Shifting from Projects to Products: How NYC H+H is Using Data to Improve Patient Care

Shahran Haider [00:00:00]:

There are only three things that happen naturally in organization organizations. It’s chaos, confusion and underperformance. And everything else takes leadership. And you know, that’s so powerful because if you think about it like leadership is about change, about making new things happen, supporting the organization, helping them get there. It doesn’t happen without leadership.

Narrator [00:00:25]:

From Healthcare IT Leaders, you’re listening to Leader to Leader with Ben Hilmes. Our guest today is Shahran Haider, Deputy Chief Data Officer for NYC Health + Hospitals. In our conversation, Shahran discusses the importance of data in improving patient care and the shift from project to product mindset in healthcare analytics.

Ben Hilmes [00:00:44]:

Shahran, thanks for joining the podcast. This is going to be a lot of fun. We got a lot of ground to cover, so let’s get going. I was thinking about this and trying to put myself in your shoes and it dawned on me that you kind of wake up every day at the intersection of data and the largest municipal health system in the United States. So New York Health + Hospitals. And I quickly thought that’s got to be one of the most exciting yet challenging jobs in the industry. Tell me a little bit, I tell our audience a little bit more about your organization. I’d love to hear if you can share some perspectives on just the broad aspects of data you guys are collecting and in particular, what problems are you trying to solve.

Shahran Haider [00:01:28]:Yeah, absolutely. Thanks for the question, Ben. So yeah, I think it’s always helpful to know the context of the work that we do for the organization that we do it for. So like you said, New York City Health + Hospitals is the biggest public health system in the U.S. we, we have 11 acute care hospitals, five post acute care long term care SNF facilities and 70 plus primary care delivery locations across the five boroughs of New York. So that’s massive scale in terms of the presence and in terms of the members and the patients we serve, it’s almost a million and a half patients every single year. So that’s the context of the organizational breadth and depth if you will. But the most important part is actually the people that we’re caring for and what our mission is, which is care for New York City, New Yorkers without exception.

Shahran Haider [00:02:24]:

So regardless of race, religion, ethnicity, background, ability to pay, you know, that is really core to the work that we do and it’s a great unifying mission and makes the work very rewarding and fulseling. Now that being said for from a data analytics standpoint, you ask like what are the problems that we’re trying to solve? And if I had to boil it down, it would really be around how do we make care better for the patients that we’re serving? And as you can imagine, it’s a complex population with a multitude of different needs, based on chronic conditions, based on ability to pay, based on various ethnicities and backgrounds. So the thing to remember about healthcare is that it really varies based on the specific complexities around the what make you up as an individual. So it’s not enough to look at a person from their race and say, this is the type of treatment that makes sense to everyone in this particular category. Obviously you have to consider a bunch of different factors. That’s the background in terms of the problems we’re trying to solve. It’s really about the patients. How do we make care better for them? Across all the different functions and domains that are built around this healthcare system, there’s a lot of different domains around how payments are done, how the services are provided, and so on and so forth.

Shahran Haider [00:03:41]:

But bottom line is, how do we make care better for our patients?

Ben Hilmes [00:03:45]:

I think having the mission focused out in front, you know, helps you kind of stay focused and frame up how you work every day. And I heard you talk recently about this concept called the product mindset shift, which was really interesting. I like simplifying things like that. So when it comes to your overall operating model, can you explain what that means, how you use it, how you apply that to your focus?

Shahran Haider [00:04:15]:

Yeah, absolutely. So, yeah, I also like to simplify things if possible. And I think it’s kind of a survival instinct, if you will. You have to have it when you’re trying to deal with this scale of information and technology complexity and things like that. But if you can imagine this for a second, like an organization of our scale, and there are other organizations, of course, of our scale and bigger and so on, everyone’s trying to make decisions with data, and everyone has a little bit of a different priority in terms of what they need to use that information for. Some of it is for operations, for pure clinical operations, some of it is for finance, some of it is for back office management, IT systems and things like that. And if everyone in the organization had one question every single day, that would be like 50,000 questions a day, right? Just based on the number of employees that we have. And as a data organization, how do you then support being able to deliver that need? Because everyone has questions that they’re asking for some reason because they want to run their operations better, to serve the patient.

Shahran Haider [00:05:26]:

It’s all connected to the patient at the end of the day. But then you have to think about prioritization. So how do you prioritize? The prioritization has to be in terms of value. Where does the value come from? The value comes from the strategic priorities of the enterprise. And it takes time to think about what that is, identify, align from an organization standpoint and say these are the big rocks, these are the blue chips that are really important to the organization. And in terms of delivering data, we want to create products and not projects. And the difference there is a product is something that is meeting a deep business need, provides customer value and is ongoing. It can be sustained in terms of delivering value.

Shahran Haider [00:06:14]:

And if it doesn’t sustain over time, then at some point you shut it down. But basically it’s not a project. A project mindset is very start and stop. Like you have a start date and you have a stop date and you get it done. And at the end of the day you’re like the work is done. But with products the mindset is really different. It’s focused on value, it’s focused on customer engagement because at the end of the day you’re building it for a customer. A set of stakeholders that are going to use the solution for something to improve operations, improve care, etc.

Shahran Haider [00:06:43]:

And then you also have to make sure that you have a focused, dedicated team that is trying to work on the right priorities instead of trying to do 10,000 things at the same time, which can happen if you let yourself overwhelmed by just requests, instead of priorities that are focused on value. So that’s the product mindset, really customer value oriented.

Ben Hilmes [00:07:06]:

That’s fascinating. I came from the product world at Cerner Corporation and you know, a lot of that is projects are, you know, they have a start, they have a stop, but a product is iterative and so there’s always this ongoing mindset around we’re going to get it to a point, then we’re going to take it to another point, then we’re going to take it to another point. And I think that’s the journey people are on in particular around data, you know, is the constant iteration of what, what you’re gathering, how you’re using it, what problems you’re solving, et cetera. So I love that. I think that’ll be fantastic. We, we get asked a lot to where do I start? In particular around the concept of AI. And we say look, I, I think you need to really start with organizing, you know, a cross functional team in order to get, you know, kind of an end to end perspective on, on how the org should be thinking about AI. You guys, you’re ahead of the game.

Ben Hilmes [00:08:04]:

You’ve already stood up your AI advisory board and you, you kind of have three key focus areas. I think it’d be really insightful for the listeners to understand that framework. How did it come about and then the three areas of focus.

Shahran Haider [00:08:19]:

I think we’re making progress. I don’t know if we’re further ahead than other, other organizations, but part of it is by design we’re a public health system. We have to be naturally learned, more cautious and more thoughtful about where we spend the funding in order to do these things. What we’ve done so far is we have had lots and lots of conversations about what are the strategic priorities of the organization. And they are things around improving clinical efficiency, clinical care delivery efficiency. So the things that basically create a lot of friction in the doctor to patient interaction in terms of basically capturing information about the patient care, about communicating that, the next set of instructions to patients and so on. Doctors, you know, as you can imagine, are overwhelmed with the amount of work that they’re faced with and the information is complicated. So how do we make it easier for them? Meanwhile, there’s also a lot of other supporting functions across the enterprise in terms of revenue cycle, managed care policy, policies, IT back office operations.

Shahran Haider [00:09:22]:

Even in the data organization, there’s productivity gains that we can have across the board by leveraging AI. And so how do we create, you know, these access to these platforms where we can safely and easily interact with sort of an advisor, right? Like I think of ChatGPT and other large language models as an advisor. Like if we’re trying to learn information about something or if we want to synthesize and summarize information really quickly, or even builds programs and code really quickly, that allows us to do it, not get it right necessarily, but you can accelerate that process so much more by having something like that at your fingertips. So the priorities of the organization really are around having safe, responsible AI. Whatever we do, especially when it comes to clinical care, we have to be really thoughtful and clear about the implications for the patient. If it is some sort of patient facing AI, then we have to have real world translation, translation and implementation evaluation frameworks, which means like we can just take a model and expect that it’s going to work a certain way. Safety, ethics, bias, responsibility is really, really, really important. It’s up there in terms of overall, we think about it in terms of governance.

Shahran Haider [00:10:39]:

The other aspect I mentioned at the beginning, which is really in terms of return on investment. So we have to make sure that whatever we’re doing that, at the end of the day, there’s going to be some benefit associated with it. Comes back to the priority, again, where are the priorities of the organization? And the third thing that I would say is really around making sure that we continue to do this in a scalable, repeatable way and create educational literacy across the organization so that anything that we do, it can be used and then reused and then others can learn from it. Then we focus on the part about adoption because it’s not enough to roll out any technology solution, any implementation. Doesn’t matter if it’s AI, analytics, databases, application, whatever. At the end of the day, if people don’t trust the solution or understand the solution, it’s not going to provide the value. So that is the third area of focus.

Ben Hilmes [00:11:33]:

So I’m thinking about this and the scalability point comes to mind because I think that’s where a lot of the organizations we work with get stuck. We’ve seen really good specific use cases. You know, think radiology lab, et cetera. There’s some specific use cases where AI has provided some real tangible value. Where I think a lot of our organizations are struggling is when they try to go to scale and then they recognize that they’ve got, you know, data quality issues. Now we got interop issues. There’s a whole plethora of other things that have to be considered. And then one of those is for sure, adoption.

Ben Hilmes [00:12:13]:

And you know, and then I think about the variance between kind of the payer world and the provider world. The margin structures are a little different. So the whole ROI conversation becomes kind of paramount is you can’t just run into it. You got to be really thoughtful and aim a lot, think a lot. Pulling these boards together to stay focused on the core, you know, key components that your organization is focused on. I, I think is key. So sounds like you’re way ahead of the game there. And you may not think you are, but you are.

Ben Hilmes [00:1245]:

I can tell you just in the number of interactions, you guys are way down the path. I’d love to talk a little bit about. So you guys are a big Snowflake client and their AI Data Cloud. And I recently read your team was able to use Snowflake to reduce membership and claims data delivery time from five days to five minutes. That’s it. Talk about that. I mean, that’s an incredible outcome. Take me through from kind of identification of problem to what you did about it and how you achieved the result.

Shahran Haider [00:13:20]:

I think this is a really good example of being clear on the enterprise priority and the need. And what is the reason that we should do this in the first place? Because, as you can imagine, that five days to five minutes didn’t happen overnight. And it didn’t happen, you know, very easily, you know, with some small shifts in, you know, process or technology. It took a lot of effort, it took a lot of design and, you know, hard work and so on. But the reason that we stood up that initiative in the first place was to serve a initiative called Special Populations. And the Special Populations initiative was to provide holistic care in terms of not just medical, but also financial well being, transportation, food, housing, assistance for people experiencing homelessness in New York City. And so that was the impetus for the program. And the reason that the data part became really important was there’s a lot of different agencies that need to share information.

Shahran Haider [00:14:23]:

When someone shows up at a homeless shelter, as you can imagine, they are there with complex needs. They may not understand the language, they may be experiencing mental stress, they may be handicapped, and they just basically need a lot of support. And there’s not one single agency that does it. All right? So there’s the shelter, there’s Department of Housing, there’s Department of Social Services. New York City Health and Hospitals was tasked with providing managed care services to basically take care of the holistic health of the member. And so as data needed to get exchanged across these agencies, the typical ways that it gets done is file transfer protocol. Emailing stuff here and there creates a lot of friction between data teams because the formats are different. Interoperability is the point that you make, right? So in healthcare, we say we use HL7 and FHIR.

Shahran Haider [00:15:18]:

We assume that it’s gonna work. It doesn’t. Because every organization transforms that in a little bit different way, right? So one epic instance for one client is different from another epic instance, another client, because we adapt the workflows, we change the fields, we change the logic. So the point was, how do we make this process of getting data from the peers just as a start, that’s the piece that we did. But the intention was to share data really easily across these various agencies in an automated way, in a governed and secured way. And it had to be bidirectional. So not only could we take in data in an easy to exchange format from our partners one way into New York City Health and hospitals, but we wanted to do it the other way back and share data. We wanted a mechanism to be able to do that.

Shahran Haider [00:16:06]:

So we built something called the Data Hub. And so there were a lot of different processes for working with different pair data feeds, we’re working with multiple pairs and every pair, as you can imagine, sends it a little bit differently. Even though it’s like all membership data and all claims data. Same thing. Everyone has different logic, different formats, different frequencies, so on and so forth. So what we did was we took that all the process mapping. We had a great reason for wanting to do this, which was at the end of the day, we got to get this to case managers and clinical care people as quickly as possible, manage care teams so they understand what members are coming through that require these services from the homeless population. So we went through that exercise and basically we built a different type of architecture which is cloud-based.

Shahran Haider [00:16:48]:

It was built on Snowflake. It standardizes data, automates the ingestion process, and then distribute automates the data distribution process as well. So doing all of that basically allowed us to automate this manual process and take it to five minutes from what had previously been five days. At its best, it was usually probably like more around seven days, eight days because it was completely person dependent.

Ben Hilmes [00:17:12]:

That’s incredible. It speaks to a passion of yours which is around data democratization and right data, right place, right time. Talk to me about, or talk to our listeners about that passion of yours, how that ties directly to the mission of your organization. But then at the same time it probably creates real challenges around overall data security risk, et cetera. Those are a lot of things to tackle as you try to accomplish the democratization of the data.

Shahran Haider [00:17:43]:

I think that word has been used for a long time, data democratization. And really what it means is, you know, put data in the hands of people that need it to get their job done, whatever that job is, right? But the complexity is how do you get data in one place where people know how to use it, they can answer their questions on their own. Sounds simple, but obviously there’s a lot of complexity, there’s a lot of integration work that needs to happen. So using again, health and hospitals as an example, if every single person in health and hospitals asked a question every single day, 50,000 employees ask the question, every single day, 50,000 questions. How would we answer that? A central team is unable to do that. They just cannot keep up with that volume.

Shahran Haider [00:18:29]:

So you have to create systems that enable that self service. But it’s not enough to build the system. There has to be a lot of organization around that. You have to put a data catalog, which basically is like going to the library and knowing which book you want to read and which shelf to go and how to Navigate there. You have to have literacy so that people in these different parts of the organization have the tools and the skills and the knowledge. There needs to be leadership, alignment, understanding. So it starts with the vision, obviously, but from the vision there’s a lot of what I would call need for persistence and commitment because it doesn’t happen overnight. It was, for us, it was a three to four year journey where we went from being an organization where different teams across the organization were doing different things with similar data and coming up with the different answers to the same question, which creates a lot of issues.

Shahran Haider [00:19:26]:

Obviously, you can imagine you can’t make the right decisions. From there we went to a central team, but then central team becomes a bottleneck. So now we have to have a little bit of both. We have to have central and we have to have enable the rest of the organization. And so we can do it in parts. Like we have 11 different hospitals. How do we allow each one of them at a time? We built up something called the Data Champions program. Give them the tools, the education, the support, ongoing.

Shahran Haider [00:19:51]:

And then basically we also have to monitor and check in from time to time and see what we’re delivering them. Is it really helping? And I think the easiest way to think about this is the results, the metrics that we’re seeing. So for example, when we went live with all of the different, let’s say data champion groups that we have at New York City Health and Hospitals, all of the 11 hospitals and multiple other groups dispersed throughout the enterprise, we have 15 different groups like this. Compared to the central team, which is our data and analytics organization, within the space of one year, these various groups throughout the enterprise are now producing dashboards and reports at a faster rate at a greater number than the central team. And then looking at Snowflake, if we think about it in terms of queries, which are like questions being asked of the data, out of like a million queries that were run last year, 60 to 70% were actually by these decentralized teams outside in the enterprise versus the 30 to 40% in the central team. That is a clear indication of data democratization working and empowering and enabling everyone else to get, you know, get information on their own versus relying on the central team.

Ben Hilmes [00:21:03]:

Are you finding now that the demand is then growing on your team, right? So 60 to 70% is coming from people that want more. I now assume they want more and more data.

Shahran Haider [00:21:14]:

It’s a never ending thirst for information. So absolutely, yeah. So more data, we need more in the enterprise platform, need more support and also need to be able to do more things. So previously we were talking about giving them data and allowing them the ability to create their own dashboards and reports. Now we’re also talking about enabling users that have the right skill sets. If they don’t, then we help train, get them to the right skill set so they can actually work on the, you know, more on the foundation stuff. Which means that all the work around bringing in data, modeling, engineering, getting it to the right structure, we’re also looking at ways to enable teams with the right skill set to be able to do that again. So moving more of the power to them versus the central team, which is therefore the guardrails.

Shahran Haider [00:22:02]:

It’s kind of like the App Store, if you want to think about it like that. The App Store, you’ve got the guardrails, you’ve got the garden, right? And then you enable the teams with the proper skills to deploy stuff on their own. That’s the model we’re going after.

Ben Hilmes [00:22:15]:

It’s data used to live in this one little silo inside of an organization. And really, it needs to be in the hands of people that can actually do something with it. So I love the. I love the approach. We always like to land the podcast with a topic around leadership. You’ve had an incredible career. It’s only expanding. The demands on you, I’m sure, are growing.

Ben Hilmes [00:22:37]:

You know, I’m curious about how you approach leadership. Also very curious about the types of skill sets and mindsets that you’re recruiting. I’m sure your team is growing and changing almost, you know, certainly annually. And so just curious about that whole ecosystem of how you lead and then the types of people you’re. You’re trying to attract to enhance what you’re doing.

Shahran Haider [00:23:03]:

Yeah, thank you. Thank you for the question. Very kind of you to say those things. So, the way that I think about leadership, I think there are really two quotes that I think, you know, exemplifies this really well. At least that’s what I try to think about. And it resonates with me all the time. So one is, you know, a quote from Peter Drucker, and he said there are only three things that happen naturally in organizations. It’s chaos, confusion, and underperformance.

Shahran Haider [00:23:32]:

And everything else takes leadership. And, you know, that’s so powerful because if you think about it like leadership is about change, about making new things happen, supporting the organization, helping them get there, and doesn’t happen without leadership. So that’s one. The other quote that I, that I really treasure is one from Maya Angelou, and she said, people will forget what you said, what you said. People will forget what you did, but they will never forget how you made them feel. And so it’s really important to remember the human aspect of this. It’s not about just getting things done. It’s about respecting the feelings of the people that you’re working with while you’re trying to get it done.

Shahran Haider [00:24:14]:

That’s really important, in my opinion. Now, you also asked about the skills that we’re looking for in the mindset and so on. Technology is changing so fast. I don’t think we know the jobs that will exist like three, four, five years from today, especially with AI, and how it’s going to change responsibilities of everyone in the organization, no matter what function it is. I mean, Goldman Sachs has done analysis on this, McKinsey has done analysis on this, where, you know, 30, 40% of most people’s jobs will completely go away. And there are certain jobs that will almost certainly completely disappear. So what does that mean? It basically means that we have to be able to be adaptable, you know, pick up new things and have a mindset of curiosity. Like it’s easy to get scared, I think, sometimes at the pace of change.

Shahran Haider [00:25:03]:

But I think if we’re open to change, be willing to learn, you know, just continuously digging into new things, I think that’s the most important skill. And then throughout all of this, I think what’s going to become even more important, given all the technology and automation stuff that is coming our way, is being able to work well with people. It doesn’t matter how brilliant you are, how good you are at technology, at the end of the day, the human relationships and the human impacts, I really care.

Ben Hilmes [00:25:28]:

That’s awesome. And I think your guidepost of Peter Drucker and Maya Angelou are fascinating, are fantastic and aligned very much to kind of where you’re going with your team. And I recently read a study where the shelf life of a skill set is now below two and a half years. So if you think about just how rapidly that is changing, just the constant challenge to our teams to reimagine and be creative, be curious, not be afraid to say yes to trying something new. It’s going to be a differentiator for those that lead, for sure. Well, this has been great. I’d love to continue the dialogue at some point, just to continue to see how you guys are moving further on your overall data and value journey and how that continues to tie to the core mission of the organization. I, I think you guys are getting it right. So this has been fascinating. I know the listeners are going to really enjoy hearing what you had to say, so thanks for coming on today.

Shahran Haider [00:26:27]:

Thank you Ben. I enjoyed it.

Ben Hilmes [00:26:32]:

Shahran is doing innovative work at NYC Health + Hospitals. Here are a few key takeaways from our conversation. 1. Shahran’s a big believer in our product mindset. Instead of one and done projects, he’s focused on long term solutions that keep delivering value. 2. Shahran is committed to democratizing data, giving stakeholders the tools and training to access data and create their own reporting and insights. 3. His advice for young engineers is to be adaptable and curious. Those are key attributes for success in this new era of artificial intelligence. So what did you think? What were your big takeaways from the episode? I’d love to hear from you on our social media channels or drop me an email from our website.

Narrator [00:27:18]:

Thanks for joining us for Leader to Leader. To learn more about how to fuel your own personal leadership journey through the healthcare industry, visit healthcareitleaders.com. Don’t forget to subscribe so you don’t miss any insights and we’ll see you on the next episode.