Latest / Becker’s Healthcare Podcast / Allen Hsiao, MD, FAAP, FAMIA, Chief Health Information Officer at Yale School of Medicine and Yale New Haven Health

Transcript

Allen Hsiao, MD, FAAP, FAMIA, Chief Health Information Officer at Yale School of Medicine and Yale New Haven Health · Becker’s Healthcare Podcast

• 0:02
  - There's a small circle of people who truly
• 0:04
  - understand what it takes to lead a health
• 0:05
  - system right now. The decisions are enormous.
• 0:08
  - The variables are constantly shifting, and the stakes
• 0:11
  - of getting a call wrong don't stay contained.
• 0:14
  - That's not a complaint. It's a structural reality.
• 0:17
  - And it's why the most effective c suite
• 0:19
  - leaders are deliberate about where do they go
• 0:21
  - to pressure test their thinking.
• 0:22
  - Becker's fourteenth annual CEO and CFO roundtable
• 0:26
  - brings together more than 1,500
• 0:28
  - executive level attendees
• 0:30
  - and nearly 400 speakers for four days built
• 0:32
  - around growth, financial sustainability,
• 0:35
  - workforce strategy, and the leadership questions that don't
• 0:38
  - have easy answers.
• 0:39
  - Join us November second through the fifth in
• 0:41
  - Chicago.
• 0:42
  - For the agenda and event details,
• 0:45
  - visit beckershospitalreview.com
• 0:47
  - and click on the events tab in the
• 0:49
  - upper right.
• 0:51
  - Welcome to the Becker's Healthcare podcast. I'm Naomi
• 0:54
  - Diaz, and today I'm joined by doctor Alan
• 0:57
  - Hsiao, chief health information officer
• 1:00
  - at Yale School of Medicine and Yale New
• 1:02
  - Haven Health, where he also serves as vice
• 1:05
  - chair of clinical systems in the department of
• 1:07
  - biomedical informatics
• 1:09
  - and data science. Doctor Shao, thank you so
• 1:11
  - much for joining us today. Well, thank you
• 1:14
  - so much for having me.
• 1:15
  - Awesome. Well, doctor Shaw, why don't you get
• 1:17
  - us kicked off with our first question? If
• 1:18
  - you could just introduce yourself to our audience
• 1:21
  - and tell us a little bit about your
• 1:22
  - background.
• 1:24
  - Sure. For folks who might be interested, I'm
• 1:26
  - happy to share. Although, I I don't think
• 1:28
  - I'm that interesting. But,
• 1:30
  - my my background is, I guess, you know,
• 1:32
  - I train as a pediatrician,
• 1:34
  - initially,
• 1:36
  - and then, decided that I want to do
• 1:38
  - a fellowship in emergency medicine,
• 1:41
  - as a specialty of pediatrics. And,
• 1:44
  - during that time, my first year of,
• 1:47
  - of my fellowship and also kind of drawing
• 1:50
  - back on my, you know, residency,
• 1:52
  - I just kind of became more and more
• 1:54
  - interested in how, you know, the transition from
• 1:57
  - paper records to electronic medical records,
• 2:00
  - could be important. So I'm dating myself a
• 2:02
  - little bit, because we did we're using paper
• 2:04
  - records in a lot of the areas where
• 2:06
  - I was a physician early on.
• 2:09
  - And,
• 2:10
  - one
• 2:11
  - one patient really,
• 2:13
  - I think, it kind of
• 2:15
  - catalyzed that
• 2:17
  - interest of mine, where we had a very
• 2:19
  - sick infant come in,
• 2:21
  - who had low oxygen, you know, levels and
• 2:23
  - was really having trouble breathing. And we, you
• 2:25
  - know, we're pretty sure we're gonna need to
• 2:27
  - intubate him and and, you know, we're,
• 2:30
  - and just we're trying to struggle to understand
• 2:32
  - how to best treat him because we didn't
• 2:33
  - know what was going on,
• 2:35
  - and asked the parents,
• 2:37
  - for, you know, more medical history.
• 2:39
  - And, you know, they said, well, you know,
• 2:42
  - he he's healthy. And, as we're undressing him
• 2:45
  - to get a better exam,
• 2:47
  - physical exam, we noticed that he actually had
• 2:52
  - a pretty well healed,
• 2:53
  - you know, sternal incision.
• 2:56
  - So he clearly had some sort of major
• 2:58
  - heart surgery
• 2:59
  - of some sort, but his parents had told
• 3:01
  - us he was healthy.
• 3:03
  - And we were still on paper records in
• 3:05
  - the ED, but because I had trained at
• 3:08
  - Yale,
• 3:09
  - as a resident, I had access to our
• 3:11
  - relatively new,
• 3:13
  - electric electronic medical record, in our ambulatory
• 3:16
  - areas.
• 3:17
  - And so,
• 3:19
  - I was able to hop onto that system,
• 3:22
  - and quickly find this chart and found,
• 3:25
  - you know,
• 3:26
  - some really useful information and understood what his
• 3:29
  - surgery was, that it wasn't,
• 3:31
  - congenital heart defect where he still had a
• 3:33
  - mixing lesion, but in fact, they completely corrected
• 3:36
  - his transposition of the great arteries.
• 3:38
  - And that let us, you know information really
• 3:41
  - led us to know that,
• 3:45
  - his low oxygen saturations were not normal.
• 3:49
  - And for those who might not,
• 3:51
  - know, if you have,
• 3:53
  - you know, certain types of congenital heart defects
• 3:55
  - and you have a mixing lesion,
• 3:57
  - if you give a patient oxygen, you actually
• 3:59
  - can cause shunting in the wrong direction and
• 4:02
  - actually make them even sicker. And in fact,
• 4:04
  - you know, that could kill them.
• 4:06
  - And but in fact, we knew that his
• 4:08
  - had been he had no shunting,
• 4:11
  - currently, that that low oxygen level was certainly
• 4:13
  - should be treated like normal anatomy.
• 4:16
  - And he, in fact, needed very aggressive,
• 4:19
  - you know, pulmonary
• 4:21
  - support.
• 4:22
  - But just, you know,
• 4:24
  - just that kind of
• 4:26
  - clicked in my head that just having a
• 4:28
  - little piece of information at the right place
• 4:30
  - at the right time can dramatically improve patient
• 4:33
  - care.
• 4:34
  - And so we would either have to just,
• 4:36
  - you know, practice medicine in the dark if
• 4:38
  - we hadn't had access to his electronic records,
• 4:40
  - which normally we wouldn't have had.
• 4:43
  - And
• 4:44
  - but if but because we did, we could
• 4:46
  - really treat it much more aggressively and know
• 4:49
  - that we were providing the right treatments. And
• 4:51
  - so
• 4:52
  - that led me to,
• 4:53
  - grow more and more interested in,
• 4:56
  - in in clinical medical informatics.
• 4:59
  - And I was lucky that Yale offered, that
• 5:01
  - as a training program as well. And, talked
• 5:04
  - to my very supportive fellowship director, Karen Santucci.
• 5:09
  - And,
• 5:10
  - you know, she was very supportive,
• 5:12
  - as well as my my chief of emergency
• 5:14
  - medicine. And, they were very supportive of me
• 5:17
  - combining fellowships. And so I did some extra
• 5:20
  - years of training and did an informatics,
• 5:22
  - fellowship as well as an emergency medicine one.
• 5:24
  - And,
• 5:25
  - that just led kind of organically without any
• 5:28
  - plan of mine over time,
• 5:30
  - to take on more and more of an
• 5:31
  - operational role as well as supporting our,
• 5:34
  - our our our
• 5:35
  - clinical investigators
• 5:37
  - at the School of Medicine,
• 5:39
  - as, you know, electronic health record data, you
• 5:42
  - know,
• 5:43
  - have been growing in importance,
• 5:45
  - to the role, you know, now where I
• 5:47
  - serve as the chief health information officer for
• 5:49
  - the health system and the school.
• 5:52
  - Oh, doctor Shaul, thank you so much for
• 5:54
  - sharing that wonderful story about how you came
• 5:57
  - into this role. I think super inspiring there.
• 6:00
  - Another question I really wanted to ask you
• 6:02
  - today. I think healthcare is just going through
• 6:03
  - such an enormous transformation.
• 6:06
  - What are just some of opportunities and or
• 6:08
  - headwinds do you kind of have your eye
• 6:10
  - on right now within your role?
• 6:12
  - Yeah. Yeah. Great question. I mean, I think
• 6:15
  - if you ask any myself or any of
• 6:17
  - my colleagues, the one that probably comes foremost
• 6:20
  - to to all of us is is really
• 6:22
  - AI. Right? The impact artificial intelligence can,
• 6:26
  - make and and and and improve clinical care.
• 6:30
  - But at the same time, it's also a
• 6:31
  - headwind,
• 6:32
  - because,
• 6:34
  - you know, it's we all we all understand
• 6:36
  - that, you know, the the data in health
• 6:38
  - care is very dirty, very messy.
• 6:41
  - And AI, you know, may not necessarily realize
• 6:44
  - that, but it's being trained upon it. So
• 6:46
  - how do you really trust,
• 6:48
  - you know, the recommendations
• 6:50
  - or or the the clinical decision support that
• 6:53
  - that AI is trying to provide for you?
• 6:57
  - And and then,
• 6:58
  - I think another headwind too with with, you
• 7:01
  - know, kind of weighing
• 7:03
  - on the opposite side of of the great
• 7:05
  - opportunities of AI,
• 7:07
  - is just that our frontline physicians and other
• 7:09
  - clinicians,
• 7:10
  - are already overwhelmed. And and somewhat ironically,
• 7:14
  - you know, the the people that could probably
• 7:16
  - most benefit from the support that AI can
• 7:19
  - offer them
• 7:20
  - are I think we're finding
• 7:22
  - those often, you know, the least or, you
• 7:25
  - know, willing or or most reluctant,
• 7:27
  - to try something new because it is so,
• 7:30
  - you know,
• 7:31
  - somewhat foreign.
• 7:33
  - They don't can't quite, you know, wrap their
• 7:35
  - heads around how it works,
• 7:37
  - may not know, you know, all the potential
• 7:40
  - pitfalls and things they need to look out
• 7:41
  - for. So
• 7:43
  - so so it's sort of a it's a
• 7:44
  - it's a truly challenging balance.
• 7:46
  - But,
• 7:47
  - you know, I I think
• 7:49
  - none of us have seen so much change
• 7:51
  - happen so quickly in health care,
• 7:54
  - as we probably have with, you know, AI
• 7:56
  - in the last few years, especially with things
• 7:58
  - like ambient,
• 8:00
  - you know, documentation
• 8:01
  - where, you know, AI is now helping,
• 8:06
  - I I think it's probably already up in
• 8:07
  - the hundreds of thousands of physicians,
• 8:10
  - to more efficiently
• 8:12
  - document their,
• 8:14
  - patient,
• 8:15
  - physician,
• 8:16
  - you know, encounters.
• 8:18
  - And so that a lot now a lot
• 8:20
  - of physicians can turn away from that keyboard
• 8:23
  - and and face the patient directly,
• 8:25
  - be able to, you know,
• 8:27
  - even even hold their hands, read their body
• 8:29
  - language, explain things in detail,
• 8:31
  - and and understand maybe better how a patient
• 8:34
  - is reacting to something,
• 8:36
  - than if they were just, you know, kind
• 8:38
  - of head behind the screen.
• 8:40
  - And that's, you know, really, I think, brought
• 8:43
  - back a lot of joy,
• 8:45
  - in practicing, you know, medicine to a lot
• 8:47
  - of physicians.
• 8:49
  - It's not perfect. You know, I think that's
• 8:50
  - one of the challenges too as AI, as
• 8:52
  - as those who, you know, outside of health
• 8:54
  - care or in health care, you know, have
• 8:56
  - played with different large language model tools,
• 8:59
  - and and know that how quickly,
• 9:02
  - AI can generate, you know, content for for
• 9:05
  - lack of a better word.
• 9:07
  - It takes a lot of time to review
• 9:10
  - and, and and and and correct, you know,
• 9:13
  - things that it may have gotten wrong.
• 9:15
  - And I think that's not to be underestimated.
• 9:17
  - So you're sort of forced with just skimming
• 9:19
  - really quickly or maybe not even reviewing it
• 9:21
  - in the detail you need to.
• 9:23
  - Or, if you are very meticulous and you're
• 9:25
  - reading every word,
• 9:28
  - it it may not save you that much
• 9:29
  - time. And so that that's, you know, I
• 9:31
  - guess a headwind in in a in a
• 9:33
  - sense is,
• 9:35
  - how quickly AI can generate things and add
• 9:38
  - new,
• 9:39
  - tasks, if you will, to the physician. Because,
• 9:42
  - you know, I think for, you know,
• 9:45
  - at least now for the immediate future and
• 9:47
  - then some, I think, we really need the,
• 9:49
  - you know, clinician,
• 9:51
  - in the loop,
• 9:53
  - as as my, you know, chief data digital
• 9:55
  - health officer likes to, you know, remind folks
• 9:58
  - is we need to have, you know, a
• 10:00
  - clinician who
• 10:02
  - is, you know, an expert in that area
• 10:04
  - reviewing what the AI has come up with.
• 10:07
  - But that takes time. That that takes effort
• 10:09
  - as well. So,
• 10:10
  - you know, I think I think, you know,
• 10:12
  - there's a delicate balance there that we need
• 10:14
  - to find that I'm pretty sure we haven't
• 10:16
  - found yet.
• 10:18
  - But, you know, I think,
• 10:21
  - I I sort of shared in my initial
• 10:23
  - story,
• 10:24
  - you know, how that how having the right
• 10:26
  - information at the right time can can, you
• 10:27
  - know, can dramatically improve and save lives even,
• 10:32
  - prevent us from making a mistake.
• 10:34
  - And those of us, you know, who've got
• 10:36
  - some gray hair recommend you know, remember the
• 10:39
  - days where
• 10:40
  - everything was locked, you know, you know, paper
• 10:43
  - charts were the norm and information was locked
• 10:45
  - away and paper charts literally locked away maybe
• 10:48
  - in a filing cabinet,
• 10:50
  - in a filing medical records room,
• 10:53
  - at a different hospital, a different clinic, inpatient,
• 10:56
  - outpatient, ED,
• 10:58
  - you know, all had different charts,
• 11:00
  - that were kept in different areas.
• 11:02
  - And so, you know, electronic records were so
• 11:04
  - exciting because now, you know, things could be
• 11:07
  - in one system. Or if not in one
• 11:10
  - system, at least that one computer potentially could
• 11:13
  - log into multiple systems and get access to
• 11:15
  - them.
• 11:16
  - But then
• 11:17
  - you fast forward to what we have today,
• 11:19
  - which is where we're now faced with this
• 11:21
  - whole information overload. We have too much information,
• 11:24
  - somewhat ironically. Right? No one can possibly read
• 11:28
  - the hundreds of notes written by nurses, physicians,
• 11:32
  - advanced practice practitioners,
• 11:34
  - medical assistants,
• 11:36
  - that, you know, have been written about that
• 11:38
  - patient in the inpatient, the outpatient, the OR,
• 11:41
  - the ED,
• 11:42
  - areas. You just don't have the time, especially
• 11:45
  - if you're, you know, a physician seeing a
• 11:47
  - patient in the outpatient area,
• 11:49
  - and you have, you know, fifteen minutes, for
• 11:51
  - a visit. But that patient maybe has since
• 11:53
  - your last visit with you has seen three
• 11:56
  - other specialists,
• 11:57
  - has been hospitalized twice, has had a couple
• 11:59
  - of ED visits.
• 12:01
  - Who has time to read all those notes?
• 12:03
  - And so I think that's where, you know,
• 12:05
  - the last decade, physicians have felt very overwhelmed,
• 12:09
  - had a lot of cognitive load,
• 12:11
  - placed upon them.
• 12:13
  - And I've even I've even talked about things
• 12:15
  - like, you know, the the huge onslaught of
• 12:17
  - patient messages through through patient portals.
• 12:20
  - But I think now we're all kind of
• 12:22
  - very, very cautiously hoping AI
• 12:25
  - can really help, you know, surface what's really
• 12:27
  - important by reading all those hundreds of notes
• 12:30
  - and other messages and things like that,
• 12:32
  - and hopefully understanding the context of the current
• 12:36
  - visit, the current, you know, know, history of
• 12:37
  - present illness and surfacing what we need to
• 12:40
  - know,
• 12:41
  - for us.
• 12:42
  - But again, you know,
• 12:45
  - it's it's it's it's hard to, you know,
• 12:48
  - also, you know, also imagine how AI can
• 12:51
  - get that a 100% correct,
• 12:53
  - as, you know, can we know AI can
• 12:56
  - confabulate, can make mistakes.
• 12:59
  - And so what's that right balance? And how
• 13:02
  - do we know that we can trust it
• 13:04
  - without reading those hundreds of notes, you know,
• 13:06
  - ourselves,
• 13:07
  - that we can miss something. So
• 13:10
  - I think, you know, that that's probably the
• 13:12
  - biggest, you know, thing on the radar right
• 13:14
  - now for for myself and many of my
• 13:16
  - colleagues is
• 13:17
  - AI, the great opportunities it offers,
• 13:20
  - but also, you know, it itself is also
• 13:22
  - a headwind,
• 13:24
  - you know, that that, you know, we have
• 13:26
  - to be very careful of how we deploy
• 13:28
  - it, how we validate it, what is our
• 13:30
  - governance,
• 13:31
  - for it, and then how quickly do we
• 13:32
  - roll it out, how do we roll it
• 13:34
  - out even for our already overburdened clinicians.
• 13:38
  - Doctor Hsiao, I think I I love the
• 13:41
  - way you phrased that. It it both has
• 13:43
  - its opportunities and really reducing that administrative burden,
• 13:46
  - but as well as that you mentioned there,
• 13:48
  - you know, a lot of caution around scale
• 13:51
  - and ensuring that human in the loop there.
• 13:53
  - I think really, really great examples.
• 13:56
  - I think another thing I wanted to tap
• 13:57
  - on to you as well is, obviously, a
• 14:00
  - lot of health systems as well are in
• 14:02
  - growth mode for you. How are you thinking
• 14:04
  - about growth and really adding value back to
• 14:05
  - your organization this year? Yeah. Adding value back
• 14:06
  - to your organization this year?
• 14:09
  - Yeah. That's a great question. I mean, I
• 14:11
  - think so for you know, we're an academic
• 14:13
  - health care,
• 14:14
  - you know, system,
• 14:15
  - and and, and, you know, academic you know,
• 14:19
  - I guess it's
• 14:20
  - repetitive,
• 14:21
  - but, you know, redundant. But but, you know,
• 14:23
  - academic, you know, academically very strong medical
• 14:27
  - school,
• 14:28
  - as well that, you know, are very symbiotic.
• 14:31
  - And so I think for us, it's, you
• 14:33
  - know,
• 14:34
  - where I think about adding growth and and
• 14:36
  - value is just really embedding, you know, research,
• 14:41
  - you know,
• 14:43
  - into
• 14:44
  - our clinical care and and embedding more of
• 14:46
  - our clinical informatics
• 14:48
  - into our research infrastructure.
• 14:50
  - Right? And really cross pollinating. Because I think
• 14:52
  - in many ways, they're they're one and the
• 14:53
  - same,
• 14:54
  - in the sense that, you know, we are
• 14:56
  - this, you know, academic learning health care system.
• 14:59
  - We've got a school of incredible amount of
• 15:01
  - number of, you know,
• 15:03
  - brilliant investigators who are doing cutting edge research
• 15:06
  - in, you know, clinical sciences as well as
• 15:09
  - informatics.
• 15:10
  - You know, we we we recruited doctor Luzseil,
• 15:13
  - you know, Iona Machado to
• 15:15
  - start, our,
• 15:17
  - found our department,
• 15:19
  - of biomedical informatics and data science.
• 15:22
  - And
• 15:23
  - just the faculty there and and their expertise
• 15:25
  - in in AI,
• 15:27
  - is just really, really impressive. And for us,
• 15:30
  - we've got this whole
• 15:31
  - clinical laboratory in our health care system,
• 15:35
  - you know, with with robust amount of data
• 15:37
  - and,
• 15:38
  - and, you know, seven hospital campuses,
• 15:41
  - patients that I think, you know,
• 15:43
  - really, you know, had could benefit,
• 15:46
  - and and are benefiting from,
• 15:49
  - a lot of the science, that that we
• 15:51
  - have at the school, the informatics expertise.
• 15:54
  - And and I think really, you know, more
• 15:57
  - cross pollinating between the two,
• 15:59
  - is really where we're gonna, you know, kinda
• 16:01
  - grow and add value.
• 16:03
  - Especially, I think, you know, as we think
• 16:04
  - about the future, we think about how do
• 16:06
  - we, you know, bring this great AI research
• 16:09
  - to the bedside,
• 16:10
  - and how do we actually operationalize and implement
• 16:13
  - that so that the clinicians,
• 16:14
  - you know, trust it,
• 16:16
  - are using it effectively, that they can actually
• 16:18
  - use it within their workflow.
• 16:20
  - Because that's, you know, I think something that
• 16:22
  - my colleagues and I, we all know workflow
• 16:24
  - is king. And you can have the best
• 16:26
  - tool in the world, but if it's not
• 16:28
  - easy to use and within the workflow of
• 16:30
  - that very, very busy physician,
• 16:33
  - nurse, or advanced practice practitioner,
• 16:36
  - you know, they're they're not gonna they're not
• 16:37
  - gonna think about it. They're gonna have time
• 16:39
  - to use it,
• 16:40
  - because, you know, they they are so pressured,
• 16:43
  - for for time and resources.
• 16:45
  - And so,
• 16:47
  - yeah. So I think I think, you know,
• 16:50
  - really,
• 16:51
  - embedding informatics into our research mission,
• 16:54
  - and then bringing our more of our research,
• 16:58
  - you know, discoveries
• 16:59
  - to the bedside,
• 17:01
  - on on the on the clinical side is
• 17:03
  - is really where, you know, we're gonna add
• 17:05
  - a lot of value.
• 17:07
  - Yeah. Doctor. Schall, thank you for breaking that
• 17:09
  - down first. I think, again, excellent points there
• 17:11
  - and especially with the old New Haven being
• 17:13
  - an AMC as you mentioned there. Another thing
• 17:15
  - I wanted to kind of hone in as
• 17:17
  - well is
• 17:18
  - from your seat, what do you feel like
• 17:20
  - is maybe one risk or investment worth making
• 17:22
  - this year, especially as you mentioned earlier with
• 17:25
  - that rapid transformation happening
• 17:27
  - in health care today?
• 17:29
  - Yeah. Yeah. Great question.
• 17:31
  - So I think,
• 17:32
  - you know, I I think an investment we're
• 17:34
  - we're we're we're making,
• 17:37
  - and, you know, there's no I think, unfortunately,
• 17:39
  - there's no, like, you know, easy playbook, you
• 17:42
  - know, because this is this is so new
• 17:44
  - for for for all of us,
• 17:46
  - is really just,
• 17:48
  - kind
• 17:49
  - of investing
• 17:50
  - in,
• 17:51
  - our AI governance,
• 17:53
  - and how we tackle,
• 17:56
  - things,
• 17:56
  - and bring in AI tools,
• 17:59
  - as as an enterprise.
• 18:01
  - And, you know, our, doctor Lee Schwamm, our
• 18:04
  - digital health, you know,
• 18:06
  - officers really,
• 18:08
  - been
• 18:09
  - working hard to create,
• 18:11
  - AI governance infrastructure for us so we could
• 18:13
  - vet our internal tools or external tools,
• 18:16
  - tools from our EHR vendor,
• 18:19
  - that, you know, include AI, which now now
• 18:21
  - almost everybody's putting AI into almost everything.
• 18:25
  - And and for us, you know but it
• 18:26
  - it's a lot of work upfront, but getting
• 18:28
  - the right leads senior leadership,
• 18:31
  - together,
• 18:32
  - investing that governance. And then I think as
• 18:34
  - a subset of that governance,
• 18:36
  - also thinking about, you know, how we,
• 18:39
  - implement and and and I think, you know,
• 18:42
  - roll out some of these AI tools. And
• 18:44
  - in doing so,
• 18:45
  - creating, you know, what, you know, doctor Schwab,
• 18:48
  - you know, is calling our,
• 18:51
  - you know, AI center of excellence.
• 18:54
  - And so this is something that we're working
• 18:55
  - on and I think we need we need
• 18:57
  - to invest in quite a bit over this,
• 18:58
  - you know, this coming year
• 19:00
  - is is that, you know, would include the
• 19:02
  - the governance part, but also the implementation part,
• 19:05
  - which a large part of the thing is
• 19:07
  - gonna be is is
• 19:08
  - how do we upskill,
• 19:11
  - you know, our very, very busy nurses and
• 19:13
  - doctors and advanced practice practitioners,
• 19:17
  - to understand more about AI, to understand that
• 19:20
  - it's not perfect, that it may be biased,
• 19:22
  - but if used carefully, you can save them
• 19:25
  - a lot of time and effort, can make
• 19:27
  - them much more efficient.
• 19:30
  - And and then how do we get the
• 19:31
  - right training? How do we what's the right
• 19:32
  - amount of time? How do we find that
• 19:34
  - time in their busy schedules to upscale them?
• 19:37
  - And then how do we then roll out
• 19:39
  - and implement, you know, more of these AI
• 19:42
  - tools?
• 19:43
  - Because I think I I think I mentioned
• 19:45
  - earlier,
• 19:46
  - you know,
• 19:47
  - somewhat ironically,
• 19:49
  - the, you know, the
• 19:50
  - the AI tools that we do have in
• 19:52
  - place that are being used, you know, most
• 19:55
  - heavily are by,
• 19:57
  - those folks who are really comfortable with technology.
• 20:00
  - They've always loved being on the cutting edge.
• 20:02
  - They are the ones who, you know, are
• 20:04
  - able to, you know, build all these amazing
• 20:07
  - smart phrases and customizations
• 20:08
  - in the EHR and be very, very efficient.
• 20:11
  - Now they're even more efficient.
• 20:13
  - But I think our clinicians who are not
• 20:16
  - as technically savvy,
• 20:18
  - are, you know, probably haven't played with ChatGPT
• 20:21
  - that much themselves,
• 20:23
  - but really could benefit from AI really helping
• 20:26
  - them. Things like ambient, you know, documentation really
• 20:28
  - saving them time,
• 20:30
  - are the ones who will not really understand
• 20:32
  - how it works, will not be very trusting
• 20:34
  - of it, will not be really willing to
• 20:36
  - just play with something.
• 20:38
  - And so I think they're gonna need more
• 20:40
  - dedicated support,
• 20:42
  - to help them,
• 20:44
  - upskill, get comfortable,
• 20:46
  - and and and and start to adopt some
• 20:48
  - of these tools, which I think once they
• 20:49
  - do,
• 20:50
  - they'll be so happy and and so much
• 20:52
  - more efficient,
• 20:54
  - and really will improve their wellness. But,
• 20:57
  - we we face a
• 20:59
  - a a higher
• 21:00
  - challenge, you know, I think,
• 21:02
  - with with with with those folks. So that's
• 21:03
  - why I think, you know, investing in our
• 21:07
  - AI governance and our AI center of excellence,
• 21:10
  - and how we can best reach out to
• 21:12
  - these
• 21:13
  - folks, is is is the investment worth making,
• 21:16
  - you know, this year.
• 21:19
  - Doctor Shao, yes. Very similar to what I'm
• 21:21
  - just hearing across the board is that governance
• 21:23
  - and up scaling piece, I think super important
• 21:25
  - as AI continues to be implemented, as you
• 21:28
  - mentioned, across workflows, I think super important and
• 21:31
  - to keep an eye out on. For you
• 21:32
  - as well, I wanna tap on where do
• 21:34
  - you see the best opportunities
• 21:36
  - for growth in the future for Yale New
• 21:38
  - Haven as you look to towards the future?
• 21:43
  - I mean, I think it's a lot of
• 21:45
  - the things I've been discussing earlier is, you
• 21:47
  - know, thinking about how do we harness AI
• 21:50
  - safely,
• 21:52
  - in the right way and, you know,
• 21:56
  - where if we you know, and it's it's
• 21:58
  - gonna be a multiyear,
• 22:00
  - amount of work, you know, with us as
• 22:01
  - well as with our colleagues, all of us
• 22:03
  - working together. But, you know, if we can
• 22:05
  - really harness AI to
• 22:09
  - really
• 22:10
  - do things that we couldn't do really well
• 22:12
  - at scale. Right? I think not just, you
• 22:14
  - know, not just kind of churning through the
• 22:16
  - EHR data, but also combining that with imaging
• 22:19
  - data and genomic data,
• 22:21
  - you know,
• 22:23
  - wearables that we, you know, don't have the
• 22:25
  - capacity to do a lot of,
• 22:28
  - you know, with you know,
• 22:30
  - do a lot with that information today for
• 22:32
  - patients who have their own, you know, Apple
• 22:34
  - Watch or Fitbit or or whatnot at home,
• 22:37
  - but be able to kinda combine all that
• 22:39
  - information,
• 22:41
  - with the with the use of AI and
• 22:42
  - this, you know, kind of a multimodal
• 22:44
  - AI, if you will. I think there's a
• 22:46
  - lot of potential there, but that's, you know,
• 22:48
  - that's gonna be growth that we're all gonna
• 22:50
  - need to learn, figure out how to do
• 22:51
  - that well,
• 22:52
  - together in in the future.
• 22:55
  - You know, for us as a researcher for
• 22:57
  - in, you know, institution and and and learning
• 22:59
  - healthcare system,
• 23:00
  - I think,
• 23:02
  - you know, doing bringing more of these, you
• 23:04
  - know, informatics and AI tools to help support
• 23:07
  - clinical research,
• 23:09
  - whether it's identifying patients for clinical trials, for
• 23:12
  - help helping patients identify their own,
• 23:15
  - you know,
• 23:16
  - with, you know, on their own clinical trials
• 23:19
  - that we might have, that would be, you
• 23:20
  - know, valuable to them,
• 23:22
  - as well as making it, you know, lowering
• 23:24
  - the bar, making it very easy for,
• 23:26
  - our busy physicians to be aware of, clinical
• 23:29
  - trials opportunities
• 23:30
  - that their patients are a really good fit
• 23:32
  - for.
• 23:33
  - I think that's a that's a great opportunity
• 23:35
  - in the future as well.
• 23:37
  - You know, there's there's so many other things,
• 23:39
  - but I think those are kind of two
• 23:40
  - of the top ones that that that come
• 23:42
  - to mind for me.
• 23:45
  - Yeah. Doctor Shao, I think we hit on
• 23:46
  - a lot today. I appreciate you breaking everything
• 23:48
  - down. Such a transformative moment right now in
• 23:51
  - health care, but also as you mentioned,
• 23:53
  - as well. Just a lot going on at
• 23:55
  - a rapid pace. I appreciate you breaking it
• 23:57
  - down for us, your role, a little bit
• 23:59
  - about what you Yale New Haven is looking
• 24:01
  - at, for the rest of the year. And
• 24:03
  - just thank you so much for joining Becker's
• 24:05
  - podcast today. We really appreciate your insights.
• 24:07
  - Oh, you're so welcome, and and thanks for,
• 24:09
  - having me on today. And I really appreciate,
• 24:11
  - our our conversation today.
• 24:14
  - Thank you, doctor Hsiao. Have a good one.
• 24:16
  - You as well. Thank you.