1
00:00:00,000 --> 00:00:17,000
All right.
2
00:00:17,000 --> 00:00:22,000
Hey everyone and welcome to Enabling Digital with Systems Plus.
3
00:00:22,000 --> 00:00:23,000
I'm your host Amir.
4
00:00:23,000 --> 00:00:28,000
And today we have Eric from the Willemotte Cancer Institute.
5
00:00:28,000 --> 00:00:32,000
He is the executive director of technology innovation.
6
00:00:32,000 --> 00:00:36,000
Thank you for having me.
7
00:00:36,000 --> 00:00:39,000
All right. So Eric, you've had a very interesting career.
8
00:00:39,000 --> 00:00:47,000
It would be great to understand what has led to your current role in healthcare IT today.
9
00:00:47,000 --> 00:00:50,000
And what does your role encompass today?
10
00:00:50,000 --> 00:00:55,000
Yeah. So it's been a pretty unique career, especially for someone in the healthcare space.
11
00:00:55,000 --> 00:01:02,000
I started my career out in the outpatient pharmacy here at the University of Rochester Medical Center,
12
00:01:02,000 --> 00:01:05,000
which a lot of people don't know because we don't put it anywhere.
13
00:01:05,000 --> 00:01:10,000
But I did this while I was in college doing a computer science degree.
14
00:01:10,000 --> 00:01:13,000
I did not see that on your LinkedIn.
15
00:01:13,000 --> 00:01:17,000
Yeah, usually there's a couple of things that I left out.
16
00:01:17,000 --> 00:01:23,000
But I did that while in the computer science degree program.
17
00:01:23,000 --> 00:01:30,000
And so I think it's where I first got my first exposure to healthcare, which kind of set up the rest of it.
18
00:01:30,000 --> 00:01:37,000
But eventually I graduated and found a job in the pathology department.
19
00:01:37,000 --> 00:01:41,000
This is one of the first computer programming jobs that I had.
20
00:01:41,000 --> 00:01:48,000
It's also where I realized that at least back then, so this is over 20 years ago,
21
00:01:48,000 --> 00:01:54,000
computer programming as defined by healthcare is a little different than what I was.
22
00:01:54,000 --> 00:02:00,000
So I was a true developer, I could back end full stack and all that kind of stuff.
23
00:02:00,000 --> 00:02:06,000
But I think back then, it was more systems configuration is what we would consider it,
24
00:02:06,000 --> 00:02:09,000
dealing with the HRs and all that kind of stuff.
25
00:02:09,000 --> 00:02:12,000
So I kind of used that.
26
00:02:12,000 --> 00:02:18,000
And I like to joke that I'm kind of the grandfather of incident electronic eligibility.
27
00:02:18,000 --> 00:02:26,000
So before the HRs had all of that ability to do that, I had that up and running for strong, URMC.
28
00:02:26,000 --> 00:02:36,000
And that kind of caught the eye of a large software vendor in the United States who ended up offering me a position there.
29
00:02:36,000 --> 00:02:45,000
So it's sort of hard because I think the reason that a lot of people don't leave the healthcare space once they're in it is that stability.
30
00:02:45,000 --> 00:02:49,000
I mean, there's very rarely like layoffs or anything like that.
31
00:02:49,000 --> 00:02:54,000
But I had a really great mentor who kind of walked me through that and was like encouraging me to go.
32
00:02:54,000 --> 00:03:01,000
So I took that job and it was one of the first of several cross-country moves that I did in my career.
33
00:03:01,000 --> 00:03:08,000
So I ended up moving down to Florida and I started a job there as a traveling programmer.
34
00:03:08,000 --> 00:03:13,000
So I was flying all around the United States three times a week.
35
00:03:13,000 --> 00:03:15,000
Traveling?
36
00:03:15,000 --> 00:03:26,000
Yeah, we were going and we were entering in different healthcare organizations, attempting to help them with whatever it is that they needed.
37
00:03:26,000 --> 00:03:32,000
And the travel part was you had to be on site because of just like security issues?
38
00:03:32,000 --> 00:03:39,000
I don't know that it was necessarily security. Some of it was. I think that people just wanted someone there at the time.
39
00:03:39,000 --> 00:03:45,000
I mean, this is still early in my career, 20-ish years ago, maybe 19 years ago.
40
00:03:45,000 --> 00:03:48,000
So I don't know, but they definitely wanted to be on site.
41
00:03:48,000 --> 00:03:56,000
So I was in, no exaggeration, probably 40 to 50 of the top institutions in America.
42
00:03:56,000 --> 00:04:02,000
It's where I really learned what worked and what didn't. That informed a lot of my later career.
43
00:04:02,000 --> 00:04:09,000
But eventually traveling that much, I mean, it's the little things that get you.
44
00:04:09,000 --> 00:04:14,000
Like I was sick a lot because you're in and out of these organizations and traveling three times a week.
45
00:04:14,000 --> 00:04:27,000
It wears on you a little bit. So I ended up taking a job back in New York at the University of Rochester again when one opened up after a few years of that.
46
00:04:27,000 --> 00:04:37,000
So then I ended up back there. But it was great. I mean, there's a reason I've gone back three times now because it's a great organization.
47
00:04:37,000 --> 00:04:44,000
But I started to realize I wasn't made for just doing a singular development job.
48
00:04:44,000 --> 00:04:49,000
After traveling all over the place, I wanted to be more innovative and all that kind of stuff.
49
00:04:49,000 --> 00:04:56,000
And healthcare IT in general, at least as like EHRs go and those things, isn't really that space.
50
00:04:56,000 --> 00:05:02,000
So while I was working there, I decided that it might also be interesting to try a different industry.
51
00:05:02,000 --> 00:05:13,000
So I went back and got a cybersecurity degree. And that was the third cross-country trip. I ended up taking a job down in Louisiana this time.
52
00:05:13,000 --> 00:05:23,000
And I was lucky and privileged, I guess, that I could start over because at that point I was in senior roles and had some sort of riser capacity as a developer.
53
00:05:23,000 --> 00:05:32,000
But I started over as an analyst in cybersecurity space and eventually worked my way up to leading teams and doing penetration testing all over the United States again,
54
00:05:32,000 --> 00:05:40,000
which is really fun because you're legally breaking into buildings and hacking people and all that kind of stuff, those engagements.
55
00:05:40,000 --> 00:05:54,000
But much the way that it got all traveling as a programmer, that also gets really, really challenging as well.
56
00:05:54,000 --> 00:05:57,000
So while you were traveling, hacker, it was essentially...
57
00:05:57,000 --> 00:06:04,000
Yeah, it was a lot of fun and I saw a lot of crazy things. I learned a lot too that I still hold today.
58
00:06:04,000 --> 00:06:11,000
But eventually at that point, I knew what I wanted to do. So now we're kind of mid-career or to top mid-career.
59
00:06:11,000 --> 00:06:18,000
And I ended up having the great opportunity to come back to when Wilma opened a position.
60
00:06:18,000 --> 00:06:25,000
And after I met with all the leadership there, they really...I mean, I think it was exactly what I wanted.
61
00:06:25,000 --> 00:06:34,000
I could be innovative, but I also could make an impact on the community because I realized that that's sort of a big aspect of what I wanted to do as well.
62
00:06:34,000 --> 00:06:45,000
So in cancer, in the cancer space, this technology can do so much. So that's sort of how I got to here.
63
00:06:45,000 --> 00:06:55,000
Okay. So super interesting. And part of what you mentioned was during your first traveling stint working in healthcare,
64
00:06:55,000 --> 00:07:00,000
you saw a lot of organizations and you saw what worked and what didn't work.
65
00:07:00,000 --> 00:07:12,000
And I know you have some opinions today on healthcare IT, where it is today, where it should be, how should people be operating, the level of innovation,
66
00:07:12,000 --> 00:07:21,000
some of those things. It would be great if you could share your thoughts on just the overall space today and where some of the gaps are.
67
00:07:21,000 --> 00:07:29,000
Yeah, I don't think this would be too controversial of a statement, but I think healthcare technology in general is roughly 10 years behind the industry,
68
00:07:29,000 --> 00:07:37,000
part of it, whatever. And I think there's some valid reasons for that and then some reasons that are a little bit more challenging.
69
00:07:37,000 --> 00:07:46,000
I think valid reasons in the gap, with the advent of AI, everything's AI right now, right?
70
00:07:46,000 --> 00:07:57,000
There's some ethical concerns in healthcare, like how to use that and how to do these things. So they slow that roll down where industry can kind of steamroll forward.
71
00:07:57,000 --> 00:08:07,000
But from a gap perspective, I really think that, and I talk a lot about this, but I really think that the gap is really foundational.
72
00:08:07,000 --> 00:08:16,000
And that's what people aren't necessarily recognizing, because I think there might be a lot of, I guess, what's called toxic positivity around this.
73
00:08:16,000 --> 00:08:26,000
We have severe data problems in healthcare, data interoperability problems. We have fire and stuff like that, but there's a reason that, and this would be hours of conversation if we were to go into that.
74
00:08:26,000 --> 00:08:40,000
That is not why it's been adopted. But I think that we haven't really figured out en masse a lot of the data issues, curation, validation, and all of that in healthcare.
75
00:08:40,000 --> 00:08:54,000
And I think what happens is because of these grant mechanisms that some places have, because of whatever, they're working on AI, they're working on blockchain, they're working on whatever, VR.
76
00:08:54,000 --> 00:09:00,000
But they're working on that sort of with a broken foot. Because that's foundational.
77
00:09:00,000 --> 00:09:02,000
Yeah, there's no foundation.
78
00:09:02,000 --> 00:09:16,000
So I think that we kind of ignore that. Well, not everyone, obviously, but I think it's often overlooked. But I think that's probably the biggest gap right now that I see.
79
00:09:16,000 --> 00:09:28,000
Right. And so, you know, one of the things you have said is easy access to data has been a problem. The foundations are fundamentally not there. And because of that, everything just gets delayed.
80
00:09:28,000 --> 00:09:35,000
A simple process may take three to five days, but one could take a day.
81
00:09:35,000 --> 00:09:42,000
You know, what are some of the innovations you and your team are pioneering at Wilmot to solve some of these problems?
82
00:09:42,000 --> 00:09:53,000
Because of everything I just mentioned and because of, you know, that breadth of experience that I saw, like, I came into this with an idea of what we would need to do to prevent that, at least internally.
83
00:09:53,000 --> 00:10:03,000
This only applies to us because you would need national levels different. But we built a data platform system we called Hyperion.
84
00:10:03,000 --> 00:10:18,000
And the general idea of that was to correct some of the foundational issues so that we could more rapidly develop and iterate after that was all set. And the way that we did that was to sort of identify some of the core
85
00:10:18,000 --> 00:10:22,000
gaps in that foundation. One of them being
86
00:10:22,000 --> 00:10:27,000
oftentimes IT groups don't have the staffing and they lack the funding to
87
00:10:27,000 --> 00:10:38,000
to accomplish all the goals. And if you have, so University of Rochester is something like 40,000, probably over a state of 35,000, 40,000 employees.
88
00:10:38,000 --> 00:10:51,000
You have a lot of requests coming in. And Wilmot itself has several thousand and all that kind of stuff as part of it. So you have to have some mechanism that you can lower the skill for low enough that you can get this data,
89
00:10:51,000 --> 00:11:10,000
these reports, these complex analytics back out to the user in a reasonable time frame, you know, less than a day, hopefully. And so we did that by semantically redefining everything that was coming into the system from the HR, from the clinical trial systems.
90
00:11:10,000 --> 00:11:18,000
Because that's sort of the other problem too. You have, you know, 30 disparate systems, all that don't talk to each other, all these things.
91
00:11:18,000 --> 00:11:24,000
So it took a year or so to do this. We did it with,
92
00:11:24,000 --> 00:11:26,000
Just a year, that's quick.
93
00:11:26,000 --> 00:11:32,000
Yeah, we worked pretty, there were a lot of nights doing this too.
94
00:11:32,000 --> 00:11:47,000
But we redefined it all into really easy to define terms. So one of the examples I love to use is, you know, just a simple like poll of disease and age or whatever in both in the most common HR on the market would take, you know,
95
00:11:47,000 --> 00:11:57,000
you're crossing 12 tables and it's all over the place that the things aren't named appropriately. In our system, it's, you know, it's like a one liner.
96
00:11:57,000 --> 00:12:14,000
So you don't necessarily have to even know healthcare to deal with that. And that lowered the skill for enough where now we can sort of focus on what matters more, you know, getting, you know, actionable intelligence and all that kind of stuff.
97
00:12:14,000 --> 00:12:18,000
And so how big was your team when you were building out?
98
00:12:18,000 --> 00:12:30,000
Yeah, so when I started, I was employee one on the team and then we started building out. So I would say we did it with about four FTs that first year.
99
00:12:30,000 --> 00:12:35,000
Yeah, they were very, the team's brilliant and they're very dedicated.
100
00:12:35,000 --> 00:12:50,000
And so we built it out with that. And then there were some iterations, obviously it was up in the first year, but it's not what it is today. I mean, that that required more but for sure. Yeah, eventually we now have 12.
101
00:12:50,000 --> 00:13:10,000
And, and so is, is Hyperion based off, you know, some of the big sort of cloud providers or some infrastructure that's out there or you know, is it because you're in healthcare, you know, you can't use some of the probably common technology that we use in other enterprises, right?
102
00:13:10,000 --> 00:13:23,000
So what are some of the challenges that you faced? Yeah, so it's completely custom. The data model is completely custom that the base layer the Esperanto layer, whatever you want to call it, is, is custom.
103
00:13:23,000 --> 00:13:35,000
I think that was one of the biggest challenges though because we were, we, we originally had built this with an MSSQL system and all that kind of stuff.
104
00:13:35,000 --> 00:13:54,000
And then, okay, right. So like hosted hosted on prem in your data center. Yeah, and we were, we were able to expand that technology quite a bit. We so the way that we had originally done it was it was on very little powered equipment.
105
00:13:54,000 --> 00:14:14,000
I mean, I think we had spent $2,000 total on some. So we had to find a way to import billions of lines of data off that originally and the way that we spent a lot of time, we coded a multi threaded multi process, like, approach with like, upset routines and all these things were going on to try and do that.
106
00:14:14,000 --> 00:14:24,000
Now, now the equipment's quite a bit larger, but it worked, you know, it worked. I was able to keep things up to date in near real time.
107
00:14:24,000 --> 00:14:29,000
And still still on prem managed.
108
00:14:29,000 --> 00:14:44,000
Yeah, it's still on prem, although we can burst the cloud now. So we do have a cloud and the reason for that is genomics. So healthcare data people, people often think is big data, but it's really not unless you're getting into genomics and stuff like that.
109
00:14:44,000 --> 00:14:50,000
You're, you're talking, you know, a few billion rows, but like, I'm from equipment can handle that.
110
00:14:50,000 --> 00:15:03,000
You know, when you're getting the genomics and the genome sequencing and stuff, now you're talking terribly today. So now that we needed, and that's why we switched to build a burst to the cloud platforms for that kind of stuff.
111
00:15:03,000 --> 00:15:07,000
Yeah, most of it's still on prem.
112
00:15:07,000 --> 00:15:24,000
All right. And so you have Hyperion, it's set up, you've got that good foundation, like you said, and now you get to do some of the more fun stuff which is visualizing it and then maybe, you know, attacking use cases and I believe you built a system called Canvas around that.
113
00:15:24,000 --> 00:15:29,000
Can you sort of help us? Yeah, so now that that foundation is done.
114
00:15:29,000 --> 00:15:50,000
We, and that's how we've been able to push out so much because it's so easy to code in it. So one of the systems was Canvas. It was a geospatial analytical system that would allow any user to go in there and look up, look up anything about our patient base internally.
115
00:15:50,000 --> 00:16:01,000
And if it was internal, then it has to be someone to do the PHI. Or New York State's registry, which we also had imported. So all the incidents rates, like New York State wide and all that kind of stuff.
116
00:16:01,000 --> 00:16:12,000
They could select any demographic, any disease they wanted, if these people have been on clinical trials or the mortality rates, all that stuff, and then plot those on a map.
117
00:16:12,000 --> 00:16:26,000
And that kind of led way to quite a few different discoveries. One being around lung cancer in a certain region here in upstate New York where there seemed to be some disparity going on.
118
00:16:26,000 --> 00:16:35,000
But they used that data and that app to open a multi-million dollar facility in that area to try to help support whatever was going on there.
119
00:16:35,000 --> 00:16:46,000
So it's been used, that technology specifically has been used in many, many different research ways. So it's pretty powerful.
120
00:16:46,000 --> 00:16:50,000
And is it the same four FTEs that built this from scratch?
121
00:16:50,000 --> 00:17:01,000
No, now we have 12. Although I would say Canvas itself was built by one very brilliant developer in obviously support.
122
00:17:01,000 --> 00:17:09,000
I need to meet these people. They're all brilliant. I constantly joke that they're all giving me imposter syndrome.
123
00:17:09,000 --> 00:17:14,000
Even though I've been developing for 20 years.
124
00:17:14,000 --> 00:17:27,000
Oh man, this is honestly phenomenal. Doing some of this stuff in a year, you don't hear that. And then one guy building some of this, which is providing tremendous value.
125
00:17:27,000 --> 00:17:39,000
That's awesome. One of the things I did notice about your team when I was sort of doing my research was you and you've posted about this as you have like a head of product, which you don't see.
126
00:17:39,000 --> 00:17:49,000
You don't see that. We work with a lot of IT teams today. I would love if there's more sort of product people in those spaces, but you don't see it.
127
00:17:49,000 --> 00:17:58,000
So like what led you to hire that person and have that person in your team and realize the importance? Yeah, so this comes from that breadth of experience too.
128
00:17:58,000 --> 00:18:05,000
He was actually, I think the third hire I did or fourth maybe somewhere around there. He was very early on.
129
00:18:05,000 --> 00:18:11,000
I think that a lot of our success is attributed to a few items, this being one of them.
130
00:18:11,000 --> 00:18:28,000
I think that anytime that, especially in the healthcare space where you do have thin budgets and you do need to hire consulting and you do need to do all these things, and priorities are shifting sometimes daily, you need someone, you need a head of product.
131
00:18:28,000 --> 00:18:32,000
You need someone that is overseeing that.
132
00:18:32,000 --> 00:18:46,000
Especially with some of the leadership style that we use too. We're not a command and control leadership style, so we have to be sure about, and we're running six different projects all at the same time and all these things.
133
00:18:46,000 --> 00:18:55,000
So someone has to be there to sort of keep me in check as well to not go overboard. And that's product.
134
00:18:55,000 --> 00:19:11,000
Also someone who can go out there and interact with the different clients, figure out the scoping so the developers can focus on exactly what they need to do, which is also how we're sort of doing this with how Canvas can be built with one.
135
00:19:11,000 --> 00:19:16,000
We already had a career scope for that. He had it already designed and all that kind of stuff.
136
00:19:16,000 --> 00:19:36,000
So I think it's a huge success part of what we have done, and I would really recommend that most people, if you're going to be in the technology innovations within healthcare, should consider getting some kind of product manager.
137
00:19:36,000 --> 00:19:56,000
I think in not just healthcare, right? I think a lot of the industries, there's fundamentally a disconnect between the technology team and the business oftentimes, and having somebody to help bridge that and then just manage scope and prioritize things is important.
138
00:19:56,000 --> 00:20:01,000
Otherwise, you just get too many moving threads at once and nothing ever gets done.
139
00:20:01,000 --> 00:20:11,000
Yeah, that's very helpful. We also have the benefit though of my co-director is a practicing oncologist.
140
00:20:11,000 --> 00:20:27,000
So we have a couple, and the product manager as well had 30 years of bench research. So we have between both of them, we do have sort of that bridge, which has helped this team immensely too, because I don't always understand whatever clinical aspect they're telling me is.
141
00:20:27,000 --> 00:20:44,000
And it's very helpful to have two people, especially like my co-director, who's also a brilliant data scientist, brilliant in general. But having both of them, it really helps. It really does.
142
00:20:44,000 --> 00:21:05,000
And so you build Canvas, which was of immense value, but from what I understand, 2D was not enough, right? And you've said, I need 1FT to build me something now, your next amazing engineer to come and build things in 3D.
143
00:21:05,000 --> 00:21:10,000
You're going to hate this, but it was 1FT.
144
00:21:10,000 --> 00:21:21,000
It's not always the case with all of our projects, but the BMTs we're going on has 4FTs on it right now. But yes, this one was 1.
145
00:21:21,000 --> 00:21:41,000
Yeah, no. So I think the idea was with AR technology, you can do a lot. And I don't think VR will ever obviously replace traditional methods, but it's meant to enhance it in certain ways.
146
00:21:41,000 --> 00:21:59,000
There's a few things that I think it really enhances. One, in that geospatial suite, you theoretically could code in some kind of collaboration in there, but to do this in VR is really at another level than 2D.
147
00:21:59,000 --> 00:22:17,000
So being able to have a researcher, which does happen in California and then in one in New York, or across the world, it doesn't really matter, both in the same geospatial suite, moving that same map around and plotting, like, I think we have a cancer problem in the southern tier or something.
148
00:22:17,000 --> 00:22:29,000
That is way more powerful, I think, than doing that in 2D. It's just not as easy because you'd have a Google Doc style, like people could see things, which is really great if you like text.
149
00:22:29,000 --> 00:22:44,000
But when you're dealing with stuff like geospatial analysis, I think that that's quite a value added over. The other part of it is with VR, our current canvas suite, you can have multiple maps side by side to compare and whatever.
150
00:22:44,000 --> 00:23:02,000
But you only have so much screen space to deal with. In VR, your viewpoint is much larger, so we're able to bring not just the basic functionality of the canvas in, but you could theoretically then bring in a lot of the analytics and have that side by side and stuff like that all in that environment.
151
00:23:02,000 --> 00:23:10,000
I think the expanded view and what you can show and how you can show it is actually quite a bit more powerful in VR.
152
00:23:10,000 --> 00:23:19,000
And so do you have sort of use cases where you're taking advantage of that, essentially that third dimension, right? That you can visualize data?
153
00:23:19,000 --> 00:23:32,000
We do. It's not in production yet. We had changing priorities and whatever. We had planned to go back to it and we will. It does work.
154
00:23:32,000 --> 00:23:44,000
There are a few, we have to, there are a few analytic items that need to be added in and combining a certain systems because VR obviously is a different coding mechanism as well.
155
00:23:44,000 --> 00:23:47,000
So yeah, there's a bunch of use cases.
156
00:23:47,000 --> 00:23:57,000
And then designing for it, were you able to manage designing for it? Was it completely different? I can barely see.
157
00:23:57,000 --> 00:24:10,000
Yeah, so I think that's it. So as a developer in healthcare, I can tell you there's very little that you could ask me or tell me that I probably haven't seen in some way or couldn't look up.
158
00:24:10,000 --> 00:24:21,000
This is a bit different though. I would say that from my perspective, the reason it's different is because you still have that development layer when you're coding in VR and AR, traditional coding.
159
00:24:21,000 --> 00:24:30,000
But then there's a spatial awareness layer because you're in 3D. And I don't think I really have that, at least internally.
160
00:24:30,000 --> 00:24:41,000
You have to imagine it, I guess, right? Like there's no way for you to say like this, because you would still design or potentially put it out in 2D and then somehow just have to imagine how it works.
161
00:24:41,000 --> 00:24:58,000
Yeah, and then you have to have the math to make that work and all the stuff that are there are one that you did. But they think that way already. And I think that, you know, it's not that I couldn't do it, but I think people are, if you're going to go down that road, you kind of do need someone that thinks that way.
162
00:24:58,000 --> 00:25:06,000
And it could just be because I never learned that way. Like my coding path is C++, Python, whatever.
163
00:25:06,000 --> 00:25:15,000
And I just, you know, I didn't think in those 3D terms, but yeah, it's definitely a different build, for sure.
164
00:25:15,000 --> 00:25:27,000
That's right. And you mentioned it's not just that you guys are building VR, but there's like collaborative VR, which is like, I mean, I want to meet this FTE.
165
00:25:27,000 --> 00:25:33,000
Well, we can set up meetings.
166
00:25:33,000 --> 00:25:51,000
But I guess, you know, in general, let's say we get very close near real time collaboration in the AR VR space over next year, which I think is possible with Apple, you know, launching Vision Pro and pushing that, hopefully, you know, further down.
167
00:25:51,000 --> 00:25:57,000
Like, how do you see this benefiting healthcare as an industry?
168
00:25:57,000 --> 00:26:10,000
I think, you know, I think like, like, I kind of had mentioned that I don't think it'll ever replace fully, but I think that, you know, those collaborations are really huge.
169
00:26:10,000 --> 00:26:18,000
There's also that whole, you know, like, training for surgeries and stuff that VR has always done, which is already amazing.
170
00:26:18,000 --> 00:26:37,000
But I think we'll really haven't gotten to that data space with VR AR yet. I think that's where we could benefit the most. I mean, we definitely know that the use cases and it's in reality for training and those mechanisms, but I don't think really people have thought through a lot of that data collaboration or anything like that.
171
00:26:37,000 --> 00:26:49,000
I mean, you could also look at it from the standpoint of identifying security vulnerabilities. You could look at it from, I think one of the things that you could take a look at is genomics.
172
00:26:49,000 --> 00:26:56,000
You know, having having those large data sets there visually color coded, whatever, are going to be easy with VR as well.
173
00:26:56,000 --> 00:27:09,000
Yeah, I think that's sort of the future. Like I said, probably it won't fully replace anything, obviously, but there's a lot of benefit. Yeah.
174
00:27:09,000 --> 00:27:27,000
Right. And then maybe looking outside of data, you know, at the industry in general, what are some of the core trends you see over the next five years between, you know, of course, there's AI, there's VR, but anything else?
175
00:27:27,000 --> 00:27:36,000
I really hope that people get to those foundational layers and that that's not really a glamorous answer.
176
00:27:36,000 --> 00:27:46,000
I mean, but it's reality. I think to really get to these these more glamorous things that people really should be focusing on the data, but they're not.
177
00:27:46,000 --> 00:28:02,000
And I think, you know, some of the things that I'm interested in for the future of the five years are CRDTs, stuff like that, sort of because I think that that's, I think blockchain, for example, I was really interested in that for a while.
178
00:28:02,000 --> 00:28:12,000
But I think that there's just too much hesitation around that because of crypto, even though it's not the same of what we were, we were pretty much.
179
00:28:12,000 --> 00:28:13,000
Yeah, it's back up again.
180
00:28:13,000 --> 00:28:23,000
Yeah, you don't trust it, people don't ever, but I think, you know, CRDTs as a data type kind of sort of accomplish the same goals there. So I'm interested in that.
181
00:28:23,000 --> 00:28:32,000
I think AI, I mean, obviously that's going to be, I don't think we're going to stop hearing about that for any time soon.
182
00:28:32,000 --> 00:28:43,000
Yeah, but you need the, like you said, I think from a like proprietary security perspective, you don't want to send your data necessarily outside.
183
00:28:43,000 --> 00:28:54,000
So if you want to build your own models inside, you need those foundational layers, which a lot of companies, I mean, they just don't have. And in healthcare, you also need clean data.
184
00:28:54,000 --> 00:29:11,000
And to trust the model if your data, with what your data looks like. I mean, we, like internally, we've taken a lot of steps to clean data, but I mean, what if most institutions, you know, the race ethnicity data is very bad.
185
00:29:11,000 --> 00:29:23,000
And, you know, stuff like that. So yeah, there's a lot of concern with AI, I think, at least internally.
186
00:29:23,000 --> 00:29:28,000
All right. I think I may have lost you for those last like 10 seconds.
187
00:29:28,000 --> 00:29:42,000
Yeah, I think you're right.
188
00:29:42,000 --> 00:30:06,000
Is there, I guess, if anybody in healthcare today wanted to go through this data transformation, any recommendations based on your own learnings and then how do they continue to keep data clean, which is impossible, but like what can they, what are some of the practices they can put in place to ensure this, you know, they don't land up back in this messy data situation?
189
00:30:06,000 --> 00:30:19,000
Yeah, I think we handled, we tackled this in another way as well. I didn't mention it, but we built in off thousands and thousands of lines of data validation scripts. So you got to be careful.
190
00:30:19,000 --> 00:30:28,000
Like, you can't automatically fix things, but the idea would be to proactively flag people if something looks off, so that it can be corrected very quickly.
191
00:30:28,000 --> 00:30:55,000
And there's many ways that you can do that, that the EHR really should be doing, but they're not, you know, we have access to, I think we're at 90 something interfaces now, but things like the registries or the clinical trial systems or whatever it is that you, if a name and a race and a gender and ethnicity gets entered and it's different from the clinical trial system, you know you've got some problem.
192
00:30:55,000 --> 00:31:10,000
One system or the other, one would think the clinical trial system is correct because, you know, that's the more audited one, but you never know. So you should be flagging those for someone to ask the patient or to follow up on.
193
00:31:10,000 --> 00:31:24,000
There's things like address information. So because we do so much with Canvas and geocoding, those addresses have to be right because we're doing social determinants of health off that, we're doing all these things.
194
00:31:24,000 --> 00:31:39,000
So we're also plugged into the Census Bureau and other places. So if address gets entered immediately in the epic, we know if it's a valid address or not, even if they flipped a letter or something, that's fine. But if the address just doesn't exist, maybe it's a new address or whatever.
195
00:31:39,000 --> 00:31:58,000
So I think it's really important to have, whatever you're doing, a lot of those validation type scripts and routines built to kind of try to keep it clean. Although it's, you know, like you said, sort of impossible. You're just trying the best you can.
196
00:31:58,000 --> 00:32:02,000
Yeah, yeah. We try to be as good as we can.
197
00:32:02,000 --> 00:32:04,000
Awesome.
198
00:32:04,000 --> 00:32:09,000
All right. Well, thanks Eric for joining and appreciate your time.
00:00:00,000 --> 00:00:17,000
All right.
2
00:00:17,000 --> 00:00:22,000
Hey everyone and welcome to Enabling Digital with Systems Plus.
3
00:00:22,000 --> 00:00:23,000
I'm your host Amir.
4
00:00:23,000 --> 00:00:28,000
And today we have Eric from the Willemotte Cancer Institute.
5
00:00:28,000 --> 00:00:32,000
He is the executive director of technology innovation.
6
00:00:32,000 --> 00:00:36,000
Thank you for having me.
7
00:00:36,000 --> 00:00:39,000
All right. So Eric, you've had a very interesting career.
8
00:00:39,000 --> 00:00:47,000
It would be great to understand what has led to your current role in healthcare IT today.
9
00:00:47,000 --> 00:00:50,000
And what does your role encompass today?
10
00:00:50,000 --> 00:00:55,000
Yeah. So it's been a pretty unique career, especially for someone in the healthcare space.
11
00:00:55,000 --> 00:01:02,000
I started my career out in the outpatient pharmacy here at the University of Rochester Medical Center,
12
00:01:02,000 --> 00:01:05,000
which a lot of people don't know because we don't put it anywhere.
13
00:01:05,000 --> 00:01:10,000
But I did this while I was in college doing a computer science degree.
14
00:01:10,000 --> 00:01:13,000
I did not see that on your LinkedIn.
15
00:01:13,000 --> 00:01:17,000
Yeah, usually there's a couple of things that I left out.
16
00:01:17,000 --> 00:01:23,000
But I did that while in the computer science degree program.
17
00:01:23,000 --> 00:01:30,000
And so I think it's where I first got my first exposure to healthcare, which kind of set up the rest of it.
18
00:01:30,000 --> 00:01:37,000
But eventually I graduated and found a job in the pathology department.
19
00:01:37,000 --> 00:01:41,000
This is one of the first computer programming jobs that I had.
20
00:01:41,000 --> 00:01:48,000
It's also where I realized that at least back then, so this is over 20 years ago,
21
00:01:48,000 --> 00:01:54,000
computer programming as defined by healthcare is a little different than what I was.
22
00:01:54,000 --> 00:02:00,000
So I was a true developer, I could back end full stack and all that kind of stuff.
23
00:02:00,000 --> 00:02:06,000
But I think back then, it was more systems configuration is what we would consider it,
24
00:02:06,000 --> 00:02:09,000
dealing with the HRs and all that kind of stuff.
25
00:02:09,000 --> 00:02:12,000
So I kind of used that.
26
00:02:12,000 --> 00:02:18,000
And I like to joke that I'm kind of the grandfather of incident electronic eligibility.
27
00:02:18,000 --> 00:02:26,000
So before the HRs had all of that ability to do that, I had that up and running for strong, URMC.
28
00:02:26,000 --> 00:02:36,000
And that kind of caught the eye of a large software vendor in the United States who ended up offering me a position there.
29
00:02:36,000 --> 00:02:45,000
So it's sort of hard because I think the reason that a lot of people don't leave the healthcare space once they're in it is that stability.
30
00:02:45,000 --> 00:02:49,000
I mean, there's very rarely like layoffs or anything like that.
31
00:02:49,000 --> 00:02:54,000
But I had a really great mentor who kind of walked me through that and was like encouraging me to go.
32
00:02:54,000 --> 00:03:01,000
So I took that job and it was one of the first of several cross-country moves that I did in my career.
33
00:03:01,000 --> 00:03:08,000
So I ended up moving down to Florida and I started a job there as a traveling programmer.
34
00:03:08,000 --> 00:03:13,000
So I was flying all around the United States three times a week.
35
00:03:13,000 --> 00:03:15,000
Traveling?
36
00:03:15,000 --> 00:03:26,000
Yeah, we were going and we were entering in different healthcare organizations, attempting to help them with whatever it is that they needed.
37
00:03:26,000 --> 00:03:32,000
And the travel part was you had to be on site because of just like security issues?
38
00:03:32,000 --> 00:03:39,000
I don't know that it was necessarily security. Some of it was. I think that people just wanted someone there at the time.
39
00:03:39,000 --> 00:03:45,000
I mean, this is still early in my career, 20-ish years ago, maybe 19 years ago.
40
00:03:45,000 --> 00:03:48,000
So I don't know, but they definitely wanted to be on site.
41
00:03:48,000 --> 00:03:56,000
So I was in, no exaggeration, probably 40 to 50 of the top institutions in America.
42
00:03:56,000 --> 00:04:02,000
It's where I really learned what worked and what didn't. That informed a lot of my later career.
43
00:04:02,000 --> 00:04:09,000
But eventually traveling that much, I mean, it's the little things that get you.
44
00:04:09,000 --> 00:04:14,000
Like I was sick a lot because you're in and out of these organizations and traveling three times a week.
45
00:04:14,000 --> 00:04:27,000
It wears on you a little bit. So I ended up taking a job back in New York at the University of Rochester again when one opened up after a few years of that.
46
00:04:27,000 --> 00:04:37,000
So then I ended up back there. But it was great. I mean, there's a reason I've gone back three times now because it's a great organization.
47
00:04:37,000 --> 00:04:44,000
But I started to realize I wasn't made for just doing a singular development job.
48
00:04:44,000 --> 00:04:49,000
After traveling all over the place, I wanted to be more innovative and all that kind of stuff.
49
00:04:49,000 --> 00:04:56,000
And healthcare IT in general, at least as like EHRs go and those things, isn't really that space.
50
00:04:56,000 --> 00:05:02,000
So while I was working there, I decided that it might also be interesting to try a different industry.
51
00:05:02,000 --> 00:05:13,000
So I went back and got a cybersecurity degree. And that was the third cross-country trip. I ended up taking a job down in Louisiana this time.
52
00:05:13,000 --> 00:05:23,000
And I was lucky and privileged, I guess, that I could start over because at that point I was in senior roles and had some sort of riser capacity as a developer.
53
00:05:23,000 --> 00:05:32,000
But I started over as an analyst in cybersecurity space and eventually worked my way up to leading teams and doing penetration testing all over the United States again,
54
00:05:32,000 --> 00:05:40,000
which is really fun because you're legally breaking into buildings and hacking people and all that kind of stuff, those engagements.
55
00:05:40,000 --> 00:05:54,000
But much the way that it got all traveling as a programmer, that also gets really, really challenging as well.
56
00:05:54,000 --> 00:05:57,000
So while you were traveling, hacker, it was essentially...
57
00:05:57,000 --> 00:06:04,000
Yeah, it was a lot of fun and I saw a lot of crazy things. I learned a lot too that I still hold today.
58
00:06:04,000 --> 00:06:11,000
But eventually at that point, I knew what I wanted to do. So now we're kind of mid-career or to top mid-career.
59
00:06:11,000 --> 00:06:18,000
And I ended up having the great opportunity to come back to when Wilma opened a position.
60
00:06:18,000 --> 00:06:25,000
And after I met with all the leadership there, they really...I mean, I think it was exactly what I wanted.
61
00:06:25,000 --> 00:06:34,000
I could be innovative, but I also could make an impact on the community because I realized that that's sort of a big aspect of what I wanted to do as well.
62
00:06:34,000 --> 00:06:45,000
So in cancer, in the cancer space, this technology can do so much. So that's sort of how I got to here.
63
00:06:45,000 --> 00:06:55,000
Okay. So super interesting. And part of what you mentioned was during your first traveling stint working in healthcare,
64
00:06:55,000 --> 00:07:00,000
you saw a lot of organizations and you saw what worked and what didn't work.
65
00:07:00,000 --> 00:07:12,000
And I know you have some opinions today on healthcare IT, where it is today, where it should be, how should people be operating, the level of innovation,
66
00:07:12,000 --> 00:07:21,000
some of those things. It would be great if you could share your thoughts on just the overall space today and where some of the gaps are.
67
00:07:21,000 --> 00:07:29,000
Yeah, I don't think this would be too controversial of a statement, but I think healthcare technology in general is roughly 10 years behind the industry,
68
00:07:29,000 --> 00:07:37,000
part of it, whatever. And I think there's some valid reasons for that and then some reasons that are a little bit more challenging.
69
00:07:37,000 --> 00:07:46,000
I think valid reasons in the gap, with the advent of AI, everything's AI right now, right?
70
00:07:46,000 --> 00:07:57,000
There's some ethical concerns in healthcare, like how to use that and how to do these things. So they slow that roll down where industry can kind of steamroll forward.
71
00:07:57,000 --> 00:08:07,000
But from a gap perspective, I really think that, and I talk a lot about this, but I really think that the gap is really foundational.
72
00:08:07,000 --> 00:08:16,000
And that's what people aren't necessarily recognizing, because I think there might be a lot of, I guess, what's called toxic positivity around this.
73
00:08:16,000 --> 00:08:26,000
We have severe data problems in healthcare, data interoperability problems. We have fire and stuff like that, but there's a reason that, and this would be hours of conversation if we were to go into that.
74
00:08:26,000 --> 00:08:40,000
That is not why it's been adopted. But I think that we haven't really figured out en masse a lot of the data issues, curation, validation, and all of that in healthcare.
75
00:08:40,000 --> 00:08:54,000
And I think what happens is because of these grant mechanisms that some places have, because of whatever, they're working on AI, they're working on blockchain, they're working on whatever, VR.
76
00:08:54,000 --> 00:09:00,000
But they're working on that sort of with a broken foot. Because that's foundational.
77
00:09:00,000 --> 00:09:02,000
Yeah, there's no foundation.
78
00:09:02,000 --> 00:09:16,000
So I think that we kind of ignore that. Well, not everyone, obviously, but I think it's often overlooked. But I think that's probably the biggest gap right now that I see.
79
00:09:16,000 --> 00:09:28,000
Right. And so, you know, one of the things you have said is easy access to data has been a problem. The foundations are fundamentally not there. And because of that, everything just gets delayed.
80
00:09:28,000 --> 00:09:35,000
A simple process may take three to five days, but one could take a day.
81
00:09:35,000 --> 00:09:42,000
You know, what are some of the innovations you and your team are pioneering at Wilmot to solve some of these problems?
82
00:09:42,000 --> 00:09:53,000
Because of everything I just mentioned and because of, you know, that breadth of experience that I saw, like, I came into this with an idea of what we would need to do to prevent that, at least internally.
83
00:09:53,000 --> 00:10:03,000
This only applies to us because you would need national levels different. But we built a data platform system we called Hyperion.
84
00:10:03,000 --> 00:10:18,000
And the general idea of that was to correct some of the foundational issues so that we could more rapidly develop and iterate after that was all set. And the way that we did that was to sort of identify some of the core
85
00:10:18,000 --> 00:10:22,000
gaps in that foundation. One of them being
86
00:10:22,000 --> 00:10:27,000
oftentimes IT groups don't have the staffing and they lack the funding to
87
00:10:27,000 --> 00:10:38,000
to accomplish all the goals. And if you have, so University of Rochester is something like 40,000, probably over a state of 35,000, 40,000 employees.
88
00:10:38,000 --> 00:10:51,000
You have a lot of requests coming in. And Wilmot itself has several thousand and all that kind of stuff as part of it. So you have to have some mechanism that you can lower the skill for low enough that you can get this data,
89
00:10:51,000 --> 00:11:10,000
these reports, these complex analytics back out to the user in a reasonable time frame, you know, less than a day, hopefully. And so we did that by semantically redefining everything that was coming into the system from the HR, from the clinical trial systems.
90
00:11:10,000 --> 00:11:18,000
Because that's sort of the other problem too. You have, you know, 30 disparate systems, all that don't talk to each other, all these things.
91
00:11:18,000 --> 00:11:24,000
So it took a year or so to do this. We did it with,
92
00:11:24,000 --> 00:11:26,000
Just a year, that's quick.
93
00:11:26,000 --> 00:11:32,000
Yeah, we worked pretty, there were a lot of nights doing this too.
94
00:11:32,000 --> 00:11:47,000
But we redefined it all into really easy to define terms. So one of the examples I love to use is, you know, just a simple like poll of disease and age or whatever in both in the most common HR on the market would take, you know,
95
00:11:47,000 --> 00:11:57,000
you're crossing 12 tables and it's all over the place that the things aren't named appropriately. In our system, it's, you know, it's like a one liner.
96
00:11:57,000 --> 00:12:14,000
So you don't necessarily have to even know healthcare to deal with that. And that lowered the skill for enough where now we can sort of focus on what matters more, you know, getting, you know, actionable intelligence and all that kind of stuff.
97
00:12:14,000 --> 00:12:18,000
And so how big was your team when you were building out?
98
00:12:18,000 --> 00:12:30,000
Yeah, so when I started, I was employee one on the team and then we started building out. So I would say we did it with about four FTs that first year.
99
00:12:30,000 --> 00:12:35,000
Yeah, they were very, the team's brilliant and they're very dedicated.
100
00:12:35,000 --> 00:12:50,000
And so we built it out with that. And then there were some iterations, obviously it was up in the first year, but it's not what it is today. I mean, that that required more but for sure. Yeah, eventually we now have 12.
101
00:12:50,000 --> 00:13:10,000
And, and so is, is Hyperion based off, you know, some of the big sort of cloud providers or some infrastructure that's out there or you know, is it because you're in healthcare, you know, you can't use some of the probably common technology that we use in other enterprises, right?
102
00:13:10,000 --> 00:13:23,000
So what are some of the challenges that you faced? Yeah, so it's completely custom. The data model is completely custom that the base layer the Esperanto layer, whatever you want to call it, is, is custom.
103
00:13:23,000 --> 00:13:35,000
I think that was one of the biggest challenges though because we were, we, we originally had built this with an MSSQL system and all that kind of stuff.
104
00:13:35,000 --> 00:13:54,000
And then, okay, right. So like hosted hosted on prem in your data center. Yeah, and we were, we were able to expand that technology quite a bit. We so the way that we had originally done it was it was on very little powered equipment.
105
00:13:54,000 --> 00:14:14,000
I mean, I think we had spent $2,000 total on some. So we had to find a way to import billions of lines of data off that originally and the way that we spent a lot of time, we coded a multi threaded multi process, like, approach with like, upset routines and all these things were going on to try and do that.
106
00:14:14,000 --> 00:14:24,000
Now, now the equipment's quite a bit larger, but it worked, you know, it worked. I was able to keep things up to date in near real time.
107
00:14:24,000 --> 00:14:29,000
And still still on prem managed.
108
00:14:29,000 --> 00:14:44,000
Yeah, it's still on prem, although we can burst the cloud now. So we do have a cloud and the reason for that is genomics. So healthcare data people, people often think is big data, but it's really not unless you're getting into genomics and stuff like that.
109
00:14:44,000 --> 00:14:50,000
You're, you're talking, you know, a few billion rows, but like, I'm from equipment can handle that.
110
00:14:50,000 --> 00:15:03,000
You know, when you're getting the genomics and the genome sequencing and stuff, now you're talking terribly today. So now that we needed, and that's why we switched to build a burst to the cloud platforms for that kind of stuff.
111
00:15:03,000 --> 00:15:07,000
Yeah, most of it's still on prem.
112
00:15:07,000 --> 00:15:24,000
All right. And so you have Hyperion, it's set up, you've got that good foundation, like you said, and now you get to do some of the more fun stuff which is visualizing it and then maybe, you know, attacking use cases and I believe you built a system called Canvas around that.
113
00:15:24,000 --> 00:15:29,000
Can you sort of help us? Yeah, so now that that foundation is done.
114
00:15:29,000 --> 00:15:50,000
We, and that's how we've been able to push out so much because it's so easy to code in it. So one of the systems was Canvas. It was a geospatial analytical system that would allow any user to go in there and look up, look up anything about our patient base internally.
115
00:15:50,000 --> 00:16:01,000
And if it was internal, then it has to be someone to do the PHI. Or New York State's registry, which we also had imported. So all the incidents rates, like New York State wide and all that kind of stuff.
116
00:16:01,000 --> 00:16:12,000
They could select any demographic, any disease they wanted, if these people have been on clinical trials or the mortality rates, all that stuff, and then plot those on a map.
117
00:16:12,000 --> 00:16:26,000
And that kind of led way to quite a few different discoveries. One being around lung cancer in a certain region here in upstate New York where there seemed to be some disparity going on.
118
00:16:26,000 --> 00:16:35,000
But they used that data and that app to open a multi-million dollar facility in that area to try to help support whatever was going on there.
119
00:16:35,000 --> 00:16:46,000
So it's been used, that technology specifically has been used in many, many different research ways. So it's pretty powerful.
120
00:16:46,000 --> 00:16:50,000
And is it the same four FTEs that built this from scratch?
121
00:16:50,000 --> 00:17:01,000
No, now we have 12. Although I would say Canvas itself was built by one very brilliant developer in obviously support.
122
00:17:01,000 --> 00:17:09,000
I need to meet these people. They're all brilliant. I constantly joke that they're all giving me imposter syndrome.
123
00:17:09,000 --> 00:17:14,000
Even though I've been developing for 20 years.
124
00:17:14,000 --> 00:17:27,000
Oh man, this is honestly phenomenal. Doing some of this stuff in a year, you don't hear that. And then one guy building some of this, which is providing tremendous value.
125
00:17:27,000 --> 00:17:39,000
That's awesome. One of the things I did notice about your team when I was sort of doing my research was you and you've posted about this as you have like a head of product, which you don't see.
126
00:17:39,000 --> 00:17:49,000
You don't see that. We work with a lot of IT teams today. I would love if there's more sort of product people in those spaces, but you don't see it.
127
00:17:49,000 --> 00:17:58,000
So like what led you to hire that person and have that person in your team and realize the importance? Yeah, so this comes from that breadth of experience too.
128
00:17:58,000 --> 00:18:05,000
He was actually, I think the third hire I did or fourth maybe somewhere around there. He was very early on.
129
00:18:05,000 --> 00:18:11,000
I think that a lot of our success is attributed to a few items, this being one of them.
130
00:18:11,000 --> 00:18:28,000
I think that anytime that, especially in the healthcare space where you do have thin budgets and you do need to hire consulting and you do need to do all these things, and priorities are shifting sometimes daily, you need someone, you need a head of product.
131
00:18:28,000 --> 00:18:32,000
You need someone that is overseeing that.
132
00:18:32,000 --> 00:18:46,000
Especially with some of the leadership style that we use too. We're not a command and control leadership style, so we have to be sure about, and we're running six different projects all at the same time and all these things.
133
00:18:46,000 --> 00:18:55,000
So someone has to be there to sort of keep me in check as well to not go overboard. And that's product.
134
00:18:55,000 --> 00:19:11,000
Also someone who can go out there and interact with the different clients, figure out the scoping so the developers can focus on exactly what they need to do, which is also how we're sort of doing this with how Canvas can be built with one.
135
00:19:11,000 --> 00:19:16,000
We already had a career scope for that. He had it already designed and all that kind of stuff.
136
00:19:16,000 --> 00:19:36,000
So I think it's a huge success part of what we have done, and I would really recommend that most people, if you're going to be in the technology innovations within healthcare, should consider getting some kind of product manager.
137
00:19:36,000 --> 00:19:56,000
I think in not just healthcare, right? I think a lot of the industries, there's fundamentally a disconnect between the technology team and the business oftentimes, and having somebody to help bridge that and then just manage scope and prioritize things is important.
138
00:19:56,000 --> 00:20:01,000
Otherwise, you just get too many moving threads at once and nothing ever gets done.
139
00:20:01,000 --> 00:20:11,000
Yeah, that's very helpful. We also have the benefit though of my co-director is a practicing oncologist.
140
00:20:11,000 --> 00:20:27,000
So we have a couple, and the product manager as well had 30 years of bench research. So we have between both of them, we do have sort of that bridge, which has helped this team immensely too, because I don't always understand whatever clinical aspect they're telling me is.
141
00:20:27,000 --> 00:20:44,000
And it's very helpful to have two people, especially like my co-director, who's also a brilliant data scientist, brilliant in general. But having both of them, it really helps. It really does.
142
00:20:44,000 --> 00:21:05,000
And so you build Canvas, which was of immense value, but from what I understand, 2D was not enough, right? And you've said, I need 1FT to build me something now, your next amazing engineer to come and build things in 3D.
143
00:21:05,000 --> 00:21:10,000
You're going to hate this, but it was 1FT.
144
00:21:10,000 --> 00:21:21,000
It's not always the case with all of our projects, but the BMTs we're going on has 4FTs on it right now. But yes, this one was 1.
145
00:21:21,000 --> 00:21:41,000
Yeah, no. So I think the idea was with AR technology, you can do a lot. And I don't think VR will ever obviously replace traditional methods, but it's meant to enhance it in certain ways.
146
00:21:41,000 --> 00:21:59,000
There's a few things that I think it really enhances. One, in that geospatial suite, you theoretically could code in some kind of collaboration in there, but to do this in VR is really at another level than 2D.
147
00:21:59,000 --> 00:22:17,000
So being able to have a researcher, which does happen in California and then in one in New York, or across the world, it doesn't really matter, both in the same geospatial suite, moving that same map around and plotting, like, I think we have a cancer problem in the southern tier or something.
148
00:22:17,000 --> 00:22:29,000
That is way more powerful, I think, than doing that in 2D. It's just not as easy because you'd have a Google Doc style, like people could see things, which is really great if you like text.
149
00:22:29,000 --> 00:22:44,000
But when you're dealing with stuff like geospatial analysis, I think that that's quite a value added over. The other part of it is with VR, our current canvas suite, you can have multiple maps side by side to compare and whatever.
150
00:22:44,000 --> 00:23:02,000
But you only have so much screen space to deal with. In VR, your viewpoint is much larger, so we're able to bring not just the basic functionality of the canvas in, but you could theoretically then bring in a lot of the analytics and have that side by side and stuff like that all in that environment.
151
00:23:02,000 --> 00:23:10,000
I think the expanded view and what you can show and how you can show it is actually quite a bit more powerful in VR.
152
00:23:10,000 --> 00:23:19,000
And so do you have sort of use cases where you're taking advantage of that, essentially that third dimension, right? That you can visualize data?
153
00:23:19,000 --> 00:23:32,000
We do. It's not in production yet. We had changing priorities and whatever. We had planned to go back to it and we will. It does work.
154
00:23:32,000 --> 00:23:44,000
There are a few, we have to, there are a few analytic items that need to be added in and combining a certain systems because VR obviously is a different coding mechanism as well.
155
00:23:44,000 --> 00:23:47,000
So yeah, there's a bunch of use cases.
156
00:23:47,000 --> 00:23:57,000
And then designing for it, were you able to manage designing for it? Was it completely different? I can barely see.
157
00:23:57,000 --> 00:24:10,000
Yeah, so I think that's it. So as a developer in healthcare, I can tell you there's very little that you could ask me or tell me that I probably haven't seen in some way or couldn't look up.
158
00:24:10,000 --> 00:24:21,000
This is a bit different though. I would say that from my perspective, the reason it's different is because you still have that development layer when you're coding in VR and AR, traditional coding.
159
00:24:21,000 --> 00:24:30,000
But then there's a spatial awareness layer because you're in 3D. And I don't think I really have that, at least internally.
160
00:24:30,000 --> 00:24:41,000
You have to imagine it, I guess, right? Like there's no way for you to say like this, because you would still design or potentially put it out in 2D and then somehow just have to imagine how it works.
161
00:24:41,000 --> 00:24:58,000
Yeah, and then you have to have the math to make that work and all the stuff that are there are one that you did. But they think that way already. And I think that, you know, it's not that I couldn't do it, but I think people are, if you're going to go down that road, you kind of do need someone that thinks that way.
162
00:24:58,000 --> 00:25:06,000
And it could just be because I never learned that way. Like my coding path is C++, Python, whatever.
163
00:25:06,000 --> 00:25:15,000
And I just, you know, I didn't think in those 3D terms, but yeah, it's definitely a different build, for sure.
164
00:25:15,000 --> 00:25:27,000
That's right. And you mentioned it's not just that you guys are building VR, but there's like collaborative VR, which is like, I mean, I want to meet this FTE.
165
00:25:27,000 --> 00:25:33,000
Well, we can set up meetings.
166
00:25:33,000 --> 00:25:51,000
But I guess, you know, in general, let's say we get very close near real time collaboration in the AR VR space over next year, which I think is possible with Apple, you know, launching Vision Pro and pushing that, hopefully, you know, further down.
167
00:25:51,000 --> 00:25:57,000
Like, how do you see this benefiting healthcare as an industry?
168
00:25:57,000 --> 00:26:10,000
I think, you know, I think like, like, I kind of had mentioned that I don't think it'll ever replace fully, but I think that, you know, those collaborations are really huge.
169
00:26:10,000 --> 00:26:18,000
There's also that whole, you know, like, training for surgeries and stuff that VR has always done, which is already amazing.
170
00:26:18,000 --> 00:26:37,000
But I think we'll really haven't gotten to that data space with VR AR yet. I think that's where we could benefit the most. I mean, we definitely know that the use cases and it's in reality for training and those mechanisms, but I don't think really people have thought through a lot of that data collaboration or anything like that.
171
00:26:37,000 --> 00:26:49,000
I mean, you could also look at it from the standpoint of identifying security vulnerabilities. You could look at it from, I think one of the things that you could take a look at is genomics.
172
00:26:49,000 --> 00:26:56,000
You know, having having those large data sets there visually color coded, whatever, are going to be easy with VR as well.
173
00:26:56,000 --> 00:27:09,000
Yeah, I think that's sort of the future. Like I said, probably it won't fully replace anything, obviously, but there's a lot of benefit. Yeah.
174
00:27:09,000 --> 00:27:27,000
Right. And then maybe looking outside of data, you know, at the industry in general, what are some of the core trends you see over the next five years between, you know, of course, there's AI, there's VR, but anything else?
175
00:27:27,000 --> 00:27:36,000
I really hope that people get to those foundational layers and that that's not really a glamorous answer.
176
00:27:36,000 --> 00:27:46,000
I mean, but it's reality. I think to really get to these these more glamorous things that people really should be focusing on the data, but they're not.
177
00:27:46,000 --> 00:28:02,000
And I think, you know, some of the things that I'm interested in for the future of the five years are CRDTs, stuff like that, sort of because I think that that's, I think blockchain, for example, I was really interested in that for a while.
178
00:28:02,000 --> 00:28:12,000
But I think that there's just too much hesitation around that because of crypto, even though it's not the same of what we were, we were pretty much.
179
00:28:12,000 --> 00:28:13,000
Yeah, it's back up again.
180
00:28:13,000 --> 00:28:23,000
Yeah, you don't trust it, people don't ever, but I think, you know, CRDTs as a data type kind of sort of accomplish the same goals there. So I'm interested in that.
181
00:28:23,000 --> 00:28:32,000
I think AI, I mean, obviously that's going to be, I don't think we're going to stop hearing about that for any time soon.
182
00:28:32,000 --> 00:28:43,000
Yeah, but you need the, like you said, I think from a like proprietary security perspective, you don't want to send your data necessarily outside.
183
00:28:43,000 --> 00:28:54,000
So if you want to build your own models inside, you need those foundational layers, which a lot of companies, I mean, they just don't have. And in healthcare, you also need clean data.
184
00:28:54,000 --> 00:29:11,000
And to trust the model if your data, with what your data looks like. I mean, we, like internally, we've taken a lot of steps to clean data, but I mean, what if most institutions, you know, the race ethnicity data is very bad.
185
00:29:11,000 --> 00:29:23,000
And, you know, stuff like that. So yeah, there's a lot of concern with AI, I think, at least internally.
186
00:29:23,000 --> 00:29:28,000
All right. I think I may have lost you for those last like 10 seconds.
187
00:29:28,000 --> 00:29:42,000
Yeah, I think you're right.
188
00:29:42,000 --> 00:30:06,000
Is there, I guess, if anybody in healthcare today wanted to go through this data transformation, any recommendations based on your own learnings and then how do they continue to keep data clean, which is impossible, but like what can they, what are some of the practices they can put in place to ensure this, you know, they don't land up back in this messy data situation?
189
00:30:06,000 --> 00:30:19,000
Yeah, I think we handled, we tackled this in another way as well. I didn't mention it, but we built in off thousands and thousands of lines of data validation scripts. So you got to be careful.
190
00:30:19,000 --> 00:30:28,000
Like, you can't automatically fix things, but the idea would be to proactively flag people if something looks off, so that it can be corrected very quickly.
191
00:30:28,000 --> 00:30:55,000
And there's many ways that you can do that, that the EHR really should be doing, but they're not, you know, we have access to, I think we're at 90 something interfaces now, but things like the registries or the clinical trial systems or whatever it is that you, if a name and a race and a gender and ethnicity gets entered and it's different from the clinical trial system, you know you've got some problem.
192
00:30:55,000 --> 00:31:10,000
One system or the other, one would think the clinical trial system is correct because, you know, that's the more audited one, but you never know. So you should be flagging those for someone to ask the patient or to follow up on.
193
00:31:10,000 --> 00:31:24,000
There's things like address information. So because we do so much with Canvas and geocoding, those addresses have to be right because we're doing social determinants of health off that, we're doing all these things.
194
00:31:24,000 --> 00:31:39,000
So we're also plugged into the Census Bureau and other places. So if address gets entered immediately in the epic, we know if it's a valid address or not, even if they flipped a letter or something, that's fine. But if the address just doesn't exist, maybe it's a new address or whatever.
195
00:31:39,000 --> 00:31:58,000
So I think it's really important to have, whatever you're doing, a lot of those validation type scripts and routines built to kind of try to keep it clean. Although it's, you know, like you said, sort of impossible. You're just trying the best you can.
196
00:31:58,000 --> 00:32:02,000
Yeah, yeah. We try to be as good as we can.
197
00:32:02,000 --> 00:32:04,000
Awesome.
198
00:32:04,000 --> 00:32:09,000
All right. Well, thanks Eric for joining and appreciate your time.
