Healthy Data Podcast Robin Damschroder (HFHS) & Jordan Cooper (InterSystems)
April 17, 2026, 4:19PM
22m 26s
Jordan Cooper 0:03
the EVP and CFO and President of Value-Based Enterprise at Henry Ford Health System. Robin, thank you for joining me today.
Robin 0:10
It's a pleasure to be here.
Jordan Cooper 0:12
So for background, Henry Ford Health System is a 4,600 bed system headquartered in Detroit with 13 hospitals and 10,000 providers. And Robin sits at the intersection of finance strategy and care transformation. Robin, what I'd like to address today is how Henry Ford Health System is using data integration and AI to make healthcare more affordable and accessible. SoI think we're going to start off with your data analytics strategy. I understand that leveraging predictive analytics to drive value-based care model is a big goal for Henry Ford Health System. What are some of the use cases you're tackling first, and what data infrastructure had to be in place before those became possible?
Click to read the full transcript
Robin 0:55
Well, I would say in general for our data strategy that we're thinking about it for care delivery operations, the value-based enterprise, the insurance enterprise, you know, our home care, every facet of what we do. So the very basic things and foundational pieces we've been working on for two or three years has really been gettingthe infamous data lake in place and the ability to make data with permission sort of ubiquitous so that you can have the standards and governance over definitions so that you can get onto those big natural language questions that we all envision having Jarvis, right, and just being able to ask questions. But until then, you know, it'sIt still remains a little more complicated, but with all the amazing tools that are coming out, whether it's for your data analyst, it's for your quality analyst, your financial analyst, the folks that we have working in the value-based enterprise, really looking at the cost and trying to drive affordability.How do we make that easy and accessible? So we actually spent quite a bit of time where we had fragmentation in the value-based enterprise beyond just the data, but the tools that we were using, and really trying to line up how we talk about things. And so again, back to those definitions and standards and howhow reports or dashboards get done. So that we're working together, we're communicating together. And, you know, Jordan, the data being available, you know, we talked about it's only good as it's curated for, right? So if you've got junk going in, it's junk coming out.And when you're aggregating particularly claims data in the universe that you are doing for your clinically integrated networks, or you're doing for population healthcare management, or on the insurance side, getting that right and filling in the holes so that you kind of have the whole picture becomes important.
Jordan Cooper 2:46
Mhm.
Robin 3:05
So we actually took on a project that helped us, for the lack of a better word, build our garden for our claims data, and then also a methodology for us to tend to the garden. So because new data comes in every day, right? And so those become really important things.
Jordan Cooper 3:22
Mhm.
Robin 3:25
features. And I was just at Becker's and we were talking about how important your commitment around the base fundamentals are. I think there's plenty of opportunity when you start talking about AI to help in the whole domain and continuum of healthcare. And we put our governance in place around that.
Jordan Cooper 3:45
So...
Robin 3:45
So happy to talk about that, but I'm going to, it's your podcast, your questions.
Jordan Cooper 3:50
So Robin, you mentioned a number of incredible topics that we're going to try to get to today. You mentioned Gen AI, and we're going to get to that in a moment. You also mentioned your plan. So for our listeners who don't know, Henry Ford Health System owns a Health Alliance Plan referred to as HAP. I think you mentioned previously that 50% of Health Alliance Plan membersare actually Henry Ford Health System patients. And so you're sitting on a uniquely rich longitudinal data set spanning both clinical and claims data. You mentioned that you're building a garden for your claims data. How are you managing and integrating those two data streams today?
Robin 4:15
Yeah.Yeah, so to be very clear, the data streams that we're integrating today, because you've got to have the right prior wells, right, we can only collectively be looking at the data of that 50%. But we actually, we have put that data together. We both used to have sets of that data.
Jordan Cooper 4:38
Mhm.
Robin 4:47
But when we're speaking and talking with each other about quality, about our HCCs or our stars, or looking at the cost and affordability of maybe it's cancer care, maybe it's MSK for a commercial population, maybe we're looking at the chronic CHF or COPD patients in Medicare,that we're actually looking at the same data together so that we can drill down and have a very detailed conversation. Because more importantly, for us to get to affordability, we have to have executable strategies. And when you talk about something like Medicare Advantage, you're really looking and focusing on readmissions, right? AndHow can we provide information back into the delivery system so that we can effectuate that? Or better yet, what kind of signals can we send into our care managers to keep our chronic patients who we know are at high risk out of the hospital? So we really try to develop those signals together. And at the same time, we want signals coming back from the delivery system. So for example,any patient that leaves Henry Ford and is discharged gets a text on their phone. And then they can respond to a few questions. And that helps us gauge where they are in our risk, whether they're moving up in risk, in rising risk, or they're coming down and whether or not they need a phone call. And again, we might have done everything right on the discharge.but somehow the patient's feeling like they're at risk. And so sometimes it can be a very easy just reassuring them, yes, you have your doctor's appointment, you're ready to go. Other times they're like, well, I didn't get those meds when I left and I didn't get this and I didn't get that. We can fill in the gaps so we don't wind up with a readmission before that seven day appointment.So again, the execution becomes a very important part of what we're trying to get done with the data.
Jordan Cooper 6:35
Oh.So I love that signal that you have with the text message post discharge. I imagine that's one of your highest profile executable strategies. Can you walk us through how you were able to execute that particular project? Did you have to, I mean, from ideation to implementation to integrate the text that pushes from the EHR?and then incorporating that data back so that you have a care manager follow up with the call and then tying all of that, those actions into the quality measures to make sure you don't have that 30 days readmission.
Robin 7:16
Yeah, well, okay, you get a lot of detail you want to ask there. So I'm going to give you the high level version because I am sitting at the top of the organization. They don't actually let me go and play with the mechanics. But you know, sometimes you have a really good idea. We were using a tool called Cipher and we were using it for patient experience.
Jordan Cooper 7:20
I...Ann.Mhm.Mhm.
Robin 7:35
And because we were using it for patient experience and asking, we ask patients while they're in the hospital questions, and you know, you have to do your CMS surveys, and we use that tool for that. It's something that patients were already familiar with. We didn't want to add another tool or another layer. We wanted it to feel consumer friendly.
Jordan Cooper 7:41
Mhm.Mhm.
Robin 7:58
So we chose to extend that tool. So as they were completing their survey and getting other patient education information, it's that tool that's actually sending out the text message, not the EMR. Then it comes back into that tool. And that data then is we use Compass Rose for our care manager.then our care managers are alerted in Compass Rose that there's a call that they need to make and it's within their queue and in their work list.
Jordan Cooper 8:20
Yeah.So on this topic, but also on the related Gen AI topic, you mentioned that with Gen AI and automation, they're most valuable when they're built atop clean data. And obviously that would also be true with this text message and integration with Compass, Rose and Cypher. Where would you honestly rate Henry Ford Health System on data readiness? And what's the biggest bottleneck
Robin 8:40
Good.
Jordan Cooper 8:51
between where you are today and where you need to be.
Robin 8:53
You know, I think you have to think about the pools of data that you have. And I think that we would probably rank different parts of our system higher or lower in that. Like an example I'll give you are community care services. We have in some parts of our pharmacy,
Jordan Cooper 8:58
Mhm.Mhm.
Robin 9:12
You know, a system that is sort of separated from the EMR, which does the inpatient and outpatient. We have retail pharmacy on a different system. We have plans to bring that in. You know, when you build those bridges between those different systems, that's where you may have MRNs that aren't linked yet, or they have to go through a process to be linked, right? So any time that you can line up
Jordan Cooper 9:20
Mhm.Mm-hmm.
Robin 9:35
Your process in a similar system, ensuring you're using the same MRNs, you know, doing the basic level things to validate that you've got the same person, because that matters on the value-based care side, just like it does on the care delivery side. You got to know you got the right person. So I think it's those data connections where we would rateYou know, we've got some noise in that data, but there are like this com spec, we call it com spec, and where we have all of our claims data. We probably rate that at a higher degree of satisfaction than where we were probably four or five years ago.
Jordan Cooper 10:00
Mm-hmm.So with those MRNs, how is Henry Ford Health System doing enterprise master person identity management in order to reconcile data sources outside of the epic source of truth?
Robin 10:23
Yeah, so you're asking someone who doesn't have that level of knowledge to be fair. And we are blessed here at Henry Ford that we have one instance of Epic. And with the exception of we're 45 days away from taking 13 separate EMRs and putting them on Epic from our Ascension joint venture.
Jordan Cooper 10:27
Kat.Yeah.
Robin 10:44
We will all be on one instance of Epic. So that's a very, it's a much simpler process inside of that window than it is today in those other 13 systems.
Jordan Cooper 10:47
So, yeah, let's...Let's dive into that. So the future of health at Henry Ford is a $2.2 billion investment centered on the Destination Grand project in Detroit. It's redefining the academic medical campus and follows on the heels of a joint venture with Ascension Michigan, which was finalized October 2024, where you transitioned 8 Ascension Michigan hospital campuses and addiction treatment centerto the health system. So I'd like to ask about the greatest challenges in getting new data sources to conform to your standards quickly. And some, what does the data exchange look like with these external partners? How are you consolidating them down to one Epic instance? And where does data exchange break down?
Robin 11:34
Yeah, so there's a lot of data separation. And again, I can't speak to it all because it belongs to my Ascension colleagues. But, you know, again, we have 4 main EMRs. There's A variety of other small EMRs for the addiction hospital. You know, you have some other service lines, some other physician groups that have smaller
Jordan Cooper 11:41
Mhm.Mhm.
Robin 11:57
Smaller EMRs. And so it does break down in that sense that you have all these subsystems. And again, the ecosystem within the former Ascension had actually built a bridge to have a data warehouse to have that all come together. And it's fairly sophisticated. So when, as we lookforward, we've had to convert for whatever data that we're taking and link it to our standards. We've done several other things and tools, and I'm going to get you to talk to my CIO if you ask me about the tools. To use those tools for the data that we're going to house and use as readable so that we can look back in a
Jordan Cooper 12:22Y
eah.Ha ha.
Robin 12:39
In a period of time that we appropriately can have access to to do that. We have other things like we have a relationship with LabCorp, right? We have to share data back and forth with them. And so you work hard to agree on the standards. You may not have the same systems, but you've got to use some sort of technology to make the bridge go back and forth.Between Epic and and the other parties.
Jordan Cooper 13:04
So you mentioned two things just now. Well, you mentioned the data lake, I think, and the data warehouse. How have you been managing a data lake and data warehouse within the context of these acquisitions and EHR consolidations? Have there been data challenges associated with the two?
Robin 13:12
Mhm.Yeah, I'm sure my team would tell you there's all sorts of data challenges. I probably see it a little more simply. I mean, we sort of started fresh and anew. We had our own data warehouses, right, and you had the Ascension ones. Probably two to three years ago, we started really looking at the data lake and taking the time and preparingto cleanse data and put it in canisters. So if you think in that lake, there's a canister for the insurance company and there's one for the care delivery system. I mean, there will be one where we're storing historical data that we're allowed to keep from Ascension. And in some of those things, again, you have to have the master indices that you're going to link together to use that historical data.
Jordan Cooper 14:05
Mhm.
Robin 14:14
Data, but a lot of, because our Epic instance is unique to us and we have standards built in there, a lot of our history on data is going to be on the go forward, right? When we start being able to compare quality and the harmonization of that data, a lot of it.the vast majority of it comes from coming onto the same platform using the same workflow, the same definitions, etc.
Jordan Cooper 14:39
So when you reconcile that data in your data lake or with the different data warehouses, you consolidate the data. So then you have a robust, rich set of clinical and claims data available in the health system. But a lot of health systems talk about there being a gap between that data, which exists in the health system,and operationalizing it, making it useful to the providers at the point of care, at the time of care, the right time, the right place, the right person. How has Henry Ford Health System worked to enable, to empower clinicians to have, to be able to have actionable insights from that data at the time of care?
Robin 15:05
You.Yeah, and I'm just going to go across the health system because there's lots of people who use the data for actions and a lot of the actions with inside the clinical care workflows is driven by Epic, right, and your ability to use Slicer Dicer and a lot of the tools that they have there. We've also built
Jordan Cooper 15:30
Mm.
Robin 15:38
You know, a lot of dashboards or workflows or BPAs within Epic to help alert the positions. Obviously, there's a lot of data, there's a lot of, you know, add-on tools. There's, we are an Epic, we are an Epic first house here, like most people know that about us.
Jordan Cooper 15:44
Mm-hmm.Mhm.
Robin 15:57
But we have built things that allow us to assist the assist the physician. Obviously, we've incorporated ambient listening, which helps take the burden off the physician as they're meeting with patients and allows them to help them get through their notes more quickly. We're super excited about the new tools that Epic has sent out, you know, Emmy and Art and Penny, which are going to helpacross the entirety of that process of visits set up to when you're in the room and what does the patient want to talk about. Those tools are going to be meaningful for us.
Jordan Cooper 16:30
Sure.So Robin, you just mentioned ambient listening. Back to the Gen AI category. I know that you have a three-tier governance model for Gen AI use cases. I wonder, using ambient listening as an example to talk through, what happens at the data layer when a new use case gets approved, like ambient listening? How quickly can your team get access to and prepare the data
Robin 16:40
Mhm.
Jordan Cooper 16:54
That it means.
Robin 16:56
So when the team decides, you know, we divide the AI information into sort of three layers, like when you're working with an Epic or a Microsoft versus maybe when you're working with the next tier, it's really kind of a robust application platform.
Jordan Cooper 17:02
Mhm.Mhm.
Robin 17:15
And sometimes the Microsoft and Epic buy the things in the previous layer. So we go through a process to kind of vet the use case to understand whether or not it's actually going to be efficacious. The nice thing about some of these AI tools, it doesn't take as long as it used to.And particularly if you're doing something in Microsoft or Epic, they have these playground production areas that you can turn it on in the background and compare it to the process that's going there. And you can decide anywhere from 10 to 60 days whether you're happy with it and whether or not the clinicians who are watching the usual process and that process, do they trust it?So even if you turn some of these things on, you don't like trust it instantly. Many times you're running them side by side to learn about how the process needs to change. Wow, that didn't quite come out how we thought. Why does that happen? Maybe you find in the way you structured the data that you left something out or wow, we got to start collecting that field. We were never collecting that field before.
Jordan Cooper 17:58
Mhm.
Robin 18:20
So there's learning that comes with implementing. And I think what we have found as well, it's not just the technology and the data, it's also the people and the process. We've also found that maybe the workflow process has to change. Maybe you have to retrain your employees because, oh, I don't have to do this part.I can trust them to do it. Why can I trust them to do that? And then how is it that, you know, they get to go up to that next step and then really execute on the piece that they need to do?
Jordan Cooper 18:50
So Robin, as we approach the end of this podcast episode, I have a final question for you. And perhaps unlike the previous questions, this one's going to be definitely within the wheelhouse of the CFO. If you could solve one data problem in the next year, year and a half, that would have the biggest downstream impact on both costs and outcomes and lead to the greatest ROI for the organization,
Robin 18:57
Yeah.Okay.
Jordan Cooper 19:12
What would that data problem be?
Robin 19:14
I think solving the data gaps would be it, because, you know, on any given data analysis you get, you know, if it's 90% of the way there, 95% of the way there, like when you're in the clinical realm, people are going to make decisions. They want a really high.
Jordan Cooper 19:33
Mhm.
Robin 19:36
Really, really high. It's like thinking about epic downtime. We would never accept epic downtime not being 99.99999. And I think in the clinical realm, we're going to go through a little bit of this process where it's 80 or it's 85. Well, okay, I'm going to run it side by side. And then we're going to learn about how to
Jordan Cooper 19:41
Mhm.Mhm.Mhm.
Robin 20:19
Lots of data in a decision support system and we bring that out and we would never say on first pass, it's done. We sit down and we start looking at it and like, then we go back in the detail, we find the gaps, we manually fill in those gaps. And the question is, how will these tools help us identify the gaps, fill in the gaps, so that we can againget that efficacy far higher when you start talking about clinical decisions. For what something when you have data gaps was much more acceptable when you're talking about, you know, things that don't impact people's lives, right? You're doing a financial analysis. You can do that again in another time. But we too on the finance side are looking for efficiency too. I love it that my analysts don't have to go back.
Jordan Cooper 20:55
Mhm.
Robin 21:02
So which means it's process, right? We're going to have people focused on, wow, you found a data gap. It's way back here at the beginning of the process when the patient walked in the door that we got to fill in that gap. So just imagine that curating that garden is really about us getting much more efficacious and accurate in capturing the data in the process.And you know, that's what other amazing, amazing tools like ambient listening, things that are going to help us get down the human error, right, that comes with data collection.
Jordan Cooper 21:26
Mhm.Well, Robin, we've covered a lot of ground today. We've spoken about your data analytics strategy and leveraging predictive analytics at their data lake. We've spoken about interoperability and Destination Grand, your joint venture with Accenture Michigan, and how you've had to consolidate clinical and claims data. We've spoken about longitudinal data sets, both spanning both clinical and claims data withhealth alliance plan and having that firewall between the payer and the provider organization. And we've spoken about Gen AI maturity and kind of scaling and introducing new solutions into the workflow. I appreciate you joining us today. For our listeners, this has been Robin Damschroeder, the EVP and CFOand president of Value-Based Enterprise at Henry Ford Health System. Robin, thank you for joining us.
Robin 22:22
It was my pleasure, Jordan. Have a good day.