Latest / Becker’s Healthcare Podcast / Why Provider Data and AI Are Critical to Patient Access

Transcript

Why Provider Data and AI Are Critical to Patient Access · Becker’s Healthcare 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
• 0:07
  have you. I'm excited today to be joined
• 0:09
  by Charlie Lloyd and doctor James Whitfield for
• 0:12
  our episode
• 0:13
  on how robust provider data and AI are
• 0:15
  becoming a key part of the infrastructure
• 0:18
  for patient
• 0:20
  access. Charlie is the CEO and cofounder of
• 0:22
  Axual. He's redefining
• 0:24
  the foundational infrastructure of the health care workforce.
• 0:26
  Under his leadership, Axual has moved beyond simple
• 0:29
  verification to become a strategic
• 0:31
  data partner for the nation's leading health systems
• 0:34
  by leveraging real time provided data. Charlie and
• 0:36
  his team are transforming
• 0:38
  how organizations
• 0:39
  source, recruit,
• 0:40
  and deploy clinicians. They're focused on solving the
• 0:43
  fragmentation of the past to build a modern
• 0:46
  data driven ecosystem that accelerates onboarding and drives
• 0:49
  the long term clinical and operational transformation required
• 0:52
  in today's market. Charlie, welcome.
• 0:55
  Thank you, Lucas. Good to be here. And
• 0:57
  we also have doctor Whitfel on. He's the
• 0:59
  senior vice president of strategic partnerships and chief
• 1:02
  transformation officer at HonorHealth.
• 1:04
  He brings a unique blend of both clinical
• 1:06
  expertise and executive leadership with a strong focus
• 1:08
  on driving innovation, scaling strategic partnerships, and advancing
• 1:11
  transformation
• 1:12
  across complex health systems. His work centers on
• 1:15
  aligning technology operations and clinical strategy
• 1:18
  to improve outcomes and create a more sustainable
• 1:21
  model for
• 1:22
  care. Doctor Whitfield, also so great to have
• 1:24
  you. Thanks for being here. Lucas, awesome to
• 1:27
  be here. Really glad. I wanna hop right
• 1:29
  into our conversation. We certainly have lots to
• 1:31
  discuss. And, doctor Whitfield, I'll start off with
• 1:33
  you, and this is a a little bit
• 1:35
  of an of an awareness piece here too
• 1:37
  on the first question.
• 1:38
  Most people think that provided data governance is
• 1:41
  basically,
• 1:42
  well, just keeping a list of doctors up
• 1:45
  to date. It should be pretty simple. We
• 1:47
  have a list. We update it regularly. Done.
• 1:50
  Can you explain
• 1:51
  why that's off and what happens if an
• 1:54
  organization
• 1:55
  views it just like a list that needs
• 1:57
  to be updated?
• 1:59
  It's first of all, Lucas, this is a
• 2:00
  great place to start because it does. It
• 2:02
  seems like it should be pretty simple.
• 2:04
  Not unlike the old saying, baseball is a
• 2:06
  simple game. You just throw the ball. You
• 2:08
  catch the ball. You hit the ball. But
• 2:10
  it when it comes to our physician advanced
• 2:13
  practice provider data, that is the lifeblood of
• 2:16
  it's the information that drives
• 2:18
  so many different parts of the health system.
• 2:21
  And I think what people don't realize is
• 2:23
  that there are lots of different systems keeping
• 2:25
  those lists,
• 2:26
  but they need to keep them for different
• 2:28
  reasons, and they have different views of what
• 2:30
  the lists mean. So for exam
• 2:33
  an example that I always like to give
• 2:35
  is that imagine that if I have a,
• 2:39
  physician that does obstetrics and gynecology, right, that's
• 2:42
  a type of medicine,
• 2:44
  treating women, delivering babies,
• 2:47
  that person may be licensed in a specific
• 2:49
  way than and and credentialed, say, from a
• 2:51
  board certification.
• 2:53
  That's a piece of data that's really important
• 2:55
  because if if you have rules that that
• 2:57
  person can't practice in your environment without that
• 2:59
  certification,
• 3:00
  and you need to know that. However, when
• 3:02
  a patient is looking for somebody
• 3:05
  to care for them, they wanna know, does
• 3:07
  that pay does that doctor deliver babies?
• 3:09
  That's a different question. Obstetrics is still in
• 3:12
  the word, but, actually, now you're talking about
• 3:14
  a different piece of data that may be
• 3:16
  held in a different system.
• 3:18
  And this repeats itself in the electronic health
• 3:20
  care record, in the system,
• 3:22
  the contact management system, in the phone system,
• 3:25
  in all of these directories. And then, oh,
• 3:26
  by the way, you multiply that by 5,000
• 3:29
  or 10,000 physicians,
• 3:31
  and that information is changing sometimes
• 3:34
  daily for different individuals.
• 3:36
  Who's in charge?
• 3:38
  Who's managing it? What's the authority? It can
• 3:40
  get out of hand really quickly.
• 3:43
  Yeah. There's so many data points that need
• 3:45
  to be cross tracked, etcetera, which is so
• 3:47
  crucial. And, again, it creates mistakes that you
• 3:49
  just outlined. Now, Charlie, we're bringing
• 3:52
  AI into the mix right now, and I
• 3:54
  know in a previous conversation, we've briefly touched
• 3:56
  on on AI hallucinations,
• 3:58
  and they still exist. They're still there.
• 4:01
  But, again, you're also talking about something that's
• 4:03
  already happening in scheduling and referral tools before
• 4:06
  AI even enters the room, really.
• 4:08
  What does all of this look like for
• 4:10
  a real patient?
• 4:13
  Yeah, Lucas. I think, you know, I like
• 4:15
  to sort of start at the basis of
• 4:17
  data. And if you look at provider data,
• 4:20
  it is that directory. And and part of
• 4:23
  the issue today
• 4:24
  is that provider directory inaccuracy
• 4:27
  is really creating a problem in The US,
• 4:31
  not just from the payer side, but on
• 4:33
  the health system and provider side as well.
• 4:36
  And and and the culmination of that in
• 4:38
  many ways is you've got inaccurate and outdated
• 4:41
  provider data that's sort of making the supply
• 4:43
  and demand problem almost worse than it already
• 4:46
  is. And so case in point, matching a
• 4:49
  patient to a physician requires a lot of
• 4:52
  up to date information. It's really important to
• 4:54
  have that right. It includes things like when
• 4:57
  and where that provider actually practices,
• 5:00
  what the patient profile
• 5:02
  of that provider is, and what conditions they
• 5:05
  treat
• 5:06
  of those patients.
• 5:08
  There's also specialty and specialization.
• 5:10
  Sometimes that gets that gets missed.
• 5:13
  And to doctor Whitfield's point here earlier,
• 5:16
  you've got in some cases, you still don't
• 5:18
  need to look at that from two different
• 5:19
  angles. There's the medical terminology and then there's
• 5:21
  the consumer terminology.
• 5:23
  But nonetheless,
• 5:24
  if you think about specialty and specialization,
• 5:27
  specialty might be carotid thoracic surgery,
• 5:30
  but specialization
• 5:31
  means that they do mitral valve replacements versus
• 5:33
  CABG. Orthopedic
• 5:34
  specialty
• 5:36
  maybe that they do a specialization, maybe that
• 5:39
  they do hands versus feet or whatever that
• 5:41
  might be. And those nuances are really important.
• 5:44
  Other information that's really
• 5:46
  critical when you're making decisions about matching a
• 5:48
  patient or provider together are things like, are
• 5:51
  they what is their availability look like relative
• 5:53
  to the utilization,
• 5:55
  not just the schedule?
• 5:56
  What's their skill level? That's something really important
• 5:59
  in the referral process.
• 6:00
  But when this information isn't applied or becomes
• 6:03
  stale, you've got situations where there's mismatched appointments,
• 6:08
  there's referral opportunities that are lost or missed
• 6:10
  or poorly optimized.
• 6:12
  There's just overall waste. And so as we
• 6:14
  think about this, the root of the problem
• 6:18
  is that most directory data is created by
• 6:21
  people rather than derived by real world data
• 6:25
  with human oversight.
• 6:27
  And and that status quo in provider
• 6:30
  data upkeep
• 6:31
  is really something that's a big challenge across
• 6:34
  the space today.
• 6:35
  Director data that's entered by providers or worse
• 6:38
  yet by people once or twice removed from
• 6:40
  the provider it's gonna be inherently biased. It's
• 6:44
  gonna miss details.
• 6:45
  It's gonna be inadequate in many cases for
• 6:48
  a really good consumer search, let alone AI.
• 6:51
  And so without that automation in real world
• 6:54
  data, administrative staffs have a tough time keeping
• 6:56
  up. But if we think about AI,
• 6:59
  if you put AI on top of bad
• 7:01
  data,
• 7:02
  it's gonna come up with some bad suggestions.
• 7:05
  So if we don't want AI to hallucinate,
• 7:08
  step one is
• 7:09
  don't feed it the data equivalent to peyote.
• 7:12
  And so, you know, we just sort of
• 7:13
  look at it that way, and, like, you've
• 7:15
  gotta get that data right to the in
• 7:17
  the very beginning.
• 7:18
  Yeah.
• 7:19
  Now, doctor Whitfield, we touched on this a
• 7:21
  little bit because, again, when we're getting this
• 7:22
  data right, we talk about data governance, etcetera,
• 7:25
  but it also touches so many different elements
• 7:27
  of an organization. Right? We talk revenue cycle,
• 7:29
  medical staff, EMR,
• 7:31
  even marketing in a lot of cases that
• 7:33
  needs data to be able to do this.
• 7:34
  Right?
• 7:35
  Who owns all of this inside a health
• 7:38
  system? Who should own it? What's that ownership
• 7:40
  level? And, again, I think the more important
• 7:42
  question here too, what goes wrong when nobody
• 7:45
  can answer that, when nobody can tell, okay.
• 7:48
  I don't know who owns this.
• 7:51
  Well, first of all, it is a total
• 7:52
  recipe for disaster when you ended that situation
• 7:54
  where nobody and and by the way, it's
• 7:56
  easy to fall into that situation because you've
• 7:57
  got all these different stakeholders that have all
• 8:00
  these different perspectives.
• 8:02
  But for the most part, they're busy. They're
• 8:04
  worried about their stakeholder's
• 8:05
  dump. They don't they can't worry about the
• 8:07
  whole picture. In our I'll tell you, in
• 8:09
  our journey, in our organization,
• 8:11
  we ended up saying that we think that
• 8:14
  the data quality around our physicians and advanced
• 8:16
  practice providers is so important
• 8:19
  that we put the the that was the
• 8:21
  first element that we brought into our analytics
• 8:23
  team, and we're not and the head of
• 8:24
  our analytics department said, alright. I'm gonna chair
• 8:27
  the provider
• 8:28
  data governance group because
• 8:30
  there is no one person who owns all
• 8:32
  of this work. And when we think about
• 8:34
  the most important
• 8:36
  data elements for an organization, there's something called
• 8:39
  data stewardship. Like, who who owns this piece
• 8:41
  of data? Who's responsible for monitoring its quality?
• 8:45
  And who convenes, you know, the work to
• 8:47
  make sure that it's done correct? We knew
• 8:49
  that we needed to do that. This was
• 8:50
  the first stake in the ground. We said,
• 8:52
  provide the physician data or provider data. That's
• 8:54
  where we're gonna start.
• 8:57
  Charlie, I'm assuming you have seen this go
• 8:59
  wrong.
• 9:01
  Same question for you. Right? Who should own
• 9:03
  this inside a health system? Who owns it?
• 9:05
  And again, what goes wrong if that's not
• 9:07
  the case?
• 9:08
  Yeah. It's a good question. I mean, it's
• 9:10
  almost like a hot potato to some degree
• 9:12
  for a lot of organizations
• 9:14
  because it it's a lot to take it
• 9:16
  on.
• 9:17
  And it's been this wide surface area problem,
• 9:20
  meaning, like, so many so many cooks in
• 9:23
  the kitchen, so many people have been trying
• 9:25
  to address it. And, you know, to again,
• 9:27
  to doctor Whitfield's point earlier, depending on the
• 9:30
  use of that data, you've got different people
• 9:33
  updating it, manipulating it, adding to it, etcetera.
• 9:36
  And, again, I think with all those people
• 9:39
  touching the data, you've got just a lot
• 9:41
  of administrative costs. You've got duplication.
• 9:44
  And that duplication,
• 9:46
  we've seen this in many cases. It's really
• 9:48
  not accounted for.
• 9:50
  Yes. Sort of built into the budgets in
• 9:52
  different departments, but and they're all they've got
• 9:55
  either a so in some cases, one or
• 9:57
  two full time people working on it, or
• 9:59
  maybe it's a half a person, or maybe
• 10:01
  they're subbing it out, and they've got other
• 10:04
  folks trying to gather the data from many
• 10:06
  different sources. But there's just a lot of
• 10:08
  redundancy and cost in that that obviously isn't
• 10:11
  a good thing for health care right now
• 10:13
  right now. But moreover,
• 10:15
  you know, I've also seen if you think
• 10:17
  about where it goes wrong
• 10:19
  is you've got so much investment.
• 10:21
  And you look at other industries outside of
• 10:23
  health care, and when they add technology, efficiency
• 10:26
  improves and cost go down.
• 10:28
  We haven't been able to do that to
• 10:30
  the extent that health care deserves it. It's
• 10:32
  just in general. Right? And I think that,
• 10:34
  you know, we look at that and say,
• 10:35
  like, how can we begin to fix that?
• 10:38
  And, you know, you've got situations where
• 10:41
  you've got a very robust EHR infrastructure that
• 10:44
  that organizations, in some cases, have spent tens
• 10:47
  of millions or hundreds of millions of dollars
• 10:48
  to put in place,
• 10:50
  but they don't have the data to power
• 10:52
  all of those capabilities.
• 10:54
  It can it can be really limiting in
• 10:57
  terms of return investment that we're all looking
• 10:59
  for. I When you look at a lot
• 11:01
  of our customers are epic. Right? And it's
• 11:02
  like, to be an epic cosmos,
• 11:04
  you have to bring good data
• 11:07
  to use, some of these tools like teamwork
• 11:09
  and chairs and CRM and other things. If
• 11:12
  you really wanna get the good value out
• 11:14
  of that, you've gotta bring your data a
• 11:17
  game. And a lot of this has to
• 11:18
  do with just figuring out where in the
• 11:20
  organization that needs to sit. In my opinion,
• 11:24
  I think a lot of this really needs
• 11:26
  to be in the wheelhouse of the CMIO
• 11:28
  or CIO or in some cases, chief data
• 11:30
  officer.
• 11:32
  Yes. The CMO
• 11:33
  absolutely cares about. The CFO cares about it,
• 11:36
  but those are folks that really understand the
• 11:38
  data. And so I think we're beginning to
• 11:40
  see some of that coalesce
• 11:43
  around those types of teams because they've worked
• 11:45
  within the past, and they understand the implications
• 11:48
  from an intersystem standpoint of getting this right.
• 11:53
  You've both
• 11:54
  now touched multiple times on the importance of
• 11:57
  getting the data right in any instance. And,
• 11:59
  again, we know that clinician data
• 12:02
  is really
• 12:03
  the fuel for a lot of this. I
• 12:05
  wanna turn this around a little bit and
• 12:07
  talk about something that's happening across health care,
• 12:09
  which is the race for
• 12:12
  agentic AI, the agentic layer in health care,
• 12:16
  so to speak. Again, if that clinician data
• 12:19
  is the fuel,
• 12:21
  what happens to those AI agents
• 12:24
  when that underlying data is wrong, outdated, or
• 12:27
  just incomplete that we've just talked about here
• 12:30
  in this conversation? Charlie, I'll I'll start off
• 12:32
  with
• 12:33
  you. Yeah. Lucas, you know, I think, again,
• 12:35
  the same things. If it if if the
• 12:37
  underlying data isn't there, you're having an agentic
• 12:40
  conversation
• 12:41
  or conversational
• 12:42
  AI around
• 12:44
  finding the right physician,
• 12:45
  matching a patient phenotype with a provider phenotype.
• 12:49
  If you don't have the right information,
• 12:51
  you can't expect it to do much,
• 12:54
  other than maybe get it wrong. And so
• 12:57
  I think it it you know, that that
• 12:59
  is obviously really important, but I also wanna
• 13:01
  look at what what the potential and the
• 13:04
  value
• 13:05
  exists there.
• 13:06
  You know, to be able to not only
• 13:08
  provide a consumer interface that is conversational,
• 13:12
  that a lot of, these agentic technologies are
• 13:14
  providing, but also learning from those conversations
• 13:18
  what patients need as consumers,
• 13:21
  as caregivers,
• 13:23
  as advocates,
• 13:24
  because that's a lot of where that's happening.
• 13:26
  You know, that that is data that's really
• 13:28
  important to capture along the along the way
• 13:31
  as well.
• 13:32
  I think, you know, agentic conversations
• 13:34
  are gonna give us the ability to take
• 13:36
  these fairly
• 13:38
  you know, a lot of the organizations we
• 13:39
  work with have invested in a, a a
• 13:42
  ton of money and effort
• 13:44
  into
• 13:45
  into models that say, if if this x,
• 13:48
  y, and z exists route to this type
• 13:50
  of clinician or that other type of clinician,
• 13:52
  All that stuff is really important, but being
• 13:55
  able to layer AI on top of that
• 13:56
  and see how it operates, I think, is
• 13:58
  something that's gonna be,
• 14:00
  really interesting. And to some degree, it's how
• 14:02
  consumers are beginning to think about how they
• 14:03
  interact and have questions today. I mean, a
• 14:06
  lot of this, we can just learn from
• 14:08
  how,
• 14:09
  search and and how they're searching and how
• 14:11
  they're behaving. So you kinda need to be
• 14:13
  be able to bring these two worlds together,
• 14:15
  and I think having, again, a good data
• 14:17
  substrate to do that is gonna be really
• 14:18
  important.
• 14:19
  Yeah. Doctor Woodfill, same question for you. When
• 14:21
  we talk about agentic AI, how are you
• 14:23
  viewing that? And, again, specifically related to the
• 14:25
  data in and of itself, how important is
• 14:27
  that when we're thinking about that a AI
• 14:29
  agent concept?
• 14:31
  Well, I'll use a real world example. We
• 14:33
  are now deploying an agent that helps with
• 14:36
  patients who are coming in for a procedure.
• 14:39
  Say they're gonna have an operation in a
• 14:40
  week. There's a lot that you have to
• 14:42
  do in order to to be ready for
• 14:44
  that. And the last thing we wanna have
• 14:46
  happen is that somebody shows up on the
• 14:48
  day of their procedure and we miss something,
• 14:50
  right, and we have to cancel. That's terrible
• 14:51
  for the patient. It's not good for the
• 14:53
  health system.
• 14:54
  So one of the things that sounds very
• 14:56
  pedestrian, but the difficulty of getting medical records
• 15:00
  from lots of different places oftentimes
• 15:02
  involves
• 15:03
  phone calls to physician offices
• 15:06
  to ask for medical records that haven't been
• 15:08
  set yet. I mean, it's terrible that we
• 15:09
  live in a world like that, but the
• 15:11
  reality is is since we have an agent
• 15:13
  that is starting to do that work, just
• 15:15
  it's a back office function. It's gotta know
• 15:18
  the right physician
• 15:19
  with the right back office staff, with the
• 15:21
  right phone numbers, and the right context of
• 15:24
  how to get that work. And because we
• 15:25
  can start to do it at scale, and
• 15:27
  I think that's the part around the agentic
• 15:29
  piece. If you don't get it right, you
• 15:31
  can do a lot of,
• 15:33
  at worst, at at best, like, embarrassing
• 15:35
  kind of embarrass yourself at scale really, really
• 15:38
  quickly. And so as we think about
• 15:40
  those agents that are starting to reach out
• 15:42
  to other human beings, other offices, eventually, probably
• 15:45
  to other agents, like, making sure that they're
• 15:47
  going to the right place at the right
• 15:48
  time for the right information and getting that
• 15:50
  all straight is really critical for that agent
• 15:52
  to do its workflow.
• 15:54
  Yeah.
• 15:55
  And, doctor Whitfield, as we close out our
• 15:57
  conversation here, again, I wanna bring this back
• 15:59
  to the patient because the patient is at
• 16:00
  the center of all this. It should enable
• 16:02
  better patient care, better patient outcomes at the
• 16:05
  end of the day. That's the crucial piece
• 16:07
  to this. Right? And we know that, again,
• 16:09
  especially in areas where there are, you know,
• 16:11
  maybe physician shortages or there's not as much
• 16:13
  access. Right?
• 16:14
  Patients tend to ask for a specific doctor.
• 16:17
  That person might not be available. And then,
• 16:19
  again, to your point,
• 16:20
  they won't come back. It has an impact
• 16:22
  on the organization beyond just the clinical piece
• 16:25
  of it. It's a brand. It's reputation.
• 16:27
  It it's financial.
• 16:29
  How does smarter data governance help a system
• 16:32
  then offer a real alternative, right, without that
• 16:34
  patient feeling like, okay. I just got handed
• 16:37
  off again to somebody else that I don't
• 16:39
  know that well, that was just happened to
• 16:40
  be free.
• 16:42
  Well, Lucas, I I still see patients, and
• 16:45
  I, you know, I'm haunted by the following
• 16:47
  scenario, which happens too often. I've let's say
• 16:50
  it's a long way to get ahold of
• 16:51
  a specific kind of neurologist because I've got
• 16:53
  a patient I'm really worried about. And if
• 16:55
  that patient has to wait thirty days to
• 16:57
  get to that appointment and then they show
• 16:59
  up at the neurologist and it's the like,
• 17:01
  it turns out that that neurologist doesn't treat
• 17:03
  that condition
• 17:04
  Mhmm. It's not I I've just wasted thirty
• 17:07
  days of that patient's life. Right? And now
• 17:09
  I have to start that process all over
• 17:11
  again. I feel terrible for the patient.
• 17:13
  It's embarrassing for us as a health system.
• 17:15
  It's bad for our brand. And by the
• 17:16
  way, I wasted the time of that other
• 17:18
  neurologist who now can't see somebody else. Right?
• 17:20
  That is
• 17:21
  that happens, and that's because we don't have
• 17:24
  the right information
• 17:25
  around our physicians so that we can make
• 17:27
  that right choice.
• 17:29
  You asked a a broad piece, but I
• 17:31
  I will just shrink it down to that
• 17:33
  very tangible
• 17:34
  sort of patient story. And that's what drives
• 17:36
  us as an organization to make sure that
• 17:39
  our physician data, our advanced practice provider data
• 17:41
  is kept as accurate as possible so we
• 17:44
  can really drive more efficiency, but get it
• 17:46
  right. Because if we don't get it right,
• 17:47
  it doesn't matter how efficient we are.
• 17:50
  Charlie, really quickly before we close it, just
• 17:52
  to bring it back to what we started
• 17:53
  our conversation out with, should this be a
• 17:56
  priority for every health system across the country
• 17:59
  in terms of looking not just looking at
• 18:01
  these things as lists and and looking at
• 18:04
  data governance as a priority
• 18:06
  that needs to be looked
• 18:08
  at. Yeah. Lucas, you know, I I would
• 18:09
  end by saying this is really about maximizing
• 18:13
  access.
• 18:14
  We have a supply and demand gap in
• 18:16
  The United States relative to care, and it's
• 18:19
  just gonna get harder and harder for a
• 18:21
  lot of different reasons. And if we can
• 18:23
  use technology and good data to make better
• 18:26
  matching decisions like doctor Whitfield just said,
• 18:29
  that goes a long ways to stretching without
• 18:32
  adding burden to clinicians.
• 18:34
  That goes a long ways to improving some
• 18:36
  capacity and improving access. And it's a national
• 18:39
  imperative. We've gotta figure this stuff out.
• 18:42
  Trust is incredibly important,
• 18:44
  and we have to be thoughtful and somewhat
• 18:46
  conservative about how we outreach and and interact
• 18:49
  with patients.
• 18:50
  But it it's just something that I think
• 18:52
  we all need to do, and and it's
• 18:54
  great to see forward thinking organizations like Honor
• 18:57
  be behind this kind of thing.
• 18:59
  Well, doctor Whitfield, Charlie, so great to have
• 19:02
  you both on. Thank you so much for
• 19:03
  being here for your time and insights, and
• 19:05
  we also want to thank our podcast sponsor,
• 19:07
  ACTUAL, for bringing us together for this important
• 19:10
  conversation. As you just heard, a national imperative,
• 19:13
  and you could tune in for more podcasts
• 19:14
  from Becker's Healthcare by visiting our podcast page
• 19:17
  at beckershospitalreview.com.