Latest / Becker’s Healthcare Podcast / AI in the Exam Room, Patient Tools, Clinical Support and the Strategy Behind It

Transcript

AI in the Exam Room, Patient Tools, Clinical Support and the Strategy Behind It · Becker’s Healthcare Podcast

• 0:00
  Hello and welcome to the Becker's Healthcare podcast.
• 0:02
  My name is Chanel Bunger. Today, I'm excited
• 0:04
  to speak with Doctor. Anna Allen, and we're
• 0:07
  gonna talk about AI in the exam room,
• 0:09
  patient tools, clinical support, and the strategy behind
• 0:12
  it. And, Art, Doctor. Allen, thank you so
• 0:14
  much for joining me today. Thank you very
• 0:16
  much for having me. Please do feel free
• 0:17
  to call me Hannah. Perfect. Well, to get
• 0:20
  us started out, before we jump into anything,
• 0:22
  could you please introduce yourself and give us
• 0:23
  an overview of your work in healthcare? Thanks
• 0:26
  so much for having me. Really excited to
• 0:27
  be here. So I'm a clinician by background.
• 0:30
  So I'm a a general practitioner based in
• 0:32
  London. I've been a GP for the past
• 0:34
  fourteen years, and I've worked in the digital
• 0:36
  health space alongside of practicing frontline health care,
• 0:40
  in AI and, telemedicine and digital health for
• 0:43
  the past ten years as well. So loads
• 0:45
  of learnings, loads of challenges along the way,
• 0:47
  and I think we're at this real sort
• 0:50
  of pivotal moment in time where it's so
• 0:53
  exciting to see this novel technology coming to
• 0:56
  frontline healthcare
• 0:57
  and genuinely
• 0:58
  having an impact
• 1:00
  in how we practice as clinicians
• 1:03
  from general practice to hospital medicine to communities
• 1:06
  to emergency care, etcetera.
• 1:08
  And I guess I've been most excited
• 1:12
  about how we can kind of use data
• 1:14
  and use AI
• 1:15
  to do all of the stuff that either
• 1:17
  we aren't necessarily trained to do as doctors
• 1:20
  and the kind of admin burden associated with
• 1:22
  with being a clinician. Right? We spend a
• 1:24
  third of our working week now on admin,
• 1:27
  which is not patient facing and doesn't benefit
• 1:30
  the patient. And, actually, what we want to
• 1:32
  do is focus on the human side of
• 1:33
  health care. And that's what I think we've
• 1:35
  got. We've we've landed ourselves in a place
• 1:37
  where now we've got the tools and the
• 1:39
  AI and the technology to allow us to
• 1:41
  do that, which I'm really excited about. So
• 1:43
  my current role is as chief medical officer
• 1:45
  at Heidi
• 1:46
  for, The UK and Europe where, we're seeing
• 1:49
  a significant kind of adoption and traction
• 1:51
  over there, but also around the world as
• 1:53
  well over in The US, here and in
• 1:56
  Australia and various other places. Very exciting.
• 1:59
  Absolutely. And now that we have a sense
• 2:00
  of your background,
• 2:01
  there have been a wave of announcements about
• 2:03
  health care AI tools that provide health information
• 2:05
  and decision support.
• 2:07
  Can you break down the different categories of
• 2:09
  these tools, and who are they really designed
• 2:11
  to serve?
• 2:12
  Yeah. Of course. So, if we kind of
• 2:14
  take a step back and we kind of
• 2:16
  look at the jobs that either clinicians are
• 2:18
  doing, patients are doing,
• 2:20
  or, hospital systems are doing, there are sort
• 2:23
  of five or six ish categories that you
• 2:26
  can kind of put these things into.
• 2:29
  So one, which has been heavily covered
• 2:31
  lately, is that the transcription
• 2:34
  sort of functionality.
• 2:35
  So that is
• 2:36
  converting speech to a templated structured note for
• 2:40
  me as a clinician, and they're certainly clinician
• 2:42
  facing.
• 2:43
  You also have, obviously, kind of consumer facing
• 2:46
  patient products that we know of.
• 2:49
  And then you have things like,
• 2:51
  clinical decision support tools.
• 2:53
  So, you know, one of the tools that
• 2:55
  we've recently brought out is an evidence based
• 2:57
  tool that allows me to not have multiple
• 2:58
  tabs open, allows me to
• 2:59
  latest
• 3:02
  guidelines, latest formularies, latest ways to manage patients
• 3:05
  because it's impossible
• 3:06
  to retain all of that information in our
• 3:08
  own heads as clinicians when, you know, medical
• 3:11
  knowledge is doubling every 73 days. So we
• 3:14
  have the transcription tools, the clinical decision support
• 3:16
  tools, then we have the kind of operating
• 3:18
  system tools that focus on, you know, improving
• 3:21
  productivity, improving efficiency through the hospitals. Maybe it's,
• 3:24
  you know, how are we going to increase
• 3:26
  the throughput of patients through the emergency department,
• 3:29
  for example.
• 3:30
  And, you know, the executive boards of hospitals
• 3:33
  are interested in those kind of systems as
• 3:35
  well.
• 3:36
  And then we have patient facing tools. So
• 3:38
  they could be a whole host of different
• 3:41
  sort of layers of intelligence,
• 3:43
  etcetera,
• 3:44
  wearables, Internet of things that allow patients to
• 3:47
  have insights into their own health as well.
• 3:50
  And we've seen a whole host of patients
• 3:51
  using things like ChatGPT
• 3:53
  to support themselves of late.
• 3:55
  And then kinda lastly, the big bucket is
• 3:57
  around intelligence.
• 3:58
  Right? Like, how do we move from this
• 4:01
  reactive health care system where people get sick
• 4:04
  and they need to go to the emergency
• 4:05
  department or people get sick and they end
• 4:07
  up with me as a GP?
• 4:09
  But how do we actually stop that influx
• 4:11
  of patients coming in through that kind of
• 4:13
  top of funnel through the front door?
• 4:15
  And in order to do that, we need
• 4:17
  to look at the data. And in order
• 4:18
  to do that, we need to have a
• 4:20
  system
• 4:21
  where we can see kind of population
• 4:23
  level data
• 4:24
  to provide us with intelligent ways to manage
• 4:27
  our patients and improve our systems ultimately. That's
• 4:29
  how I kind of think of those, like,
• 4:31
  five or six systems
• 4:32
  and and who they're speaking to.
• 4:35
  Yeah. And you mentioned consumer facing tools. Digging
• 4:38
  deeper a bit on that.
• 4:40
  For the, quote, unquote, doctor AI tools aimed
• 4:42
  at patients, what guidance would you offer physicians
• 4:45
  on how to navigate conversations when patients bring
• 4:47
  those insights into the exam room?
• 4:50
  Yeah. That's a really good question. It is
• 4:51
  something I've been navigating myself in clinic. You
• 4:54
  know? The one thing I would start with
• 4:56
  is that if we take a step back,
• 4:57
  this isn't new. Right? It's a new way
• 5:00
  of bringing it to us, and it's a
• 5:02
  new sort of format
• 5:04
  that can be a bit intimidating for some
• 5:05
  clinicians, but it's not new that patients would
• 5:08
  bring
• 5:08
  accessory information with them or things that they've
• 5:11
  already validated. They already come with an idea
• 5:13
  of what they think is wrong with them,
• 5:15
  which is actually quite helpful when you embrace
• 5:17
  it.
• 5:18
  You know, since the dawn of time or
• 5:20
  the dawn of the Internet, patients were always
• 5:22
  googling their symptoms and bringing those in, which
• 5:24
  already frames how they're thinking about it. Right?
• 5:27
  So the thing that's changed is the kind
• 5:30
  of confidence, the quality, and the fluency of
• 5:32
  the information and data they're bringing in now.
• 5:34
  Now what that means for us as clinicians
• 5:37
  is that our role has changed. So we're
• 5:39
  not necessarily the person who's coming up with
• 5:41
  the first idea. Perhaps the AI has got
• 5:43
  it right, and it likely has.
• 5:46
  But how do we explain to the patient
• 5:48
  that we still need to look at them
• 5:50
  holistically? Because AI isn't doing that currently. They
• 5:53
  haven't got necessarily access to all of their
• 5:55
  medical information and their latest blood tests and
• 5:57
  their family history because I've known them for
• 5:59
  twenty years or whatever.
• 6:01
  That part,
• 6:03
  we still, as a human to human, need
• 6:05
  to explain to them and to be able
• 6:06
  to kind of relate to them on a
• 6:08
  human level.
• 6:10
  I think once we can kind of explain
• 6:12
  the wider sort of accountability
• 6:14
  framework and also the holistic three sixty viewpoint
• 6:17
  that we have of them as as a
• 6:18
  patient, as a person,
• 6:21
  then, actually, you can have a really great
• 6:23
  conversation
• 6:25
  using the information,
• 6:27
  as a clinician, but embracing that change in
• 6:29
  the role, right, where we have to be
• 6:31
  able to relate to them on a human
• 6:32
  level, but often we can be deemed as
• 6:35
  kind of gatekeepers. Right? How do we help
• 6:37
  them to navigate the information that they've got,
• 6:39
  which is probably quite accurate,
• 6:41
  but we need to
• 6:44
  part of our role is to educate them
• 6:46
  in, you know, is that source a good
• 6:48
  source? Is it a high quality source? If
• 6:51
  it's not, why is it not? What should
• 6:53
  they be looking for? And how might that
• 6:55
  mean that they've got some misinformation there? So
• 6:57
  that's part of our role now is explaining,
• 7:00
  the quality of the information that they've got
• 7:02
  already and what that means for them as
• 7:03
  a person.
• 7:04
  Absolutely. And on the flip side of that,
• 7:07
  for AI tools designed for clinicians, what are
• 7:09
  two or three key factors leaders should evaluate
• 7:12
  when vetting these tools?
• 7:15
  Gosh. I think there were quite a lot,
• 7:17
  to be honest.
• 7:18
  To kind of break it down to two
• 7:20
  or three, I think
• 7:22
  we need to stop thinking of these tools
• 7:25
  as standalone tools. Right? I think we need
• 7:28
  to look at how they fit in with
• 7:29
  a system. Just another tool in the toolbox.
• 7:32
  Exactly. Exactly. And you end up then, you
• 7:34
  know, procuring 20 different tools for 20 different
• 7:37
  purposes. That's 20 different training programs.
• 7:40
  And what if the adoption's low? You know?
• 7:42
  So which inevitably, it probably will be. So
• 7:44
  I think we need to look at how
• 7:46
  does it fit into your overarching kind of
• 7:48
  thesis and goal? What does success look like
• 7:50
  for it? And how does it work, on
• 7:53
  a kind of interoperability
• 7:54
  level? So what workflows do you have already,
• 7:57
  and how will this fit in and improve
• 8:00
  those workflows?
• 8:01
  You know, it's not just a siloed product
• 8:04
  or siloed tool for individual clinicians. We have
• 8:07
  to think bigger on this level.
• 8:09
  Absolutely. And as providers begin adopting these tools,
• 8:12
  how should they think about them within their
• 8:14
  broader AI strategy?
• 8:16
  And what does taking a a big picture
• 8:18
  approach really mean in this practice?
• 8:20
  Yeah. Great question. So I think kind of
• 8:22
  leading on from what we were saying before
• 8:24
  is not kind of siloed AI.
• 8:27
  We need to think about what does a
• 8:29
  world look like where we've got an AI
• 8:31
  enabled
• 8:32
  health care system.
• 8:34
  Right? And I encourage people to kind of
• 8:36
  think about the p's so it's easy to
• 8:39
  remember.
• 8:40
  Rather than individual products, think about problems that
• 8:43
  we're solving. And rather than individual products, let's
• 8:45
  think about pathways that we're wanting to improve.
• 8:48
  Because if we're constantly asking what problems are
• 8:51
  we having, we're thinking on a system level
• 8:53
  rather than in a siloed individual level. And
• 8:55
  that's really important when we're thinking about these,
• 8:58
  you know, AI bringing AI into existing workflows.
• 9:02
  And I would say to look for an
• 9:04
  AI partner rather than
• 9:06
  a scribe, rather than a clinical decision support.
• 9:09
  It needs to be, you know, the way
• 9:12
  that we're approaching things is looking at what
• 9:14
  problems hospital systems are having
• 9:16
  and clinicians are having and coming up with
• 9:19
  innovative solutions that utilize certain product functionality and
• 9:22
  features
• 9:23
  that will ultimately benefit the overarching
• 9:25
  goal of the hospital as as a whole.
• 9:29
  Got it. Got it. And now looking ahead,
• 9:31
  we all know that AI is moving incredibly
• 9:33
  fast. And with that, how do you see
• 9:35
  AI reshaping
• 9:36
  clinical decision support over the next two years?
• 9:38
  And what's one thing leaders need to do
• 9:40
  today?
• 9:41
  Yeah. That's a really good question. I think
• 9:44
  up until today,
• 9:45
  it's been an accessory
• 9:47
  sort of ancillary
• 9:48
  to how we're consulting and how we're practicing
• 9:52
  and how the systems are working,
• 9:54
  I think we're moving towards it becoming
• 9:58
  embedded into that system. So what does that
• 10:01
  embedded
• 10:02
  AI supported system and AI supported clinician look
• 10:05
  like?
• 10:07
  That will enable us to practice at the
• 10:08
  top of our game, at the top of
• 10:10
  our practice. Right? And I believe that in
• 10:13
  time, patients will only want to see clinicians
• 10:16
  who are AI
• 10:18
  augmented
• 10:19
  because they know they'll have a better, more
• 10:21
  accurate,
• 10:21
  more precise, more listened to experience.
• 10:25
  And so I think for leaders in this
• 10:27
  field, we need to then be looking
• 10:29
  at, like, where is this gonna have the
• 10:31
  most impact. Right? What workflow,
• 10:34
  which part of my system
• 10:36
  should I look at to start augmenting with
• 10:39
  AI first?
• 10:40
  So a lot of people look at, say,
• 10:42
  general practice because of high footfall or the
• 10:44
  emergency department, for example, but look at that
• 10:47
  on a system and workflow basis.
• 10:49
  So how can I augment the pathway and
• 10:52
  improve the pathway for the clinicians, for the
• 10:54
  hospital system overarching, and for the patients ultimately?
• 10:57
  Because that's how we're going to ultimately result
• 11:00
  in better patient care, which is what we're
• 11:02
  all here for. Well, Hannah, I wanna thank
• 11:04
  you for your time today. But before I
• 11:05
  let you go, is there anything else that
• 11:06
  listeners should know?
• 11:08
  I think the way that we are approaching
• 11:10
  things at Heide is in a very agile
• 11:12
  and innovative way. And so I think I
• 11:15
  would encourage people to look out for that
• 11:17
  next wave of innovation
• 11:19
  and, in particular, how it applies to your
• 11:22
  hospital system. So I'm thinking things like revenue
• 11:24
  cycle management, for example.
• 11:26
  We've got some really innovative partnerships and really
• 11:29
  innovative ways of approaching this that are going
• 11:32
  to be ultimately benefiting the hospital systems, and
• 11:35
  that it's coming very, very soon in the
• 11:36
  near future. So watch out for that one.
• 11:38
  Well, that's a great note to end on.
• 11:40
  I wanna thank you, Hannah, doctor Allen, for
• 11:42
  your time today. And we also wanna thank
• 11:45
  our podcast sponsor, Heidi. You can tune into
• 11:47
  more podcasts from Becker's Healthcare by visiting our
• 11:49
  podcast page at beckershospitalreview.com.
• 11:54
  Hi, everyone. This is Chanel with Becker's Healthcare.
• 11:56
  Thank you so much for tuning into the
• 11:58
  Becker's Healthcare podcast series. Today, we're gonna talk
• 12:00
  about AI in the exam room, patient tools,
• 12:03
  clinical support, and the strategy behind it. Joining
• 12:06
  me for today's discussion is doctor Hannah Allen,
• 12:08
  the chief medical officer at Heide.
• 12:11
  Hannah, doctor Allen, thank you so much for
• 12:13
  being here today.
• 12:19
  Yeah. I got it. And you mentioned the
• 12:22
  consumer facing tools
• 12:24
  or that quote, unquote doctor AI.
• 12:30
  Yeah. And you mentioned consumer facing tools. Digging
• 12:34
  deeper a bit on that.
• 12:35
  For the quote, unquote doctor AI tools aimed
• 12:38
  at patients, what guidance would you offer physicians
• 12:41
  on how to navigate conversations when patients bring
• 12:43
  those insights into the exam room?
• 12:51
  As providers begin adopting these tools, how do
• 12:54
  you think they should
• 12:55
  think about them within their broader AI strategy?
• 13:00
  Got it. Got it. And as providers begin
• 13:02
  adopting these tools, how should they think about
• 13:07
  got it. Got it. And as providers begin
• 13:09
  adopting these tools, how should they think about
• 13:12
  them within their broader air AI strategy?
• 13:15
  And what does taking a big picture approach
• 13:17
  really mean in practice?
• 13:22
  And what
• 13:23
  does
• 13:24
  and then
• 13:27
  and then what does taking a quote unquote
• 13:29
  big
• 13:32
  And then what does taking a big picture
• 13:34
  approach really mean in practice?
• 13:37
  Really mean in practice?
• 13:41
  Got it. Got it. And as providers begin
• 13:44
  adopting these tools, how should they think about
• 13:46
  them within their broader AI strategy? And what
• 13:48
  does taking a big picture approach really mean
• 13:50
  in practice?
• 13:55
  Now looking ahead, how do you see AI
• 13:58
  reshaping clinical decision support over the next few
• 14:00
  years? And what's one thing leaders need to
• 14:03
  know?
• 14:05
  Now looking ahead, how do you see AI
• 14:07
  reshaping clinical
• 14:10
  now looking ahead, how do you see AI
• 14:12
  reshaping clinical decision support over the next few
• 14:14
  years? And what's one thing leaders need to
• 14:16
  do today?
• 14:31
  Now that we have a sense of your
• 14:32
  background, let's jump into the conversation a bit.
• 14:36
  There
• 14:39
  there have been a wave of announcements about
• 14:41
  health care AI tools that provide health information
• 14:43
  and decision support. Can you break down the
• 14:46
  different categories of these tools and who are
• 14:48
  they really designed to serve?