Latest / Becker’s Healthcare Podcast / How AI is Revolutionizing Capital Planning and Supply Chain in Health Systems

Transcript

How AI is Revolutionizing Capital Planning and Supply Chain in Health Systems · 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. Today, we're talking about how AI
• 0:09
  is revolutionizing
• 0:10
  capital planning and supply chain in health systems,
• 0:14
  and I'm so excited to have her. Joining
• 0:16
  me for today's discussion is Shreleka Akula, chief
• 0:18
  data and AI officer at TriMetx. Shreleka, thank
• 0:21
  you so much for being here today. It's
• 0:23
  so great to have you.
• 0:25
  Thank you for having me, Lucas. Happy to
• 0:27
  be here. Absolutely. I do wanna start us
• 0:29
  off with introductions for those that might not
• 0:32
  know you at least yet.
• 0:34
  Could you just share a little bit about
• 0:35
  yourself and and your work?
• 0:38
  Happy to. I'm Shralika.
• 0:40
  I,
• 0:41
  I'm chief data and AI officer at, TriMetrix.
• 0:45
  I lead our product development data and AI
• 0:47
  functions here.
• 0:49
  Prior to that, I was chief technology officer
• 0:51
  for Alto Pharmacy.
• 0:53
  And before that, I led
• 0:55
  countries, one of the largest revenue cycle management
• 0:58
  software at Optum,
• 1:00
  including their, AI disruption,
• 1:02
  and prior life includes work at Google and
• 1:06
  Amazon's of the world and fintech,
• 1:08
  etcetera.
• 1:10
  I said that it's so great to have
• 1:11
  you, and that's I mean that in the
• 1:13
  truest sense because you have bring such a
• 1:15
  vast knowledge about AI
• 1:18
  and about the space with you. And I
• 1:20
  do wanna
• 1:21
  start us off with sort of laying the
• 1:23
  basics for our conversation today. Right?
• 1:27
  What is AI doing differently
• 1:30
  in capital planning today, and how should leaders
• 1:33
  think about its role
• 1:35
  alongside human decision making?
• 1:38
  Yeah. And it's it's interesting. Right?
• 1:40
  The overall, if we look at the Gartner's
• 1:43
  hype cycle on when first AI appeared as
• 1:45
  an intelligent agent, it's been roughly over three
• 1:48
  decades since it has happened.
• 1:50
  And in health care now and very specifically
• 1:53
  within capital planning, if we take stock of
• 1:55
  that,
• 1:56
  a recent study in SFMA,
• 1:59
  shows that the average age of a plant
• 2:01
  in a not for profit hospitals is roughly
• 2:03
  about 12.8
• 2:04
  in f I twenty four.
• 2:06
  That's
• 2:07
  the oldest average in at least the last
• 2:10
  thirteen years. And not surprisingly, that has showed
• 2:12
  a surge in capital spending. Right?
• 2:15
  And within that capital spending,
• 2:17
  today, if they a step back and look
• 2:19
  at how the capital planning cycle happens, it's
• 2:21
  pretty static.
• 2:23
  From the point where a CFO or a
• 2:24
  capital planning director asked a question about, hey,
• 2:27
  I have x y z budget and where
• 2:30
  do I spend it? To receive a cogent
• 2:32
  high fidelity response takes roughly two to three
• 2:35
  weeks to
• 2:37
  two to three months, actually. And God forbid,
• 2:39
  if they have additional questions on top of
• 2:41
  that, it's a few more weeks on top
• 2:43
  of that. Right? It's very static. It's a
• 2:45
  point in time.
• 2:47
  And what AI allows us to do is
• 2:49
  to have a dynamic and interactive process
• 2:53
  that is rooted in data, that is rooted
• 2:55
  in solid evidence, and that makes it more
• 2:57
  interactive and dynamic and real time.
• 3:01
  And it also allows for,
• 3:03
  scenario based modeling, which is hard to do
• 3:05
  in today's situation
• 3:06
  where if you were to make judgment calls
• 3:09
  or trade offs between different decisions and different
• 3:11
  scenarios,
• 3:12
  it it is easy to make it with,
• 3:15
  AI and health care now than what it
• 3:17
  used to be.
• 3:18
  And I'll also add this, Lucas, is,
• 3:21
  capital planning is not a once and done
• 3:23
  process. Right? It's a continuously
• 3:25
  informed process.
• 3:26
  For example, a device could be recalled. A
• 3:28
  cybersecurity attack would have happened that which needed
• 3:32
  a system to take down or new facilities
• 3:34
  are being added all the time. So capital
• 3:37
  planning is not once in a once and
• 3:39
  done, once a year,
• 3:40
  process, but it's actually a continuously informed process.
• 3:44
  And, the AI disruption in health care now
• 3:46
  allows and especially in capital planning, allows for
• 3:49
  us to have that continuously informed process rather
• 3:51
  than a static point in time process.
• 3:54
  I love that you mentioned that because I
• 3:56
  think
• 3:56
  when it comes to that, to be able
• 3:58
  to have that continuous process, what you do
• 4:00
  need is also continuous
• 4:02
  data.
• 4:03
  You need data that fuels that process to
• 4:06
  be continuous, which is so crucial as you've
• 4:08
  just outlined.
• 4:10
  What kinds of data does AI need to
• 4:13
  make better
• 4:14
  capital decisions, and why is that still
• 4:17
  so hard for health systems to pull together?
• 4:21
  Yeah.
• 4:22
  It is hard. Healthcare data is
• 4:25
  one of the most fragmented
• 4:28
  data systems
• 4:29
  in
• 4:30
  that we know of in any industry.
• 4:32
  And not only that, we know that health
• 4:34
  systems are generally
• 4:36
  vendor fatigued. Right? There are a lot of
• 4:38
  vendors in a single,
• 4:40
  value chain.
• 4:42
  And that is same for making these, capital
• 4:45
  decisions.
• 4:46
  To make an informed and well educated,
• 4:49
  deeply rooted in data and evidence decisions for
• 4:52
  capital planning, it needs the data from
• 4:56
  device intelligence,
• 4:57
  the utilization data, the device age remaining useful
• 5:01
  life, or
• 5:02
  the end of life as prescribed by the
• 5:05
  suppliers or the service history, cybersecurity,
• 5:07
  parts availability.
• 5:09
  All of these playing to making very informed
• 5:12
  decisions on capital planning.
• 5:14
  And that gives you the information about how
• 5:18
  and where a particular hospital needs to,
• 5:22
  make investments in based on their use. That's
• 5:24
  n equal to one. However,
• 5:27
  other thing that would be missing in that
• 5:28
  case would be an aggregate. How does that
• 5:31
  particular health system
• 5:33
  compare against its likeness of other health systems?
• 5:36
  That information would still be missing.
• 5:39
  So it's not only that independent health systems
• 5:42
  information, but it's also an aggregate of temporal
• 5:45
  data that is taken over decades of information.
• 5:49
  That makes it the most comprehensive
• 5:51
  informed decision for a health system to not
• 5:53
  only do it at any equal to one
• 5:55
  individual
• 5:57
  unique, insights into their capital planning, but also
• 6:00
  compare them against the rest of the ecosystem
• 6:02
  to say how well they're doing and how
• 6:04
  can they be continued to be competitive,
• 6:07
  in the landscape.
• 6:08
  And today, it's it's poor visibility. It's fragmented.
• 6:11
  It's incomplete data. All of those contribute to
• 6:14
  the difficulty in fetching this information.
• 6:17
  Yeah. We come back to what you've mentioned
• 6:18
  earlier. Right? It needs to be a continuous
• 6:20
  process and this, what you've just described, if
• 6:22
  that's not done right, that continuous process can't
• 6:24
  exist. It doesn't work for organizations that way.
• 6:27
  And I wanna turn this into action a
• 6:29
  little bit and and sort of discuss the
• 6:31
  scenario here.
• 6:32
  You've touched on the multi vendor piece here
• 6:35
  too, which is certainly crucial.
• 6:38
  Once AI
• 6:39
  predicts a likely failure, right, how should that
• 6:43
  insight then
• 6:44
  translate into real supply chain action,
• 6:48
  and where do health systems typically see breakdowns
• 6:51
  in some of this connecting
• 6:53
  predictive maintenance with that multi vendor part sourcing
• 6:56
  piece?
• 6:57
  Yeah.
• 6:58
  These are fragmented systems as we've been, talking
• 7:01
  about.
• 7:02
  Hospitals don't order parts, turns out, just for
• 7:05
  the fun of it.
• 7:06
  Similarly, a machine maintenance is not that complete
• 7:09
  until the necessary part is being replaced and
• 7:11
  it's fully repaired. It sounds silly when I
• 7:13
  say that. Right? But, however, the systems are
• 7:16
  actually built that way. They don't talk to
• 7:18
  each other. Mhmm.
• 7:20
  And bridging this gap between the two systems
• 7:23
  in this context of predictive maintenance and the
• 7:26
  failures
• 7:27
  and connecting that to the supply chain allows
• 7:29
  us to bridge that end to end value
• 7:31
  chain to meet these,
• 7:33
  failures
• 7:34
  with minimal disruption and downtime.
• 7:37
  A planned minimal downtime implies
• 7:40
  eliminating patient impact. At the end of the
• 7:42
  day, that's what we want as minimal, if
• 7:44
  at all, patient impact.
• 7:46
  And that's what it allows us to do
• 7:48
  by combining these two systems and informing one,
• 7:51
  one system to the other.
• 7:53
  The reason it is important for us to
• 7:55
  take stock of, multivendor intelligence of these parts
• 7:58
  is to allow for those predictive failures when
• 8:01
  those happen and when they are predicted well
• 8:04
  using
• 8:05
  AI. Being able to look at what is
• 8:07
  the best place and the best way for
• 8:09
  us to fetch a part to replace and
• 8:11
  maintain that, system
• 8:13
  is important. It allows for, hospital
• 8:16
  systems to be capital efficient,
• 8:18
  and it allows for, those direct savings to
• 8:20
  be translated back into the health system
• 8:23
  to repair it again, going back to the
• 8:24
  patient experience so that the planned downtime,
• 8:28
  happens.
• 8:29
  It's not just a break fix at that
• 8:31
  time responding
• 8:32
  reactively, but it's actually planned and it is
• 8:35
  procured
• 8:36
  before the, failure happened, ready to go and
• 8:39
  replaced, and that machine is repaired. And it's
• 8:42
  important that these two systems is connected to
• 8:44
  make sure that actually happens. And at the
• 8:46
  end of the day, the way it translates
• 8:48
  to is it's direct savings from plan supply
• 8:51
  chain. It's also direct impact on the patient
• 8:53
  experience, and it's preventing loss of revenue. Right?
• 8:56
  These,
• 8:57
  these
• 8:58
  hospital systems, these devices are typically seen as
• 9:01
  cost centers,
• 9:02
  but they are actually strategic assets for the
• 9:05
  hospital to make sure that patient care is
• 9:07
  not interrupted.
• 9:08
  And it's important as a result to make
• 9:10
  sure that these are not only predicted, but
• 9:13
  also connected all the way through the supply
• 9:14
  chain and procured
• 9:15
  ahead of time to minimize that and to
• 9:18
  prevent that loss of revenue.
• 9:20
  And I wanna touch a little bit more
• 9:21
  on that too because, obviously, loss of revenue
• 9:23
  is something that you've just highlighted that's really,
• 9:25
  really crucial to
• 9:27
  what are some of the other risks that
• 9:29
  you're seeing when all of this is treated
• 9:31
  in isolation, when some of these alerts are
• 9:33
  treated in isolation, some of this protective may
• 9:35
  predictive maintenance that we've touched on is treated
• 9:37
  in isolation.
• 9:38
  Again, without tying all of this
• 9:41
  to inventory availability, costs, capital planning, etcetera,
• 9:44
  what's the risk
• 9:46
  apart from just the revenue loss that you
• 9:48
  just described
• 9:49
  when we're treating all of this in
• 9:52
  isolation? Yeah. Let's walk through a scenario. Right?
• 9:55
  If these systems are not connected, what is
• 9:57
  the risk that we see operationally?
• 9:59
  It is the system's maintenance,
• 10:01
  is, is a break fix at that time.
• 10:03
  The system needs to be taken down. The
• 10:05
  patients need to be rescheduled.
• 10:08
  And that could actually lead to a patient,
• 10:11
  loss for that hospital system because, you know,
• 10:13
  they they don't want to, sit around waiting
• 10:16
  for this to come back up or to
• 10:17
  be rescheduled.
• 10:18
  And that's a real loss to the hospitals.
• 10:20
  And all the way through to inefficient supply
• 10:23
  chain systems
• 10:24
  is another risk, is procuring too many parts
• 10:27
  or procuring ahead of time instead of just
• 10:30
  in time or having the having to return
• 10:32
  them because it's not the right part or
• 10:34
  having to get at a cost, that are
• 10:36
  not optimal to the hospital system.
• 10:39
  These are all the various different risks that
• 10:41
  we see when we are looking at,
• 10:43
  not having a strategic
• 10:45
  multi vendor
• 10:47
  strategy at hospital system or when these predictive
• 10:50
  maintenance systems are not connected
• 10:52
  to proper sourcing and supply chain systems.
• 10:55
  An efficient supply chain process
• 10:58
  doesn't a solid case and justification
• 11:00
  for the hospital systems and for the capital
• 11:03
  planning directors and the CFOs and the CMOs
• 11:06
  and the CIOs to get a buy in
• 11:07
  from their executives
• 11:09
  collectively
• 11:10
  and from their board to make these decisions
• 11:13
  faster.
• 11:14
  At the end of the day, the speed
• 11:15
  to making decisions and having the confidence in
• 11:18
  the decision that they are making and give
• 11:20
  the sufficient test justification to their board is
• 11:23
  one of the most hardest things that they'd
• 11:25
  have to do year over year. And with
• 11:27
  AI in capital planning, it makes it
• 11:30
  easier, faster, better for all of them.
• 11:35
  Shralika, again, thank you so much for your
• 11:36
  time and insights today. We also want to
• 11:38
  thank our podcast sponsor, TriMedx. You could tune
• 11:41
  in to more podcasts from Becker's Healthcare by
• 11:43
  visiting our podcast page at beckershospitalreview.com.