S2E9 Hybrid Analytics and Data (ft. Joshua Bosire, Cooper University Health Care)

Bosire, Joshua 0:32
Thank you, Jody.

Jordan Cooper 0:34
Yeah.

Jordan Cooper 0:34
So I would like to talk about some projects you're working on in analytics.

Jordan Cooper 0:40
Specifically, you mentioned that you are working on a transition to a new analytics organizational model.

Jordan Cooper 0:48
A hybrid, a hub and spoke model.

Jordan Cooper 0:51
Would you tell us more about what you're doing and what led to this change?

Bosire, Joshua 0:57
Sure.

Bosire, Joshua 0:58
Absolutely.

Bosire, Joshua 0:59
Thanks for asking the question.

Bosire, Joshua 1:01
So I'm sure this will be familiar to most organizations, because I believe that the success of your analytics initiatives is by large controlled by your operating model.

Bosire, Joshua 1:14
That's how you execute your analytics.

Bosire, Joshua 1:16
Yeah.

Bosire, Joshua 1:17
And most of the organizations in healthcare, when we look at it, you I believe that most organizations do have some model existing an existing model of analytics, which might either be decentralized.

Bosire, Joshua 1:32
So it might be centralized, or it might be in between.

Bosire, Joshua 1:35
So for us, we've for a long time had more of a a model that I would describe more to be on the decentralized side and there have been some limitations that you have experienced in the past regarding that you know specifically, yeah you do run off and into some quality or trust issues with data are confusion for your end users maybe slow responsiveness because when you're in movie decentralized world uh the coordination is less because you have multiple teams trying to do analytics without a proper coordination towards a common goal.

Jordan Cooper 2:17
Umm.

Jordan Cooper 2:35
What's that?

Bosire, Joshua 2:17
So those are some of the challenges or limitations that you have had in the past that led us to this journey that we are on now over the last two years I'd said and after researching and looking at various industry best practices, we decided that the hybrid are operating model was best for us essentially that's really taking best of both worlds from a centralized and decentralized approach.

Bosire, Joshua 2:42
So we are really not trying to be a centralized organization, but we also trying to move away from being extremely decentralized.

Bosire, Joshua 2:50
Uh, so that.

Jordan Cooper 2:51
And so we started just to interrupt.

Bosire, Joshua 2:53
Sure.

Jordan Cooper 2:54
I'm wondering what was this driven by leadership or where was it?

Jordan Cooper 2:57
Users who were expressing frustration with the decentralized approach to analytics.

Jordan Cooper 3:01
Who was the main driver behind this push to move from decentralized to hybrid?

Bosire, Joshua 3:06
Good question.

Jordan Cooper 3:09
OK.

Bosire, Joshua 3:07
It's actually both because the feedback kind of comes top down and bottom up because your leaders are reacting to what they are hearing from the organization, from the analytics users, right?

Bosire, Joshua 3:20
So you have analytics users who are concerned about the confusion and slow responsiveness and equality issues when that cascades up to the leadership team, then that becomes an important aspect for the organization, necessitating us to actually try to tackle it.

Bosire, Joshua 3:41
Yeah.

Bosire, Joshua 3:47
Yeah.

Jordan Cooper 3:36
You mentioned that there are some data integrity issues or because when you said quality issues with with data, could you speak perhaps about a particular use case, maybe a certain analytical report that you have for a particular department and how those users have had with the decentralized approach experience, data quality issues and then how in the hybrid model now those issues are resolved?

Bosire, Joshua 4:03
So a very simple example will be where two different teams within the organization are ran data or created analytics about pretty much the same thing, but they analyze the same data differently and thing up with two numbers that are not exactly the same.

Jordan Cooper 4:21
Umm.

Bosire, Joshua 4:21
So you that often happens, and I'm sure that it's it's not unique to us, it happens and how it even from my peers where you walk into meetings.

Bosire, Joshua 4:30
But because of such an organizational situation, you end up with people presenting numbers that are different, but they are about the same thing.

Bosire, Joshua 4:39
So that of course does raise eyebrows from your senior leadership team saying, hey, I thought we are talking about the same things.

Bosire, Joshua 4:46
Yeah.

Bosire, Joshua 4:47
How comes the numbers are different?

Bosire, Joshua 4:48
Right.

Jordan Cooper 4:49
Yeah.

Jordan Cooper 4:58
What?

Bosire, Joshua 4:49
So part of try trying to change the model towards a hybrid model is more to introduce the Central Code National Central governance of the common elements of analytics that touch everybody.

Jordan Cooper 5:11
OK, both.

Bosire, Joshua 5:04
So in such a model that means that then we have a forum where we are bringing all these teams together to create a common understanding of those things.

Bosire, Joshua 5:14
So if it's the governance of the data and how we define the data we are doing that collectively as a single entity in as much as they analytics is being created and consumed in a rather decentralized manner.

Bosire, Joshua 5:28
So that's a way that we are that's the approach that we are taking and trying to solve this challenge.

Jordan Cooper 5:34
I see.

Bosire, Joshua 5:43
Yeah.

Bosire, Joshua 5:50
Yeah.

Jordan Cooper 5:35
UM, I uh would like to move on to the next topic of analytics here, which is analytics enablement and you spoke about I think we discussed before the episode about expanding self service capabilities to empower end users with analytics.

Bosire, Joshua 5:53
Yeah.

Jordan Cooper 5:52
So again, these end users that drove the model they wanted a more centralized, uh, analytics governance model.

Jordan Cooper 6:01
Now they're looking for to be empowered with self service capabilities.

Jordan Cooper 6:06
Tell me more about that project.

Bosire, Joshua 6:09
Yeah.

Jordan Cooper 6:19
But.

Bosire, Joshua 6:09
So, uh, we've been on an exciting journey as most organizations where as you as you get, you get your leadership team and your users interested in another analytics, there's always the appetite for more.

Bosire, Joshua 6:22
So that comes into fashions.

Jordan Cooper 6:29
Yeah.

Jordan Cooper 6:38
That.

Bosire, Joshua 6:24
You can choose to they buy approach where you are implementing various tools where it's your EMR or ERP systems which have embedded analytics but those do have certain limitations where you get to a point where certain things are not.

Jordan Cooper 6:53
Yeah.

Bosire, Joshua 6:39
The bailiff body are not available in the format that's necessary, so and additional proach you have taken is to create more the custom development platform using one of the leading software tableau, which then enables us to create these wealth of information.

Jordan Cooper 7:00
Yeah, for.

Bosire, Joshua 6:56
So in addition to what's available from within existing systems, we have been able to create a lot of cash from analytics that a crossover, multiple domains within the organization.

Bosire, Joshua 7:08
So what this has done is really yeah, it's helped keep on ohm satisfying that task for knowledge.

Bosire, Joshua 7:16
But I I believe at the more information that we present to users, the more the questions they're asking, because this is information that they they had necessary scene before.

Bosire, Joshua 7:28
So it's generating more demand for data if that makes sense.

Bosire, Joshua 7:36
Hmm.

Jordan Cooper 7:33
So I understood that at Cooper University Health Care, you've purchased a Tableau license and you're using that to create dashboards for your leadership team.

Jordan Cooper 7:45
Is that right?

Bosire, Joshua 7:48
Yes, and I so essentially it's not a life, it's a platform.

Bosire, Joshua 7:52
So it's essential we have created a user base of over 600 active users.

Jordan Cooper 8:04
No.

Bosire, Joshua 7:57
Yeah, which is mostly our leadership team where we have created this ecosystem of analytics tools where they can access the information about their business units based on the various business questions.

Jordan Cooper 8:21
Yeah.

Bosire, Joshua 8:13
So they are trying to answer and that's to complement the several analytics that we have from existing systems.

Bosire, Joshua 8:21
You know, like your ERP, your EMR, your financial systems, your patient experience systems.

Jordan Cooper 8:30
And.

Bosire, Joshua 8:27
So it's really creating these plethora of information so that your users have the options on how to easily access the information that they're looking for.

Jordan Cooper 8:35
That I see.

Bosire, Joshua 8:41
Yeah.

Jordan Cooper 8:37
So are you saying something like for example does make this concrete the Chief Financial officer in the past would have had to say, hey, Joshua, I need a report on accounts receivable and your team would have to do that and then email him the report versus Now the CFO can say, hey, I want this report on accounts receivable and he runs it himself.

Bosire, Joshua 9:02
I ideally, yes.

Jordan Cooper 9:13
Uh.

Bosire, Joshua 9:04
So it's exactly what you're describing that the information is available, though I think just to clarify that at some time some of the questions being asked are new.

Bosire, Joshua 9:14
So it might entail a little bit of wrangling of data, but using existing tools that have been created.

Bosire, Joshua 9:20
So it might mean changing something to be able to answer the specific question because what he or she is asking for may not exactly be available in that format.

Bosire, Joshua 9:33
Umm.

Jordan Cooper 9:30
What's the business impact been on Cooper University Health care of having users being able to drive their own analytics reports without needing your team as an intermediary?

Bosire, Joshua 9:46
So just to be clear it it's a journey I can talk about some of the benefits you have seen.

Jordan Cooper 9:54
Yeah.

Bosire, Joshua 9:51
I'm not trying by any means to say that we are about chieve the, you know, the the the gold standard of this.

Bosire, Joshua 9:58
This is a journey that we are on and we are evolving each day.

Jordan Cooper 10:02
Umm.

Jordan Cooper 10:09
Right.

Bosire, Joshua 10:02
So, specifically to speak of, for example, the access world where my role overlaps with, we have seen a lot of benefits talking about, for example, managing and improving how we use physician or provider capacity to improve access, right?

Bosire, Joshua 10:20
So for example, in last year we were able to improve that by three percentage points and I think a lot of this was mostly enabled by the availability of the data to be able to drive a conversation.

Bosire, Joshua 10:32
That is an opportunity here that we need to improve.

Bosire, Joshua 10:36
Uh, so other things like, you know, when you talk about access, these aspects of say in reducing your nauseous and cancellation.

Bosire, Joshua 10:45
So that then you have more capacity to use.

Bosire, Joshua 10:47
I think last year we saw almost a 1.5 percentage points decline in that right and improving self service when we are driving to push more patients to do things like scheduling and managing their appointments that we are able to drive a 5 percentage point increase in that.

Bosire, Joshua 11:06
Again, I'm not saying that this is all mainly because of analytics, but analytics is an enabler for the operational decisions that have to be made in order to drive that change.

Jordan Cooper 11:16
So I imagine as Cao at a peer institution somewhere else in United States, maybe listening to this episode and saying interesting.

Jordan Cooper 11:25
So this analytics enablement is allowing end user self service and analytics.

Jordan Cooper 11:31
Is that freeing up more of my teams time?

Jordan Cooper 11:34
Are they therefore no longer spending as much time creating ad hoc reports for leadership?

Jordan Cooper 11:40
And if so, is my team diminishing in size of resources being constricted or are my is my team being applied to other use cases and have more ability to innovate and work on their own projects?

Jordan Cooper 11:53
What has been the impact on your team of taking some of the report creation off their plates and pushing it right onto Andy End users?

Bosire, Joshua 12:03
So I get the same question for my team also, right?

Bosire, Joshua 12:08
So that's not just a question from peers.

Bosire, Joshua 12:10
But my team does ask the same question.

Bosire, Joshua 12:12
So in analytics, I think he's unable the things we are doing today are different than what will be doing tomorrow, right?

Bosire, Joshua 12:20
The questions that you're trying to answer for your business users by running the analytics you're doing today will be different than what you're doing tomorrow.

Bosire, Joshua 12:28
So I guess what I'm trying to say is that there is never scarcity of teams for us to work on.

Jordan Cooper 12:42
I love that.

Bosire, Joshua 12:35
In fact, I feel that the more we give people information, they more they look at it, the more questions that they have about other things that they wanna learn.

Bosire, Joshua 12:45
So I feel like it's actually been increasing, not reducing.

Bosire, Joshua 12:49
Having said that, I think there's that evolution in analytics as we try to move from just working on descriptive analytics to going all the way to a predictive and prescriptive analytics.

Jordan Cooper 13:02
He's not.

Jordan Cooper 13:03
What?

Bosire, Joshua 13:02
So there's a lot of interesting things that we are really planning on expanding our capabilities to.

Jordan Cooper 13:11
Umm.

Bosire, Joshua 13:08
When you start going into the realm of advanced analytics, simple examples might be a really leveraging on a emerging technologies to create more information from unstructured data.

Bosire, Joshua 13:23
A very concrete example is when you look in healthcare, a lot of data might either be test data or conversational data, but that's still data.

Bosire, Joshua 13:34
It's not in rows and columns for you to easily analyze, but that's still a wealth of information that's untapped somewhere, right?

Bosire, Joshua 13:43
So that's all potential of things that we exploring to move into as we free up the resources from just running reports.

Jordan Cooper 13:53
So have you been?

Bosire, Joshua 13:53
So I really don't see it.

Bosire, Joshua 13:54
I don't see this as a threat to the team.

Bosire, Joshua 13:57
I see this as a recognition of the value of the team so that we can continue doing more things.

Jordan Cooper 14:02
I see.

Jordan Cooper 14:03
So the team, the more the team gets in, gets reports that are useful in front of end users.

Bosire, Joshua 14:12
Yeah.

Bosire, Joshua 14:19
Yeah.

Jordan Cooper 14:09
The more end users have a dive or requesting to have more additional reports, more questions answered, I'd like to ask a few more questions that have come to mind when you talk about predictive and prescriptive analytics.

Jordan Cooper 14:24
Are you building these sorts of tools?

Jordan Cooper 14:26
You talk about unstructured data.

Bosire, Joshua 14:28
Umm.

Jordan Cooper 14:28
Are you leveraging open AI?

Jordan Cooper 14:30
LLM AMS are you are you building your own tools?

Jordan Cooper 14:35
How are you approaching predictive and prescriptive analytics?

Bosire, Joshua 14:39
So that's a new frontier that we are approaching now.

Jordan Cooper 14:46
OK.

Bosire, Joshua 14:43
We are running some small proof of concepts, but again we aren't healthcare, so there's a lot of sensitivity around how we manage the data.

Bosire, Joshua 14:52
So we are really staying away from the public systems like open AI, but the good thing is that we, Microsoft Shop and Microsoft as you're aware as availed a lot of tools through the ecosystem.

Bosire, Joshua 15:05
When you talk about Microsoft copilot and their yeah, GPT capabilities.

Jordan Cooper 15:17
Thank you.

Jordan Cooper 15:19
Yeah, you know.

Bosire, Joshua 15:12
So those are the things that we are tapping into to see how we can leverage on both LLM capabilities within a secure infrastructure that doesn't introduce any risk to our data.

Bosire, Joshua 15:31
Yeah.

Jordan Cooper 15:23
This I see on the topic of risk to your data and I want to kind of circle back to the getting of this conversation with the topic of data quality. Right.

Jordan Cooper 15:33
You know, the for decades had gigo, garbage in, garbage out, has been an acronym.

Bosire, Joshua 15:41
Yeah.

Jordan Cooper 15:41
Uh, well, well known in the data analytics world.

Bosire, Joshua 15:46
Yeah.

Jordan Cooper 15:45
You know, what are you doing to ensure you know great?

Jordan Cooper 15:48
You have excellent Tableau reporting capabilities that end users can use to create their own.

Jordan Cooper 15:54
You have new potential tools through Microsoft, like copilot, which allows you to leverage a private LLM instance that protects your Phi.

Jordan Cooper 16:04
But what are you doing to ensure that one data quality is improved?

Jordan Cooper 16:11
You know, I know you mentioned one that you moved to a kind of hybrid but sent centralized and decentralized governance model.

Jordan Cooper 16:17
But what are you doing to ensure one that the data is of the right quality and that it's deduplicated, aggregated, normalized, and then two, how are you working to ensure that all the data you are getting from many sources like copilot, like the EHR like patient notes like you know non clinical data, how do you ensure that those are interoperable?

Jordan Cooper 16:43
There's two questions.

Jordan Cooper 16:44
Clean data and interoperable data.

Bosire, Joshua 16:46
Yeah.

Bosire, Joshua 16:47
So that is an ongoing journey.

Jordan Cooper 16:52
Yeah.

Jordan Cooper 16:57
Thank you.

Jordan Cooper 17:07
Yeah.

Bosire, Joshua 16:51
The steps we are taking are we do have a very strong data team and the various systems that we have put in place which House our data have been developed in partnership with our you know, external partners to ensure that the data is structured in the best way possible.

Bosire, Joshua 17:11
But having said that, data quality issues always arise because you have data.

Bosire, Joshua 17:16
Ah, interpreting.

Bosire, Joshua 17:18
Pulling that data and interpreting it always introduces issues, so it's simple example might be like I mentioned earlier, two people going in to run the same data but with slightly different definitions of the same data.

Bosire, Joshua 17:31
So a very simple example.

Jordan Cooper 17:32
I.

Bosire, Joshua 17:32
So some of the very basic housekeeping things that we are working on is really trying to ensure that we have standard when you talk about definition of your metrics and KPIs that face tablished, so that we are we have an agreement of what the KP or the metrics mean and people can leverage that across the organization.

Bosire, Joshua 17:56
So for any analytical tools that we are developing, we ensure that that's available and we are working on making that more robust to cover more metrics across the organization.

Bosire, Joshua 18:06
So that really helps from the gate code to make sure that there's no confusion as you try to use that data.

Bosire, Joshua 18:12
But as you are looking at a interoperability and pulling in more data together, I think that's where some of now the behind the scenes confusion happens.

Bosire, Joshua 18:21
Meaning, if you're using the wrong tools or your folks don't, you know unsure about how they're doing specific things, there's potential for risk there.

Bosire, Joshua 18:31
So that's something we are approaching carefully.

Bosire, Joshua 18:34
We do have certain tools in place.

Bosire, Joshua 18:36
Help us with that on how we wrangle the data, but as we are looking to expand the data sources, we also exploring other technologies that might help us do that better.

Bosire, Joshua 18:47
We are having some discussions, for example with Microsoft to see if they are fabric solution might help us in bringing all that data better in a better fashion close together as we try to look around you know blending all these data to do more analytics in a more coordinated way.

Bosire, Joshua 19:10
Yeah.

Bosire, Joshua 19:19
Yeah.

Jordan Cooper 19:06
So as we approach the end of this episode, I'd like to ask you to hypothesize Joshua, if you could have kind of a 10 out of 10, right, you have some wish lists and you're saying this is what I'd like to see.

Jordan Cooper 19:22
If I had unlimited resources and you know no obstacles, what would you like to see in 2024?

Jordan Cooper 19:29
What would be a 10 out of 10 for, whether it's data interoperability, Queen data reporting being able to incorporate copilot?

Jordan Cooper 19:37
What are you looking to see?

Jordan Cooper 19:38
What wouldn't really make your life easier if if what would be kind of your 10 out of 10 for where you'd like to be with with your analytics team?

Bosire, Joshua 19:50
Well, you know, I'm tempted to see all the above all the things you said, because the truth is analytics is that it's an ongoing journey, right.

Bosire, Joshua 19:59
There's the things we are doing today.

Bosire, Joshua 20:01
We are not perfect at all.

Bosire, Joshua 20:03
Those things is making tremendous strides to improve them.

Jordan Cooper 20:06
Umm.

Bosire, Joshua 20:06
Uh, the ideal situation is that you have it all, you know, all the checkboxes.

Bosire, Joshua 20:11
Check, but I know that's unrealistic.

Jordan Cooper 20:18
Yes.

Bosire, Joshua 20:14
So if I was to pick like, what could I see as the biggest win for this year?

Jordan Cooper 20:20
Umm.

Bosire, Joshua 20:21
I would say if we could, if we could at attain more maturity on our governance issues, right, because I feel like data is all about trust and data.

Jordan Cooper 20:40
Umm.

Bosire, Joshua 20:33
So if we could resolve a lot of the, you know, the things you're working on so that people have full trust and confidence in data, I think that would be a big win for me.

Bosire, Joshua 20:42
I know it's foundational, but that's a big win because then that helps you.

Jordan Cooper 20:46
Problem.

Jordan Cooper 20:53
For.

Bosire, Joshua 20:47
Yeah, have a solid foundation so that people are not having any concerns and questions about the data they're using down the road.

Jordan Cooper 20:53
But and and just a little follow up.

Jordan Cooper 20:57
What?

Jordan Cooper 20:58
Which of the topics that we covered today, would you say would do the most to improve trust in the data that you're working with?

Bosire, Joshua 21:07
The organizational model the because at the end of the day, all these efforts have to be done by people and if you have an effective organization model, it helps actually drive your results.

Bosire, Joshua 21:18
So that's what we have been putting a lot of work into that.

Jordan Cooper 21:21
Perfect.

Jordan Cooper 21:22
Well, Joshua, I'd like to thank you for our listeners.

Jordan Cooper 21:24
Again, this has been Joshua Bosire, the VP of Access and Cao at Cooper University Healthcare.

Jordan Cooper 21:30
Joshua, thank you so much for joining us today.

Bosire, Joshua 21:33
Thank you, John.