Latest / Becker’s Payer Issues Podcast / How AI is Transforming the Payer Strategy

Transcript

How AI is Transforming the Payer Strategy · Becker’s Payer Issues Podcast

• 0:00
  Hi, everyone. This is Brian Zerman with Becker's
• 0:02
  Healthcare. Thank you so much for tuning into
• 0:04
  the Becker's Healthcare podcast.
• 0:06
  Today, we're going to talk about how AI
• 0:08
  is transforming the payer strategy.
• 0:10
  Joining me for today's discussion live in Chicago
• 0:12
  at Becker's Payer Issues Roundtable is doctor Mira
• 0:15
  Atkins, chief medical officer at Lyric. Doctor Erica
• 0:18
  Atkins, thank you so much for being here.
• 0:20
  Thanks for having me. So to get us
• 0:22
  started here, can you just share a bit
• 0:24
  more about yourself, the work you're doing at
• 0:26
  Lyric, just so listeners can sort of appreciate
• 0:28
  where you're coming from as we move through
• 0:29
  the conversation today? Sure. So I'm actually a
• 0:33
  board certified OB GYN, and I practiced for
• 0:36
  several years prior to transitioning out of practice.
• 0:39
  And I've been in the payer space for
• 0:41
  the last ten years, but recently joined Lyric
• 0:44
  as their chief medical officer,
• 0:46
  where my focus is really to scale clinically
• 0:48
  informed AI for payment accuracy.
• 0:51
  And we're really moving towards prepayment intelligent that
• 0:54
  just helps to preserve the provider trust.
• 0:58
  Excellent. I appreciate that background. So here at
• 1:00
  at at Becker's Parish's roundtable,
• 1:03
  you've discussed with other leaders how AI and
• 1:04
  advanced analytics are are helping payers move from
• 1:07
  reactive to proactive care and payment models.
• 1:10
  So from your perspective, how is the shift
• 1:12
  showing up in areas like payment integrity and
• 1:14
  clinical validation, which you just flagged as really
• 1:16
  important to the to the work you're doing?
• 1:18
  And, really, I guess, in that context, if
• 1:21
  you could
• 1:22
  unpack what the word proactive actually means. I
• 1:24
  think we hear that word a lot, but
• 1:26
  put some,
• 1:27
  a concrete definition for folks out there. Sure.
• 1:29
  So for decades, you know, claims review has
• 1:32
  been really a reactive
• 1:34
  adversarial
• 1:35
  exercise
• 1:36
  where
• 1:37
  we are looking at either post payment and
• 1:40
  audits, and it's just been a really
• 1:43
  blunt force.
• 1:44
  I think the future is really we're moving
• 1:47
  towards the left where clinical AI provides a
• 1:50
  lot more transparency,
• 1:51
  and it drives more appropriate payments,
• 1:55
  where it reduces provider friction, and it really
• 1:58
  focuses
• 1:59
  people on the outliers. Those cases that really
• 2:03
  matter where you're getting the the high hit
• 2:05
  rates, I guess.
• 2:07
  Yeah. And that moving away from that adversarial
• 2:09
  sort of nature,
• 2:11
  do you think that AI
• 2:13
  is really an opportunity for that finally to
• 2:15
  happen? Do you really think that this technology
• 2:17
  is kind of might be the the the
• 2:19
  solution to some of the that sort of
• 2:21
  administrative
• 2:22
  gridlock, I guess. Yes. Absolutely. And I think
• 2:25
  it takes, like, true partnership and trust in
• 2:27
  order for to make that happen.
• 2:30
  So I think that the technology
• 2:32
  is out there. I think we just the
• 2:35
  people
• 2:36
  have to have the trust
• 2:37
  and
• 2:39
  the willingness
• 2:41
  to do something a little bit different. Yeah.
• 2:43
  And and coming to that trust, you really
• 2:45
  need sort of that,
• 2:47
  clinical validation. It's gotta be clinically sound. So
• 2:50
  how do you do that? How how can
• 2:51
  peer organizations really integrate these tools in a
• 2:54
  way that
• 2:55
  is equitable, transparent?
• 2:57
  And thinking from your experience as a as
• 2:59
  a clinician and also someone who who leads
• 3:01
  digital strategy,
• 3:02
  what are sort of your guiding principles or
• 3:04
  safeguards
• 3:05
  that are are essential to achieving that balance
• 3:08
  to making sure that
• 3:09
  AI is clinically sound and then you're coming
• 3:11
  to to that place of trust together?
• 3:13
  I think the first
• 3:15
  step
• 3:16
  is that intentionality.
• 3:18
  I think you have to be really thoughtful
• 3:20
  and intentional
• 3:22
  to ensure that we're embedding
• 3:24
  our human values
• 3:27
  of being patient centric. Right? Member centric.
• 3:30
  So
• 3:31
  at Lyric,
• 3:33
  our governance model is really a clinical governance
• 3:36
  first
• 3:37
  model.
• 3:38
  So we have multidisciplinary
• 3:40
  reviews. So we have clinicians. We have data
• 3:42
  scientists. We have legal. We have compliance. We
• 3:45
  really have just the whole gamut of people
• 3:49
  to give their input
• 3:51
  on
• 3:52
  what
• 3:53
  AI
• 3:54
  should be deployable
• 3:56
  before it's actually deployed and during deployment
• 3:59
  as well.
• 4:01
  And then we have continuous monitoring
• 4:03
  for equity
• 4:05
  and safety. Right? We hold our AI to
• 4:07
  really rigorous safety standards,
• 4:10
  and
• 4:11
  we require
• 4:12
  that we have this, like,
• 4:14
  subgroup performance checks,
• 4:16
  in real time governance to detect
• 4:19
  drift because the models can actually shift based
• 4:23
  on the data that's being,
• 4:25
  inputted.
• 4:26
  So and then finally, it's like that transparency.
• 4:30
  You can't build trust without the transparency.
• 4:34
  So the automated decisions, they have to be
• 4:37
  explainable,
• 4:38
  to clinicians and providers,
• 4:40
  to the people.
• 4:41
  So,
• 4:43
  you have to be able to back them
• 4:44
  up. Yeah. And that that transparency is a
• 4:47
  continuous act too. Just like as you're tweaking
• 4:50
  the models, there's no, like, set it and
• 4:51
  it's done. You know what I mean? Like,
• 4:53
  it's it's it's always constant work. It is
• 4:56
  constant work. Right? Because things are constantly changing,
• 4:59
  constantly evolving.
• 5:01
  I mean, look at medical evidence. When we
• 5:02
  talk about clinical, the medical evidence is constantly
• 5:06
  changing and evolving, so there is no set
• 5:08
  it and forget it.
• 5:09
  Yeah. And you could make the argument too
• 5:11
  that clinical evidence and medical evidence is just
• 5:13
  is gonna start evolving faster than what we've
• 5:15
  seen in the past. Absolutely.
• 5:17
  I mean, I think there's,
• 5:20
  a lot of evidence out there that says
• 5:22
  and I
• 5:23
  think someone quoted to me in 1950,
• 5:26
  the rate of doubling of medical literature is
• 5:29
  fifty years. And I think in 2024,
• 5:33
  it's something like sixty two days.
• 5:35
  So it's astonishing.
• 5:37
  So we have
• 5:39
  to be conscious of that and really intentional,
• 5:43
  about our models. Yeah. I I heard it
• 5:45
  put to me too. That's a I mean,
• 5:46
  that's that's a wild stat. Yeah. I also
• 5:49
  heard it put to me once that it's,
• 5:51
  you know, technology advancement innovation has never moved
• 5:54
  as fast as it's moving now, and it'll
• 5:56
  never move this slow again. Right. Yeah. I
• 5:59
  I have heard the same thing. Right? Like,
• 6:00
  this is the worst that it's gonna be.
• 6:02
  Yeah. So
• 6:04
  I don't know. That could be a little
• 6:05
  scary. Yeah. Yeah. That's a lot. It's a
• 6:08
  lot.
• 6:09
  But but let's let's come back to sort
• 6:11
  of thinking about your expertise and and what
• 6:13
  you're doing at Lyric.
• 6:15
  How does all this sort of fit into,
• 6:17
  I guess, helping health plans accelerate value based
• 6:19
  care goals,
• 6:21
  which are crucial to to getting to that
• 6:23
  payer provider,
• 6:25
  equanimity, I guess, to where things are sort
• 6:28
  of in harmony. Value based care is gonna
• 6:29
  have to flourish for that to happen. Right?
• 6:31
  Yeah. You know, I mean, that partnership is
• 6:34
  so important.
• 6:35
  I think that
• 6:37
  you can use AI, like, from that predictive
• 6:39
  modeling perspective,
• 6:41
  and you'll get better risk stratification or identification
• 6:44
  of at risk patients early.
• 6:47
  So you can really direct your clinical efforts
• 6:50
  to members who are gonna benefit the most
• 6:52
  and drive those better outcomes.
• 6:55
  I think from an administrative
• 6:57
  perspective,
• 6:58
  it can really decrease the administrative burden,
• 7:01
  which ultimately means
• 7:04
  providers, doctors, clinicians
• 7:06
  are gonna be spending more time with the
• 7:08
  patients. Right? And that's really what I think
• 7:10
  we want.
• 7:11
  So,
• 7:13
  automating those activities, it really frees,
• 7:16
  clinicians up to work top of license.
• 7:19
  And then
• 7:20
  shared measure measurement. Right? There's always been
• 7:24
  just that source of truth kind of friction,
• 7:28
  who has the right source of truth. And
• 7:30
  I think AI can really help align all
• 7:33
  the stakeholders,
• 7:34
  payers, providers, patients, by really integrating the data
• 7:37
  sources so you have one source of truth
• 7:40
  in a comprehensive
• 7:42
  view of outcomes.
• 7:44
  So data has always been really difficult to
• 7:46
  do with value based care, but I think
• 7:48
  there's a real opportunity here with AI to
• 7:50
  make the data far more accessible than it
• 7:53
  has been in the past,
• 7:55
  and transparent.
• 7:57
  So and then trust. Again, trust is at
• 8:00
  the center of all of it. That that
• 8:01
  that's where where my head went to. It
• 8:03
  comes back to, like, trust and transparency.
• 8:05
  You have the best technology in the world,
• 8:07
  but if you don't have trust and transparency
• 8:09
  sort of around it, it's not gonna not
• 8:11
  gonna do what it's supposed to do. Right?
• 8:13
  Right. Nobody's gonna use it. Right. Right. So
• 8:16
  So for folks listening out there who want
• 8:17
  to sort of, I guess, perhaps
• 8:20
  accelerate their their value based care goals and
• 8:23
  and put AI sort of at the center
• 8:25
  of that strategy, what's one step they could
• 8:26
  take? What advice would you give? You
• 8:29
  know, that's a tough one.
• 8:31
  But
• 8:32
  I would say,
• 8:34
  look at what's actually actionable. Right? Take something
• 8:38
  do a pilot on it. Partner with someone
• 8:40
  just on, you know, one aspect.
• 8:43
  You know, for example,
• 8:45
  here at Lyric, you know, we look at
• 8:47
  all sorts of different types of claims,
• 8:50
  look at a high cost inpatient DRG
• 8:53
  or coordination of benefits,
• 8:56
  Measure the hit rate. Look at provider appeals.
• 8:58
  Look at time to, you know, resolution.
• 9:02
  Grab some KPIs with just a short pilot
• 9:04
  and see if it works, and then iterate
• 9:06
  on that. Right? It's not gonna work the
• 9:08
  first time. It's just those multiple iterations and
• 9:10
  learning off the iterations,
• 9:12
  and I think that you'll find it really
• 9:15
  successful.
• 9:16
  That's, I think wonderful advice.
• 9:18
  You started at you said it was a
• 9:20
  tough question, but I think you landed the
• 9:21
  plane pretty good there.
• 9:23
  Mara, is there anything we didn't touch on
• 9:25
  or final thoughts you'd like to share? Perhaps
• 9:27
  it's something you wanna reemphasize for listeners. I'll
• 9:29
  leave it to you how how you wanna
• 9:31
  close out the podcast.
• 9:32
  Yeah. I think when we build AI with
• 9:35
  clinical intelligence,
• 9:36
  the governance, and the transparency,
• 9:39
  we move from a system that has a
• 9:40
  lot of friction in it to much better
• 9:43
  alignment across the system.
• 9:45
  It can help clinicians
• 9:48
  be clinicians.
• 9:49
  So it's a tool that we can use
• 9:52
  to improve outcomes for patients while
• 9:54
  really
• 9:55
  decreasing that administrative burden.
• 9:58
  Wonderful. Mara, thank you so much for being
• 10:00
  on the podcast. Thanks for having me. We
• 10:02
  also wanna thank our podcast sponsor, Lyric. Listeners,
• 10:05
  you can tune to more podcasts from Becker's
• 10:07
  Healthcare by visiting our podcast page at beckerspodcast.com.