Latest / Becker’s Payer Issues Podcast / How Health Plans Can Build Scalable High Performance Infrastructure to Support Growing Complexity & Enable AI

Transcript

How Health Plans Can Build Scalable High Performance Infrastructure to Support Growing Complexity & Enable AI · Becker’s Payer Issues Podcast

• 0:00
  Hi, everyone. This is Lucas Voss with Becker's
• 0:02
  Healthcare. Thanks so much for tuning in to
• 0:04
  the Becker's Healthcare podcast series. It's great to
• 0:06
  have you. Today, we're talking about how health
• 0:08
  plans can build scalable high performance infrastructure to
• 0:11
  support
• 0:12
  growing
• 0:13
  complexity, enable AI, and stay ahead in the
• 0:16
  certainly rapidly evolving landscape that we're all facing
• 0:18
  right now. And joining me for today's discussion,
• 0:21
  very excited to have him is Rob Duffy.
• 0:22
  He's the chief technology officer at HealthEdge. Rob,
• 0:25
  thanks so much for being here today. It's
• 0:26
  great to have you. Yeah. Welcome, to Ben's
• 0:28
  point. Thank you very much. I really, really
• 0:30
  appreciate the opportunity to talk to you today.
• 0:32
  Yeah. Absolutely. I do wanna start with introductions
• 0:34
  here really quickly before we hop into the
• 0:36
  meat of this conversation.
• 0:37
  If you just wanna briefly introduce yourself, who
• 0:39
  you are, what you do at HealthEdge,
• 0:41
  and, what's going on with you.
• 0:44
  Amazing. So I'm Rob Duffy. I'm the chief
• 0:46
  technology officer at HealthEdge.
• 0:48
  My role is really to execute on it
• 0:50
  on the product road map in the most
• 0:52
  efficient and the high velocity way. And what
• 0:55
  we do is we, you know, as an
• 0:57
  engineering team, we really think about how we
• 0:59
  can leverage AI, how we can leverage cloud
• 1:01
  native infrastructure, how we can leverage all the
• 1:04
  things that are out there and available to
• 1:05
  us to really scale our operations and meet
• 1:08
  our customers
• 1:10
  through their scaling journey and make sure that
• 1:12
  they're getting the most leverage out of, the
• 1:14
  underlying technology and then also AI and anything
• 1:17
  that we build, from an AI perspective. So
• 1:20
  HealthEdge, for those those of you that aren't
• 1:22
  familiar with us, we are a, initially, we
• 1:25
  were a core admin platform, that does claims
• 1:27
  processing. We're a company called HealthEdge at the
• 1:30
  time.
• 1:30
  We acquired three other companies, GuidingCare, utilization management,
• 1:34
  and care management platform,
• 1:36
  Source, a,
• 1:37
  payment integrity platform,
• 1:39
  and then Wellframe, a member engagement platform.
• 1:41
  And then recently, we merged with Healthproof,
• 1:44
  who are a business process as a service,
• 1:46
  organization. And together, we have become HealthEdge's combined
• 1:49
  entity.
• 1:50
  And what that enables us to do is
• 1:53
  provide claims processing as a commodity to our
• 1:56
  customers. And what we can do because of
• 1:58
  our scale and because of our technology and
• 2:00
  because of our labor arbitrage,
• 2:02
  we can actually guarantee
• 2:03
  outcomes, guarantee the same claims processing outcomes
• 2:06
  for a a discounted rate,
• 2:09
  than the the ability for our our plans
• 2:12
  to actually process on their own. So, you
• 2:14
  know, on a per member per month basis,
• 2:16
  we are able to
• 2:18
  guarantee the same outcomes from claims processing to
• 2:20
  our customers,
• 2:22
  and do that at a discount.
• 2:23
  And the large portion of that, obviously, which
• 2:26
  we'll talk about in a little bit here
• 2:27
  is not only scale, but also,
• 2:29
  our ability to leverage AI in order to
• 2:32
  make bring more and more efficiency to the
• 2:34
  claims processing and claims operations. Yeah. I was
• 2:36
  gonna say we'll we'll walk through the way
• 2:38
  this works here in a little bit detail,
• 2:40
  but you mentioned something very important, which is
• 2:42
  high velocity.
• 2:44
  It's it's a high velocity market. Certainly, when
• 2:46
  we talk about, again, that rapidly evolving landscape,
• 2:49
  I'd love to touch on this a little
• 2:50
  bit more because, again, we know that health
• 2:52
  plans are dealing with a lot of pressure,
• 2:54
  right, growing membership, complex care models, and then
• 2:57
  you add regulatory demands onto that, I I
• 3:00
  think it's really crucial.
• 3:02
  From where you're sitting right now,
• 3:04
  what does it actually look like when a
• 3:06
  health plan's claims processing infrastructure
• 3:09
  starts
• 3:10
  to buckle under all of this weight that
• 3:13
  we're experiencing right now? I mean, you look
• 3:15
  at at any health plan right now, you
• 3:17
  know, medical loss ratio shooting through the roof
• 3:19
  either through, you know,
• 3:21
  very expensive drugs, more utilization,
• 3:24
  you know, people just getting, an aging population
• 3:27
  or even people getting smarter about their health
• 3:28
  care. I mean, the the impact of AI
• 3:30
  and chat GPT has meant that people are
• 3:32
  actually now going and sort of self diagnosing
• 3:34
  and seeing doctors where they where they wouldn't
• 3:36
  have seen before. So utilization is increasing. You
• 3:38
  know, cost of drugs is increasing. And then
• 3:40
  you have this interesting dynamic where on the
• 3:43
  revenue cycle management side of things,
• 3:46
  providers are using AI to code,
• 3:49
  their claims more effectively, and that's causing an
• 3:51
  increase in the in the the medical loss
• 3:53
  ratio. So you you think about that bucket
• 3:56
  of, you know, cost.
• 3:58
  There's really increasing
• 4:00
  in in terms of, the the MLR for
• 4:02
  our for our payers. And And then the
• 4:04
  other bucket of cost is the administrative loss
• 4:06
  ratio, which is like, how much does it
• 4:07
  just cost to own and operate the plan,
• 4:08
  process claims, do everything?
• 4:10
  And
• 4:11
  that bucket, I think, is just absolutely ripe
• 4:13
  for for disruption from AI tools, from automation,
• 4:16
  from scalable modern platforms.
• 4:18
  And that's really what we wanna bring to
• 4:19
  health care plans is a is a platform
• 4:21
  for scalability and a platform for innovation. And
• 4:24
  one where we can innovate on behalf of
• 4:25
  our customers and invent things using,
• 4:28
  all of the tools that are available us
• 4:29
  today to us today from an automation perspective,
• 4:31
  from an AI perspective, and just continually drive
• 4:34
  that ALR cost down so that our our
• 4:36
  plans, our customers
• 4:38
  can, you know, redivert some of those savings
• 4:40
  back to, you know, better health care outcomes,
• 4:42
  better customer, better, you know, patient experience,
• 4:45
  and really maintain their business and maintain their
• 4:47
  margins in the face of increasing it a
• 4:49
  lot.
• 4:50
  You mentioned innovation, which is certainly really crucial
• 4:52
  work from where you're sitting. You're innovating every
• 4:54
  single day. That's part of your job.
• 4:57
  You at HealthEdge and you certainly your team
• 4:59
  as well, you've reached an incredible milestone specifically
• 5:01
  with AWS, which most of our listeners are
• 5:04
  obviously familiar with, a big player in the
• 5:05
  space,
• 5:06
  in a single instance handling the equivalent of
• 5:08
  40,000,000 members. And, again, you cut that latency
• 5:12
  by more than half at that same time
• 5:14
  while also going through that process. Can you
• 5:16
  just walk us through what this meant for
• 5:19
  you?
• 5:20
  And, again, because you improved scale and speed
• 5:23
  at the same time,
• 5:24
  which is really, really hard to do.
• 5:27
  Yeah. And, you know, we made a decision
• 5:29
  to go all in with AWS,
• 5:31
  about eighteen months ago now. We said, hey.
• 5:33
  There's no point I was pursuing
• 5:35
  a multi cloud strategy. It's just diluting not
• 5:38
  only our purchasing power, but also the partnership
• 5:40
  opportunity we have with our cloud provider. And
• 5:42
  part of that was an acknowledgment that
• 5:45
  going all in with a with a single
• 5:47
  provider would also enable us to get really,
• 5:50
  really
• 5:52
  familiar with everything that would that provide that,
• 5:55
  cloud provider was able to offer us and
• 5:57
  really develop deep expertise and partnership with them
• 5:59
  in order to bring these outcomes to our
• 6:01
  customers. And we very early on realized that,
• 6:05
  you know, we've been pushing the boundaries of
• 6:07
  scalability
• 6:08
  using a physical hardware
• 6:10
  mindset. Right? And then we can we were
• 6:12
  previously in a private cloud and a data
• 6:14
  center virtualized there. And whenever we thought about
• 6:17
  scalability, it was like, you know, let's buy
• 6:18
  some more machines, long lead time, larger machines,
• 6:21
  whatever. Right? We we we would scale to
• 6:23
  the point where we could afford the infrastructure
• 6:26
  and and, you know, it's expensive infrastructure when
• 6:28
  it's in in a data center. So we
• 6:30
  would so we're thinking about it in a
• 6:32
  in a very non high performance compute way
• 6:34
  and then more of a sort of, like,
• 6:36
  horizontally scaling traditional data center way. When we
• 6:39
  moved to the cloud, all of a sudden
• 6:41
  we were able to take advantage of some
• 6:43
  new, you know, AWS silicon, the Graviton chips,
• 6:45
  the high performance compute instances, and we're able
• 6:48
  to dynamically scale them to meet the demands
• 6:51
  of our application and drive where we were
• 6:53
  able to scale to way beyond where we
• 6:56
  had previously been able to. And I think
• 6:58
  that's just a
• 6:59
  a a a a great indication of how
• 7:02
  deeply partnered we are with AWS and then
• 7:04
  also the the ability for that partnership to
• 7:06
  bring true scalability and and true outcomes for
• 7:09
  us from a technology perspective. And you mentioned
• 7:11
  the latency there. I mean, really,
• 7:13
  you know, AWS
• 7:15
  is a a
• 7:16
  has deep expertise
• 7:18
  in how to get their infrastructure to talk
• 7:21
  to each other and how to, you know,
• 7:24
  continuously fight against the speed of light
• 7:26
  to to to bring latency down between their
• 7:29
  their applications and between their services. And that
• 7:32
  really shows when you look at things like
• 7:33
  the network IO, the latency, the disk.
• 7:36
  You know, we had our our database administered
• 7:37
  getting a bit technical, but we had our
• 7:39
  database administered
• 7:40
  administrator,
• 7:41
  incredible lady called Manu.
• 7:43
  And she was, like, having a little bit
• 7:44
  concerned about the IO. And then after we
• 7:46
  routed, the the ability to write information to
• 7:48
  the disk, and then after we moved to
• 7:49
  the cloud, she was like, oh, wow. We're
• 7:51
  gonna be fine. This is amazing. It's it's
• 7:53
  much better than we're able to achieve. So
• 7:55
  I I think, you know, maintaining your own
• 7:57
  infrastructure,
• 7:58
  you can do well. I think moving to
• 8:01
  a a cloud hyperscaler and really leaning into
• 8:04
  one cloud hyperscaler, you can do amazing and
• 8:06
  great things with them because they will also
• 8:09
  invest in you, and you can invest in
• 8:10
  that partnership together.
• 8:12
  Applying this to
• 8:14
  the use case in and of itself, just
• 8:16
  as a quick follow-up here. Right?
• 8:18
  Why does the single instance that you just
• 8:20
  described make such a difference
• 8:23
  from a day to day operations perspective? What
• 8:25
  difference does it make for day to day
• 8:28
  operations?
• 8:29
  So traditionally,
• 8:30
  for a health plan, you would have different
• 8:33
  lines of businesses, maybe different regions on multiple
• 8:36
  different core admin platforms.
• 8:38
  So you have the the
• 8:40
  the same system, sometimes a different system, sometimes
• 8:42
  the same system, but a different version of
• 8:44
  the same system
• 8:45
  running in multiple places across across the organization.
• 8:49
  And for us, when we're trying to bring
• 8:50
  down the cost of operations and we're trying
• 8:52
  to commoditize claims processing and really sort of
• 8:54
  continue to lower the cost of operations,
• 8:58
  having multiple different systems and multiple different tools
• 9:01
  and multiple different instances
• 9:03
  really means that you can't get economies of
• 9:05
  scale or leverage from the investment that you
• 9:07
  have in that technology. And the same is
• 9:08
  true for for any plan. You know, if
• 9:10
  you look at a plan that, has been
• 9:12
  around for any length of time, there's gonna
• 9:14
  be many, many different systems, and they're gonna
• 9:15
  be thinking about this consolidation into one.
• 9:18
  When you can consolidate
• 9:20
  all these different things into one, it brings
• 9:22
  a tremendous amount of efficiency,
• 9:24
  and it brings a, you know, it it
• 9:26
  brings lower cost in the long run because
• 9:28
  you're not maintaining multiple different systems. You're not
• 9:31
  maintaining multiple different workforces that have to integrate
• 9:33
  with these different systems. You can get a
• 9:35
  single view of your data. You can start
• 9:37
  to talk about events across these, diff, you
• 9:39
  know, the different customer life cycles that are
• 9:42
  triggering things on your data platform. You don't
• 9:44
  have to worry about integrations. Like, each of
• 9:46
  these systems has to be integrated with up
• 9:48
  to 120
• 9:49
  different inside internal tools. Right? So now all
• 9:51
  of a sudden, if you have 10 systems,
• 9:53
  you have 1,200
• 9:54
  integrations that you have to maintain. You're shuffling
• 9:56
  data everywhere from different things. So it's really
• 9:59
  a a a grand simplification
• 10:01
  of the internal architecture
• 10:03
  of the health plan
• 10:04
  and the and the claims,
• 10:07
  operating system. And that enables you to do
• 10:09
  much more interesting things with AI, with data,
• 10:12
  with your visibility into, you know, how your
• 10:15
  business is performing with the latest BI tools
• 10:17
  that you can query with natural language. And
• 10:19
  it removes this sort of diaspora of systems
• 10:22
  that are very hard to integrate and and
• 10:24
  gather intelligence out of and and provides you
• 10:27
  really with an edge to continuously optimize your
• 10:29
  business and continuously,
• 10:31
  you know, reduce the cost of outcomes.
• 10:34
  A lot of leaders are pointing to
• 10:36
  that piece specifically, the integration
• 10:39
  part and the infrastructure piece as a crucial
• 10:41
  enabler for health care as we move into
• 10:43
  2027
• 10:44
  and beyond. If that's not the case,
• 10:46
  then organizations are going to have an issue
• 10:49
  moving forward. And I wanna stick with that
• 10:50
  theme a little bit because, again, I feel
• 10:52
  like especially with AI, and you've talked about
• 10:55
  the AI native enterprise, so to speak. Right?
• 10:58
  How does that infrastructure work then lay the
• 11:00
  groundwork
• 11:01
  for organizations
• 11:03
  to have
• 11:04
  AI driven automation in place?
• 11:06
  And, again, just touching back on on what
• 11:08
  we've just talked about, what's the risk of
• 11:10
  Healthfence that just try to bolt on, and
• 11:14
  they're not establishing that infrastructure that we've just
• 11:17
  talked about.
• 11:18
  Yeah. There's a lot of really great demos
• 11:20
  out there for, you know, AI startups that
• 11:23
  will bolt something on top of the system
• 11:25
  of record. You know? And they'll they'll get
• 11:26
  to a really good, yeah, you know, demo
• 11:30
  maybe in a couple of weeks, and they'll
• 11:31
  show you it. And I'll be like, this
• 11:32
  is proof concept, and you'll get really excited
• 11:34
  about the fact that that that this could
• 11:35
  actually scale out across, you know, the entire
• 11:38
  the the entire set of use cases that
• 11:40
  you want to deploy this tool in. And
• 11:42
  the reality is
• 11:43
  it that it's not gonna it's not gonna
• 11:45
  scale out because, ultimately, you have the system
• 11:47
  of record. And in order for you to
• 11:49
  expand all the use cases, you would have
• 11:51
  to, you know, take the data that's in
• 11:53
  the system of record and march out. And
• 11:54
  it's like it it becomes it becomes a
• 11:56
  nightmare anything beyond the first one or two
• 11:58
  use cases. And, you know, we believe that
• 12:00
  by deeply integrating AI into our platform and
• 12:03
  by providing a platform that is
• 12:06
  equipping AI tooling on top of the system
• 12:08
  record and inside the system record, you can
• 12:10
  get much
• 12:11
  more efficiency out of that, and you can
• 12:13
  get a much higher return on your investment
• 12:15
  from AI. And you can do it a
• 12:16
  lot faster than if you're just still marshaling
• 12:19
  all these different tools and you're trying to,
• 12:21
  you know, add bolt ons on top of
• 12:23
  the system of record. So our belief is
• 12:25
  that we can move incredibly quickly to your
• 12:27
  point earlier on about velocity
• 12:29
  in deploying AI capabilities
• 12:31
  to, you know, reduce the amount of dependence
• 12:34
  we have on on humans and human effort
• 12:36
  clicking through UIs and and moving information left
• 12:39
  and right, then we can very, very rapidly
• 12:42
  bring a truly integrated
• 12:44
  AI toolkit that is deeply connected to the
• 12:47
  system of record. And again, that that single,
• 12:51
  platform and single instance enables us to do
• 12:53
  that. We can build, you know, these these
• 12:55
  tools that enable AIs to consume information, enable
• 12:58
  AI agents to act within the within the
• 13:00
  system of record, and work in in concert
• 13:03
  with each other and with all of the
• 13:05
  other systems in the ecosystem
• 13:06
  rather than being something that's sort of bolted
• 13:08
  on from the outside.
• 13:10
  Realistically,
• 13:13
  how much time do we have then to
• 13:16
  rethink these legacy payer platforms before
• 13:20
  we hit that hard ceiling. Right? And from
• 13:23
  your perspective, from the conversations you were having
• 13:25
  with health system leaders across the country,
• 13:28
  what are the questions
• 13:29
  that leaders need to ask
• 13:31
  right now that they need to ask their
• 13:34
  technology partners to avoid running into some of
• 13:36
  the issues that we've highlighted?
• 13:38
  You know, I might I tie it back
• 13:40
  to the the performance of the company. Right?
• 13:42
  Like, when you think about and if everything
• 13:43
  that we do in technology,
• 13:45
  it's really
• 13:46
  about reducing the cost of operations. Right? Like,
• 13:48
  it's how we continually driving down the operating
• 13:51
  costs. And if I was a, you know,
• 13:53
  a business leader at a plan,
• 13:56
  the real question I would be asking is,
• 13:58
  are we going to be able to accelerate
• 14:01
  the savings that we want to realize
• 14:04
  in our administrative
• 14:05
  loss ratio
• 14:07
  fast enough
• 14:09
  to keep up with the competition?
• 14:11
  And are we gonna be able to invest
• 14:13
  enough in it
• 14:14
  to be able to deliver true outcomes and
• 14:17
  true value,
• 14:18
  or should we partner with someone who's gonna
• 14:20
  do that and who's gonna guarantee that that
• 14:22
  cost is gonna come back and who's gonna
• 14:24
  continue to invest with the best and brightest
• 14:26
  and get, you know, huge leverage out of
• 14:28
  partnerships with AWS much broader than individual plans
• 14:31
  would be able to and and hitch onto
• 14:33
  that wagon. And I think, like, that that's
• 14:34
  the way I'd be thinking about it. So
• 14:35
  can I internally develop the skill set and
• 14:38
  rapidly ship and, you know, put all the
• 14:40
  pieces in place
• 14:42
  to guarantee that my business is gonna stay
• 14:44
  healthy and that my administrative spot for ratio
• 14:46
  is gonna continue to decline,
• 14:47
  or
• 14:48
  why not partner with someone who's gonna guarantee
• 14:51
  that? And then I can focus on things
• 14:52
  like member engagement,
• 14:54
  health care outcomes,
• 14:55
  marketing,
• 14:56
  you know, taking care of the population.
• 14:59
  Because if you do that, then you're not
• 15:01
  going to run out of runway in 2026
• 15:04
  and 2027.
• 15:05
  We're just very crucial, especially in health care.
• 15:08
  Rob, it's so great to have you on.
• 15:09
  I just wanna give the floor to you
• 15:10
  too. Is there anything else that you wanna
• 15:12
  touch on? I know we've we've hit on
• 15:13
  a lot of numbers here.
• 15:15
  Again, I feel like this is about confidence
• 15:17
  for health plans, but just wanna hear any
• 15:19
  final thoughts that you have for our listeners.
• 15:22
  No. I I think, you know, it's always
• 15:24
  a pleasure to come on the on the
• 15:25
  Beckett's podcast and and talk about the the
• 15:28
  future and the future of AI. And it's
• 15:30
  a very exciting but also scary time. You
• 15:33
  know, there's excitement in the possibility,
• 15:35
  but there is the fear of not moving
• 15:37
  fast enough. And, you know, what we do
• 15:39
  is help our plans and help our customers
• 15:42
  feel like they're moving fast enough. And I
• 15:44
  think that brings an enormous amount of excitement
• 15:46
  and energy to to our customers and ourselves.
• 15:49
  Velocity
• 15:49
  is the word of the day for us
• 15:51
  today. Rob, so great to have you. Thanks
• 15:53
  so much for your time. Thank you. Take
• 15:55
  care.
• 15:56
  We also want to thank our podcast sponsor,
• 15:58
  HealthEdge. You can tune in to more podcasts
• 16:00
  from Becker's Healthcare by visiting our podcast page
• 16:02
  at beckershospitalreview.com.