
Chris Belmont, VP & CIO, University of Texas MD Anderson Cancer Center
Why would a CIO want to lead a major Epic rollout after having recently completed the task at another organization? For Chris Belmont — who was asked this very question by his son — the answer was simple: it was a chance to do it better, using the knowledge gained the first time around. In this interview, Belmont talks about how an implementation is like having a child, the two biggest challenges in a major transformation, and how an organization can benefit from disruption. He also discusses MD Anderson’s Moon Shots program, the work his team is doing with IBM’s Watson, how he hopes to improve the patient experience, and the importance of mentoring.
Chapter 2
- Rolling out Epic… again
- The power of hindsight — “Things go through a traditional life cycle and you can expect it.”
- Moon Shots Program
- Training Watson “to think like an oncologist”
- Structured data & finding “gold nuggets of knowledge.”
- Patient experience — “We can always do better.”
LISTEN NOW USING THE PLAYER BELOW OR CLICK HERE TO SUBSCRIBE TO OUR iTUNES PODCAST FEED
Podcast: Play in new window | Download (Duration: 15:53 — 14.5MB)
Subscribe: Apple Podcasts | Spotify | Android | Pandora | iHeartRadio | Podchaser | Podcast Index | Email | TuneIn | RSS
Bold Statements
The reality is it’s kind of predictable; and things just go through this traditional life cycle and you can expect it. I think the advantage I’m bringing in is how do we predict those and respond prior to them coming up so they don’t bubble up to be crises.
We’re taking more of a horizontal view of things, so it’s maybe a mile wide and a mile deep — instead of it being just a narrow focus on a specific set of cancers, we’re kind of looking at it institutionally.
When you force a lot of structured data, it impacts the workflow. So we have a delicate balance between how do we get the information we need but not impact the workflow and the ability for physicians to do their job and take care of patients? It’s a challenge.
We could apply some of these Watson technologies to ingest this and do analytic work and do some comparisons and do an automated screening. So while we’re trying to teach it to think like an oncologist, could we also teach it to think like an HR person?
Gamble: You touched a little bit on your experience at Ochsner, where you led a pretty significant Epic rollout. So my first question is, did you have any kind of reservations about okay, I’m jumping back into this huge project at another organization?
Belmont: Yeah. It happened in the interview. It happened when I announced my resignation at Ochsner, and it happened when I had my conversation with my family. They said, ‘Dad, why do you want to do this again?’ I don’t know. It was so rewarding. To do what we did at Ochsner post-Katrina and turn around and have the opportunity to do it again here at MD Anderson, on a broader scale and for a specific area, really was exciting for me, and it was frankly humbling to have the opportunity.
But then really what I tell people is that it’s like when you go to a Mexican restaurant and the waiter puts down the plate and says, ‘don’t touch it; it’s hot.’ And what do you do? The first thing you do is grab it. So I just tell people that’s why I’m doing it. Something’s wrong with me. But it is fun to rally a team around a cause and drive some transformation.
Gamble: I imagine there were a couple of lessons learned from having that fresh in your mind that you could apply.
Belmont: Yeah. At one of the first meetings I had here at MD Anderson, they were asking me questions and I made a comment like, ‘now that I’m here at MD Anderson, I can get it right.’ And the guy who hired me said, ‘I wish you would have brought that up in the interview that you got it wrong.’ But I don’t mean it that way; like you said, there are a lot of lessons learned.
On the other hand, every institution is unique. The pace and the reasons why we were doing it at Ochsner were significantly different than what we’re doing in here. So you have to adapt to that. But yes, there are still a lot of lesson learned. It’s amazing, I can see things coming and people are starting to realize it. The other day, I said, ‘Right about now, people will be having a lot of anxiety about what historical data we’ll put into Epic.’ Sure enough, a couple of days later we started to getting emails from people about what are we rolling over and what’s going to happen to my old data. So they think I’m this visionary, but the reality is it’s kind of predictable; and things just go through this traditional life cycle and you can expect it. I think the advantage I’m bringing in is how do we predict those and respond prior to them coming up so they don’t bubble up to be crises.
Gamble: Right. It’s probably not that different from parenthood. You’ve been through some of the toughest years with the first kid. And maybe you’re crazy for doing it again, but a lot of people do.
Belmont: That’s a great comparison because it kind of is. Your second child doesn’t get the same attention as your first. When they cry, you don’t respond as fast. It’s the same thing here.
Gamble: One of the interesting things about MD Anderson is just the sheer amount of data you’re dealing with. One thing I wanted to ask you before getting into that strategy a little bit was about something called the Moon Shots Program. What does that entail?
Belmont: Our president announced the program specifically focused on eight cancers where we’re going to attack and prevent and make a significant impact on those cancers. A very aggressive, a very focused approach — it’s funded specifically as Moon Shots Program. So we’ve announced those. We have a lot of initiatives chasing those, but the reality of it is there’s a lot of data that’s necessary. And what they’ve put a spotlight on was the fact that a lot of the data that we had or a lot of the data that we are going to use in leverage for the Moon Shots Program is residing in different places.
So the first thing is how do we ingest this and leverage it across the board, and then how do we drive the insights out of it? What is happening in one cancer that might apply to another? We’re taking more of a horizontal view of things, so it’s maybe a mile wide and a mile deep — instead of it being just a narrow focus on a specific set of cancers, we’re kind of looking at it institutionally. And we’re having a major impact. I think in the leukemia area in particular, we’re really having an impact on the outcomes, and the insights from the data that’s being generated is significant.
We talked about Epic a minute ago, and it’s not unrelated to this Moon Shots Program. One of the things that I realized that I’m driving for here is that many of these initiatives in the past, from an IS perspective, were kind of running parallel. They were parallel streams. But to me, it’s one information system for MD Anderson — Epic is a piece and the Moon Shots initiatives are a piece and our big data is a piece. But it should be all interrelated, and we could make the information available. As Epic comes live it’ll become a feeder for Moon Shots, so it’ll help us identify patients that might be eligible for certain clinical trials. And then the insights that we gain from the Moon Shots Program — we can drive those back in the Epic point of care so that the knowledge we gain in research can get into the hands of the clinicians much quicker. One of the statistics I heard was that the knowledge we have today will take 10 years for it to become common knowledge in the general medical industry. How do we take that knowledge and drive it in there?
We also have a couple of other initiatives affiliated with it. One is to democratize health care. Expecting patients to come to Houston to get their care is not possible for everyone, not to mention we don’t have the capacity, and there’s really not a need. So how do we take the knowledge we have and democratize it and get it out to the world? So we are aggressively moving on. We’re wrapping up our first year of taking the insights and more or less allowing others to leverage the knowledge we have.
And then we’re using IBM Watson too; that started about two years ago, actually just a little before I got here. We’re trying to train Watson to think more like an oncologist. How can it ingest and look at all of the information that’s out there — not only within MD Anderson, but all the journals and all the other literature that’s been created — and drive specific insights into patients. We’re seeing great results with that. Training is proving to be a little more difficult than we thought, but any training takes a while. So we’ve made major strides. We just did our last annual report and the progress is just amazing and significant, and we’ll accelerate our development in that area quite a bit in the coming years.
Gamble: There is some really interesting work being done and I can imagine that that has to be such a huge priority — just harnessing these data and being able to leverage it so that it’s not taking years and years for these practices to go to the bedside.
Belmont: Agreed. The big challenge we have, again like many other big data initiatives, is that our data is not always structured. So a lot of the insights we gain will come from unstructured notes that have been collected for years — how do we get that knowledge in a format that can be leveraged by the tools we have. And then how do we not build Epic in a similar fashion so that Epic is allowing more structured information? And then balancing that with the fact that sometimes when you force a lot of structured data, it impacts the workflow. So we have a delicate balance between how do we get the information we need but not impact the workflow and the ability for physicians to do their job and take care of patients? It’s a challenge, and that’s part of the change enablement piece in that helping them understand that a little bit of extra work here is going to benefit us downstream.
Gamble: Right. And then in terms of analytics, what are some of the other things that you’re doing?
Belmont: Part of our big data strategy is that while we’re focused on Moon Shots initially, because that’s our primary focus, we’re also looking at all data. Big data is not just clinical data and research data. It’s all data. I think there are insights we can gain from things like patient behaviors and patient demographics, and I’ll get into that in a minute. And then also all the information that’s been generated with PeopleSoft can help us with the operational and financial performance, and so it’s integrating supply chain with clinical and with research in getting the information; preventing our user from having to go on a scavenger hunt and building correlations and relationships between data that otherwise you wouldn’t think about.
It’s interesting because we did a little bit of that at Ochsner — and they’re doing much more. There are some gold nuggets of knowledge that pop out when you bolt things together. One example was when we actually put our phone switch data up against our scheduling data — it sounds very simple but it’s not. And many people don’t think about it this way, but we could look at call volumes down to the agent and then look at appointments being booked. We realized we have a lot of opportunity to optimize our business, and it drove not only better capacity management and better financial performance, but patient satisfaction went up. So looking at the data that way is critical. I think looking at all data is where we’re going to go. But that’s going to take a lot of time, and actually, it never ends, because there’s always new pieces of data out there.
Gamble: That’s a really good point. It doesn’t always have to be the clinical data; you can take on these other pieces and use them to improve things. So I guess that kind of speaks to your overall goals of process improvement and just changing the way some of the business is done.
Belmont: I think looking at the institution instead of looking at point solutions is going to be critical for IS. We have so many tools — so many powerful tools, and like I said earlier, we have a lot of duplicates. But then how do we take advantage of that? The way Watson works, for example, is it takes all of these data and ingest it. We get 40,000 applicants a year — I’m not sure if that’s the right number, but it’s what I remember. So can you imagine how many resumes and profiles and non-structured information is flowing into the organization all the time? And the traditional way of having a human resources screener look at all these and hopefully get the right candidate, make it through the gauntlet, and have it get to the person they need — we could apply some of these Watson technologies to more or less ingest this and do analytic work and do some comparisons and do an automated screening. So while we’re trying to teach it to think like an oncologist, could we also teach it to think like an HR person? If we wanted a Portuguese-speaking critical care nurse, could we find that very quickly or could we predict that we need a Portuguese-speaking nurse because that’s something where when you correlate that data with other patient performance, that might be a competency we’re looking for.
So again, how do we take the tools we have and use them across the whole end solution and not just keep them boxed up in one area? It’s kind of interesting and kind of exciting. People will say, ‘Is that possible?’ Absolutely. I mean, people didn’t think that Watson could think like an oncologist and we’re showing that it happened, so why can’t we just do that?
Gamble: It’s really interesting. It makes me think about when people talk about innovation — it doesn’t have to be this grand new invention, but more like different ways to do things that maybe use less resources or just do things in a better way.
Belmont: I agree. It’s not Watson, but we’re using the cognitive capacity of different platforms. It’s also improved the patient experience. We just did a quick trial. We took a cohort of about 80 or so patients and ran it through a process and we gave them to one of our cognitive learning partners and we just gave them some basic information about these 80 patients. They came back less than 60 days later and they said, here’s what we know about these patients. Here’s a guy who is coming in. He’s on this treatment (and we got permission, by the way, from these patients). He’s coming in for these treatments. He’s coming in at these frequencies. He likes basketball. The next appointment he has, the Houston Rockets are in town. Here’s another guy who likes sports cars and we noticed close by you can rent a Ferrari, so we can offer this up for him. Here’s another family who obviously likes Chinese food.
They went out to social media and they would and they just found insights about these patients. So how can we make the MD Anderson experience better — not just clinically, but from a patient experience? Because when people engage with us, they’re going to engage with us for a long time, so how can we help them through this process? Not just being here for them as an institution, but how can we make their experience better overall? Those are some of the things we’re playing with and looking at and trying to decide where they fit strategically with us. But our patient experience is really important to us. We pride ourselves in what we do, but I think we can always do a little bit better.
Share Your Thoughts
You must be logged in to post a comment.