometimes you may have a PI that’s very good at enrolling, but may need help in a different way to maintain patients and so that’s another service. We also look at during the trial was conducting, “Is the data coming out of good enough quality to be able to separate from placebo?” For example. So if that’s the case fine, but if we find sites or trends that are happening at trials that call that into question, we may say, “Well, maybe somebody should have a conversation with that.”

Andy Palmer: It’s amazing. So it’s like this data-driven clinical trial optimization.

Art Morales: Absolutely. 10 years ago, if you asked somebody, “I’m going to look at the data in the trial.” They’re like, “Stay away from me.” It’s a black box. We set it up. We spent two years running this, we set a black box, close your eyes, close your ears, look at it when your data [crosstalk]-

Andy Palmer: Mostly intuitive, right? Like they just, “Well, I know, because I’m the expert and just designed it.”

Art Morales: But don’t change anything. And that’s really the problem. And the field has been changed and especially in the last five years, from Analgesic Solutions solutions where I came from to WCG. We started asking the question, “Well, if you see that there’s something going on through the data, why not try to fix it?” You have to be careful that you’re not biased and, of course, all these pieces. But once you control for those things, you can rescue trials and we’ve shown that.

Andy Palmer: That’s amazing. Yeah, so you adjusting in real-time?

Art Morales: Exactly.

Andy Palmer: Obviously, in the context of the pandemic, these kinds of things have all kinds of dramatic implications.

Art Morales: Absolutely.

Andy Palmer: So it sounds like clinical trials themselves, the nature of how they’re run and how they’re administered is becoming a bit more iterative and a bit more dynamic.

Art Morales: It is. And it’s also one thing that we’re seeing. For example, we saw this at the beginning of the pandemic, getting patients to the clinic was not that easy when we first went with all the shutdowns. So as that started to happen, then you start to see, well, virtual trials are becoming more of a thing, but what does that mean? What is a site anymore? If you have the patient at home and you send a VNA to the patient, how do you manage that centrally? How do you do sample collection? How do you do all the logistics? So those are the things that I’m interested in for the future, but certainly, as the trial space is changing and technology is becoming much more immersed into it. Thankfully we’re not faxing that much anymore, so we can actually look at the data and see things in real-time.

Andy Palmer: Wow. That’s cool, this idea that… Yeah, if you’re data-driven, you can have this long tail of sites that maybe is as small as the patient’s home, and you’ve got data about whether or not that information coming from that site is reliable or not.

Art Morales: You can see in real-time, if that patient or site, et cetera, is providing data, that’s going to help to throve. If you see the problem, if you see a patient that’s not doing consistent data that… You’re asking them, how much pain they have and how they feel in two different scales and they say, they feel great, but they have the worst pain in the world, something’s wrong about that. Well, do you not want to fix that in the middle of the trial? Once you notice, you don’t want to fix them when you lock the database, because can’t do anything about it.

Andy Palmer: Wow. There’s so many different places now where data can be collected, so many different sources of data, are you seeing that in your infrastructure, that you’re getting data from lots of different alternative devices and different things?

Art Morales: We were seeing that. When we started with a data monitoring tools, we were the little monkey in a barrel of gorillas, right? We could never tell people, “This is what you should use.” So we have to be ultimately flexible. And we don’t care. Like, I think to me, when the data’s coming in, has the right formats and consistent formats, and there’s the right safeguards into it, to me, it’s just a data source. And as long as we can trust that data source coming through, there’s a different question of how good they are at collecting data, so some companies are better than others at being more clinically oriented because they’re dealing with a patient than others. So that’s important. But on the other hand, once the data is collected, we’re agnostic, and we see it from everywhere. Especially with the digital medicine, you start to see a lot more data coming through from devices, and so on that, that’s going to really change things quite a bit.

Andy Palmer: Wow. I would imagine… I mean, the heterogeneity of that data is extreme.

Art Morales: Yeah, and not only that, but you then start to ask the question just because you’re collecting data, is it useful? All these devices are collecting good at the consumer level, but are they good for a clinical trial to make decisions on it? And is it really helping you at the digital endpoint, for example?

Andy Palmer: Well, it’s amazing you say that because, I mean, one of the themes that we’ve heard over and over again during DataMasters this year is this idea of not taking sort of big data for granted, like big data exists, but it’s really about the right data and it’s relatively small dataset, ultimately that matters, right?

Art Morales: One of my favorite sayings is that, AI in clinical trials is really hard because you have too many columns, not enough rows. And to actually try to do any insights, just based on the entirety of the data. Number one, most of it is empty, most of it is just noisy, dirty and then you only have a subset of patients. Big trial has 10,000 patients, a huge trial. You’re talking with 200, 300 patients on a phase two trial. How do you do any insights when you have a hundred thousand columns if you connect all the data? There’s just no way to connect that. So you have to be smart about what you’re analyzing. I’m a fan of collecting the data, but I’m also a fan of saying, “Okay, let’s make sure we don’t boil the ocean.” And as data scientist, I’m always like, “Well, let’s boil the ocean and see what happens.” But in clinical, it’s just not deep enough.

Andy Palmer: Well, it’s one of the reasons we love working with you because we always know that the work that we’re doing with WCG is in context of how the data is going to be useful and consumed and isn’t just a big boil the ocean data for data’s sake kind of project. So it’s always fun working with you.

Art Morales: Thank you.

Andy Palmer: So Art, it’s been really great spending time together. Thank you so much for your great partnership and friendship over the years. And we’re both rugby fans, so do you have a prediction for the Six Nations that’s coming up?

Art Morales: I always like it when England wins, but I’m always disappointed.

Andy Palmer: All right. All right. I’m going to go with Ireland this time around, but as always, it’s going to be fun watching it together-

Art Morales: Absolutely.

Andy Palmer: … and following it. And thanks again for coming today and I really-

Art Morales: [crosstalk]

Andy Palmer: … appreciate it. All right.

Art Morales: Thank you.

About DataMasters

We believe in a world where every business is data-driven. Watch DataMasters 2021 on-demand to be inspired by what’s possible when you have clean, curated data. No matter where you are in your digital journey, the content in these sessions will help you accelerate your transformation.