Uli Chettipally: “Punish the Machine! Spare the Doctor and Save the Patient” | Talks at Google

Uli Chettipally: “Punish the Machine! Spare the Doctor and Save the Patient” | Talks at Google


[MUSIC PLAYING] SPEAKER 1: I am
really delighted to be able to welcome our guest
today, Dr. Uli Chettipally, who is both an MD and an MPH. Dr. Chettipally is a pioneer at
the intersection of two things that we all find
very interesting here at Google, which is
artificial intelligence and healthcare. He’s the co-founder and
CTO of Kaiser Permanente’s CREST network, where
he’s received the Pioneer Award from Kaiser
Permanente Innovations for his groundbreaking use
of clinical decision support technologies. He has presented at
many scientific meetings and industry events, such as
Big Data 360, Medical MEMS– I’m going to say
these all wrong– and Sensors, Connected
Healthcare, Maxim Integrated, Health Technology
Forum, and others. And he’s an assistant clinical
professor of medicine at UCSF, as well as a clinical researcher
at the Kaiser Foundation Research Institute, and
has published articles in a variety of
peer-reviewed journals. He received his medical training
from the Charles R. Drew University of Medicine
and Science in Los Angeles and has his Master’s in
Public Health from UCLA. Also, we hope that you
will keep an eye out for his forthcoming book, which
is coming out this year, which we’re hoping that he
will tell you more about during his presentation. So a round of applause
for Dr. Chettipally and we look forward to having you up. [APPLAUSE] ULI CHETTIPALLY: Thank you. Thank you, Mr. [INAUDIBLE]. And thanks for
everybody who came in, remotely and otherwise
in this room. It’s a pleasure to be here. I always wanted to
come here and talk a little bit about healthcare,
because the technology piece, you guys already know. So I’m not going to touch too
much about the technology side, more about how
healthcare has changed and how technology
can help get better at saving patients’ lives and
also, hopefully, protecting physicians from burnout and
all the difficult situations they’re in right now. As described by
Miss [INAUDIBLE],, my name is Uli Chettipally. I’m a physician,
and a researcher, and a scientist at
Kaiser Permanente. I see patients in the
emergency department. That’s my main profession. But I also do research. One of the key
things that I found was that, when you’re
seeing patients, you can make a difference in one
person, one patient at a time. When you work in
research and technology, you can make a
difference in hundreds, and thousands, and probably
millions of patients. And so that’s what gets me
really excited about working in research and technology. My group, CREST Network,
is a group of 12, or now 13, physician scientists. All of us are trained
in emergency medicine. And all of us are researchers. So we see patients and
also we do research on the things that interest us. We have a staff, about four
or five people, right here, division of research in Oakland. And we tackle some of the big,
big ticket items, like stroke, heart attacks, and sepsis,
and things like that. So I’m one of the
co-founders of that group. And also, I help design
the technology for that. Let’s see. I teach medical students
and residents at UCSF. They come to us for
their medical rotations in the emergency department. And I also run a
group called Society of Physician Entrepreneurs. It’s a large group of
healthcare entrepreneurs who are interested in
changing healthcare. I run the San Francisco
Bay Area chapter. And we meet every
third Thursday. And so all of you are welcome. It’s open to everybody. You can check out the
invitation on LinkedIn. Let’s see. Disclosures– I have
nothing to disclose. Right now, I am employed
by Kaiser Permanente and I do research. But the views that
I’m expressing here are totally mine. And I don’t get any industry
money or anything like that. When I was thinking
about doing this talk, I know you guys know
technology very well. But I thought I would give you
a little brief introduction to healthcare in
the United States and how it has
changed over time, and what are the challenges
now, and how to deal with some of those challenges. So I’m going to divide my talk
into three eras, the Golden Era, where about a
hundred years ago or so, when medical science
started revealing to us the secrets of disease
and how disease happens and how to prevent it. And the second era I
call the Steel Era. Steel Era is where
you’re looking at industrialization
of countries and huge growth of
large companies, large industries, large hospital
systems, diagnostic abilities and surgeries, so
everything that has to do with
building large things. And that’s the Steel Era
that we talk about today. And then, comes the Digital Era,
that’s more recent times, how digitization of
medicine has changed and will change in the future. So that’s talking
more about the future and how technology can help. The Golden Era, this is
probably 1850s or so, it starts around that time,
where the germ theory came about, where
scientists figured out that germs are the
ones that cause disease and how to treat them, how to
kill them with antibiotics. And a lot of causes of
diseases were found in that, I would say, about
100 years time frame. And you know, being an
epidemiologist by training, a lot of my heroes
are from this era. So I’ll talk about
a few of those. First one that comes to
my mind is John Snow. He was a physician in England. And during the 1950s, there
was a bad cholera epidemic in England. And so he was the one who
started investigating this. He was one of the earliest
scientists, or data scientists, I could say, in healthcare. So what he did was he
took a map and started putting dots on where people
were dying of cholera. And so this is the original
map that he worked on. And you can see those black
dots along the streets. And those are the places
where people died. And then, what he
did was he looked at one area, which was a
very dense area, where there were a lot of deaths happening. And that’s on Broad Street. And then, he figured out that
there was a pump, a water pump, called the Broad
Street pump, which he felt that somehow that
water coming from that pump was the one that is
causing this disease. And so he put out
this theory first. Until that time, people
were thinking that the air is what causes cholera. And so eventually,
what he did was he removed the
handle on the pump. You know, these pumps
had handles where you have to pump to get water. He removed the handle on that
pump and lo and behold cholera was declining. And then, later
on, they found out that it was the source of
water that was contaminated and that was causing cholera. And that pump still is up
there in England, in London. The second one, I’m sure that
you know this person, Florence Nightingale. She was a Duchess, actually
they called her Duchy. She was the daughter of a
rich person in England, again. She was good at mathematics,
another data scientist. And she went and worked
with the wounded veterans during the war. And this was in the 1950s also. And what she did was she did
enumerate the causes of death in these soldiers. And she prepared
this beautiful graph. You know, this is one
of the first, I guess, infographics, where
she described– So each of those slices is a
month’s time and the gray area, or the blue area, as
she describes it as, preventable deaths. The black area is where other
causes, or unknown causes. And the pink areas, or the
red areas, in her description, are the deaths from the
wounds, from bleeding. And remember, this was even
before there was antibiotics or the germ theory wasn’t
even published yet. So just by observation,
she was able to figure out a lot of things
using data science. Luther Terry was a physician
in the United States in the JFK era. He was the Surgeon General
of the United States. He had a tough choice
of putting out a warning that you see on the
cigarette labels. And this was the
first time that people realized that cigarette smoking
was injurious to health. And he is the one that is the
person behind this committee that sat together, looked
at all the evidence coming from the public health data. And soon they realized that,
wow, we are all smoking. And probably 75% of those
physicians, including Terry, was a smoker. And imagine how
hard it must have been trying to look at
that evidence, which shows that smoking is causing
cancer and other diseases. And so he was one of my heroes. Thomas Dawber was
another physician who was the lead investigator
in the Framingham Study. Framingham is a town
in Massachusetts. And so in the ’50s, what they
did was they picked this town and enrolled all the
people in the town, all the adults in the
town, about 5,000 of them, and then tracked their
health histories, tracked their diet,
tracked their activity, tracked their blood tests,
blood pressures, and everything. And there were
about 1,000 papers that came out of that study. And they studied for
years and years and years. And they’re still studying
that population, which is now in the third generation. And so this is where we
got a lot of the good data about what causes heart disease. And these are the risk factors
that they found, which we still practice today. So those were all
the great things that happened in those days that
changed the lives of millions of people, that actually
saved a lot of people too. Coming to the Steel Era, this
is the industrialized nations, starting anywhere from the
’50s, ’60s, to early 2000s. So this is where
everything is growing, big hospitals, big
hospital systems. Electronic health
records were introduced. And they were doing all kinds of
surgeries saving lives and very complicated procedures. And everything blossomed. And healthcare became
this huge, huge industry. It is all great, except that
it was getting out of hand. And so I’ll show you
some of the things. I think, based on my 25
years of experience, 50% of what we do in the
clinics, in the US, is a total waste of
money and effort. It’s unnecessary,
ineffective, or dangerous. Now, you must be thinking,
looking at that data, I don’t believe it. I don’t think that’s true. I don’t think this guy has
25 years of experience. But it’s true. I do have 25 years
of experience. And I think 50% of
it is unnecessary. So when you go to
your doctor, I’m sure you must be thinking, wow. What part of this 50%,
what part of this care, is not necessary or dangerous? So I’ll go over a few
things to show you what is happening in
the industry right now. This is from “Harvard
Business Review.” The population is aging. So one of the problems
with aging population is that the cost
of care goes up. If you’re a 30-year-old, maybe
it’ll cost about $5,000 a year to take care of you. But if you’re a 65-year-old,
it’ll cost $20,000. And so if everybody
is getting older, that means the cost of care
will get higher and higher. This other graph, I love,
because it shows the largest economies in the world. The tall, big, blue one
is the United States. And as you guessed right, the
second orange one is China. And then, comes Japan,
in the gray one. And the yellow line is Germany. And then, of course, UK and
France and other countries follow. Now, healthcare,
if you look at it, it’ll fit right there,
the US healthcare. That’s the red
one in the middle. We spend that much money
on healthcare every year. We spend almost as much as
the whole economy of Germany. And you know how big that is. That is the fourth
country, fourth biggest country in the world, as
far as GDP is concerned. And so that is what I’m
talking about when I talk about the cost of healthcare. This is a consumer price index
of various things that we buy. The ones in the bottom
are clothing, food, and other transportation,
things like that. The purple one at the
top is healthcare. Healthcare costs keep
going up and up and up. This is another graph, talks
about quality and spending. And healthcare is right there. All the other industrialized
countries, developed countries spend less and have
a better performance or better quality of care. We spend more and our
quality of care is less. And the knowledge keeps growing. It’s not like we don’t know the
science behind the diseases. And these are the number of
studies, the clinical trials, that happen in the US. And that keeps going up. And the 2017 number
is only up til August. So actually, 2018 is probably
shooting up out of the screen. Physicians are facing
a difficult time, mainly because most
of the compensation comes from a
fee-for-service world. And so they are worried
about their reimbursement. That means they have to work
harder, see more patients, do more things to make
the same amount of money. And then, of course,
there’s a lot of pressure from the payers
or the insurance companies. The regulations are
getting more complex. That means they have to do more
and more and check more boxes. And same thing with
the medical-legal risk, where they’re worried
about getting sued. So they order more tests
so that they are covered in case something goes wrong. And then, that increases their
workload, and their effort, and their stress. This is a survey
from Medscape, which shows that 51% of physicians,
they complain about burnout. That’s a sad thing, if
half of our physicians are not feeling good about their
jobs, about their workload. On the technology side, as
you know, data availability, there’s a lot of data being
collected, 1,000 fold increase. Hardware speeds are
increasing and algorithms are getting better and better. So now, we come to the current
age, which is the Digital Era. This is where I’ll talk
about some of the solutions, or how to tackle
these problems that we have seen that have grown out of
the Industrial Era or the Steel Era. Now, 80% of the medical
records are digitized. EHRs are pretty prevalent. Hopefully the efficiency
will get better and the quality will get better. And of course,
artificial intelligence seems to be getting inroads
into healthcare, which looks like a promising era. Within my research, I do
work with several groups here, locally and nationally. But we know that this
is where, I think, the biggest bang
for the buck lies. Let me share a story
with you that kind of portrays our current
healthcare situation. There was a town, which has
a river flowing through, a small town. And there’s a bridge there. One day, one person was
walking across the bridge and then saw a child
floating in the water. Actually, the child
was trying to swim and could not swim and kind
of floating away in the river. And so he tries to put
his hand and throws a rope and then saves the child
and brings him out. And then, soon
after, another child is coming along the river. And then he saves
that child too. And then, more and
more people are coming, trying to survive
this flood of water that is going under the bridge. And soon enough, he
calls his friends, and everybody is trying to save
these people out of the water. Soon they set up some
tents, some fire, some food, and everything. And they realize that
there’s more people coming. And so they set up a
nice building, hotels, and they build all
this infrastructure around that bridge, because
they see people floating. And some of them are dying. And so one wise person
comes by and says, hey, do you guys know where these
people are coming from? Maybe we should go
there and stop that. That thought of preventing
disease and illness and death is what I’m
talking about when I talk about how technology can help. If you look at healthcare
industry in general, it’s very complex. You have the physicians on
one side and the patients, which are the two most
important pieces in healthcare. But then, you have the
pharma, the medical devices, the academia, regulators,
malpractice, and payers. But each of those groups have
something else in their minds. The pharma, they are looking
at getting their FDA approval. They’re looking at
selling these drugs and eventually making profits,
which is a good thing. The same thing with devices. And in academia, you see doctors
trying to publish papers, because if you don’t
publish, you don’t survive. But one of the big
things that you realize is that, well, is their
primary goal to help patients? Same thing with regulators. They’re trying to control the
quality, and the malpractice, same thing. The lawyers are trying to
control the malpractice so that we can weed
out the bad doctors. Insurance company’s
trying to hold down costs. But then, in the mix, the
physicians and patients are getting lost. They are not the main focus of
all these other institutions that have grown around
this healthcare. All physicians want is to be
able to take care of patients and then, hopefully, decrease
all these other burdens that come in the way. So one of the best ways is– now, I’m talking about
solutions and how to steer this wave of increased
costs and decreased quality. Number one, I think the
business model has to change. And it’s already
changing in healthcare, from volume-based care
to value-based care. So when you talk about
volume-based care is, the more procedures
you do, the more money you make, the more
operations, the more drugs, the more hospitalizations. I think it has to
be flipped, where when you keep the
patients healthy, that’s when you make money. And that switch is
already happening. And that has to happen. The second thing is that
we have to go upstream. When I say upstream,
we don’t want to spend too much effort on
people who are really, really sick, but try to see how we
can prevent the sickness, even before it happens. So you have to look at
not just the hospital, but you have to look at
the environment, the home, the city where they
live in, the zip code. They say zip code
is more important than your genetic code. And then, you go upstream. Are there any genetic factors? We look at those things later. But the idea is
to go upstream, so that you can attack
the problem early on, before it becomes
a major problem. Because the more to the right
you go, the more expensive it becomes to deal
with the problems. The scientific
method has changed. Or let me put it this way. The scientific method is there. You typically ask a question. You collect a sample of data. And then you get an answer. But then with AI,
you do have the data. You don’t have to collect the
sample, although you could. The key is to ask
the right questions and be able to get the answers. One thing I keep saying is that
healthcare is an information business. And the sooner we realize it,
the more lives we can save. Because the way you deal
with the information, the way you can analyze
the information, the way you can extract
knowledge out of it becomes key when you
want to save patients. I like this paper that
came out a long time ago. Chris Anderson wrote
this, “The End of Theory– The Data Deluge Makes the
Scientific Method Obsolete.” In a way, we are
seeing that right now, because one of the problems
with scientific method is that you take samples
from a big population. You collect that study sample. And then, you study it. And then, you learn things
and apply that knowledge to your target population. Right now, you have the
whole target population. You go through these steps,
collecting data, analyzing, and translating,
implementing science. But then, it takes a long time. It takes about 18
years on average from an idea to studying it, to
analyzing it, implementing it, and then actually
coming to the bedside where the patients are
actually using that. I want this to speed up. It should take
less than a second to be able to get that data. In this day and age, we should
have access to all that data. We should have analytics and
engines that can work harder. And this is what I mean
by punish the machine, is the idea that we
have this technology and we’re just
letting it sit there and not be used for
the good of mankind. A lot of value-based
systems, as you might have noticed
or understood from, initially there used to
be a lot of procedures. I don’t know if you
remember, but we used to do a lot
of tonsillectomies and hysterectomies and
all kinds of procedures. But we soon realized
that, oh, we could treat that with medication. Maybe soon we’ll
realize that a lot of the sickness and illnesses
can be treated with diet. And now, we are talking about
more of the social determinants of health, how the
social situations will contribute to illness and
how we can prevent that. And so we have to go
up that value chain. I want to talk a little
bit about what I do. So my main key important
people are the physicians and the patients. And that’s all I care about. And how can we
design systems that will help the physicians take
better care of the patients? So we study data and we go back
and look at the retrospective data– because right now,
we have more than 10, 15 years worth of electronic
health record data– and see what happens to
these disease processes. How do people get sick? And then, you study
how you can predict it. What will happen? If you know that this person
has these risk factors and soon they’ll end up
here, well, hopefully, you can do something to
change those risk factors and prescribe something that
will change the trajectory of that disease process. And so that’s the idea. So we have
descriptive analytics, predictive analytics, and
prescriptive analytics. So what we have done
is we’ve taken data and put it through our engines. By the way, this
is a whole platform that is now active in
Kaiser Permanente Northern California, which is 21
hospitals and 21 emergency departments. And this is what we did. We created a platform that
can then have the engine, have the modules for
each disease process, and then have the clinical
decision support system, which means that it’ll
tell the physician what the next step should be. For example, if a patient comes
into the emergency with chest pain, we get a lot of patients
in the emergency departments. 21 emergency
departments, we have more than a million
patients that go through our
departments every year. Chest pain is probably number
two or number three chief complaint we call it. So people with chest
pain, maybe 100,000 people come through these
departments with chest pain. Only a few, maybe 5%, 10%,
15%, 20% have serious disease. The other 80% do
not have disease. But right now, there
are no systems– I mean, yeah, we do look
at our Framingham Study. But then, that’s
a little bit old. Because if we can figure
out what their risks are, so predicting the risk, and
then mitigating those risks, giving that information
to the physician, so that they can implement
those strategies. So when a patient comes
in with chest pain, when the physician clicks a
button in our medical record, it goes to our website. It opens up in the browser
within the health record. And it walks them
through a few steps and eventually giving
the physician the answer. The answer says this
person has a chance of 0.003% of having a heart
attack in the next 60 days. And so you should
do A, B, and C. Now, what it does
is, number one, you’re assessing the risk
exactly for this patient, based on your knowledge of
100,000 patients before. So you can actually guide this
physician in the patient’s care so that the best outcome can
be had with the least cost and least work. And so we are very
excited about this. And right now, we are looking at
implementing an AI-based engine and replacing what
we have right now, so that it becomes even faster
and more accurate, hopefully. And we do clinical trials
looking at these technologies. So half of the medical
centers have the technology. The other half do not. And then, we track
these patients to see what changes have
happened within these entities. And so we are very excited
and very happy about this. And we got an award for this
from our innovation folks. So all the answers are there. All the data is there. The technology is there. Everything is there. All we need to do is
ask the right questions. And when we ask the
right questions, we get those answers. Well, this is my
hope and my dream, is that when we have
an AI-enabled care, the bad outcomes will go
down and the good outcomes will go up. In the past, it used to
be very doctor-specific, where, well, there is this great
doctor in Houston, or Boston, or wherever. People used to go there. Now, it is more system-specific. In other words, wow,
this is a great system. Oh, Cleveland Clinic,
oh, Kaiser Permanente, they have great quality. Eventually, it becomes
more patient-specific, where you can actually, using
the data of all these patients in the past, you can
really fine-tune the care of each individual patient. In the past, it used to
be high touch and no tech, in the olden days. Now, it’s kind of lo
touch and lo tech. The current EMRs are very– what’s the word? Hard to use. In the future, it’ll be
more high touch again. But high tech will
be in the background. Because the physicians
won’t be directly interacting or entering
data, because the data is already there. So what does the future hold? Well, there are some great
opportunities in healthcare to be able to make a
difference in people’s lives. This story, I saw it on American
Heart Association’s website. They put this story
out about Justin, because they see this as the
triumph of the healthcare system. Justin had four heart attacks. He had five bypass surgeries. And believe it or not,
he had 35 stents placed in his heart blood vessels. That’s unbelievable. It’s great that Justin
survived this and is doing OK, I hope, after all this. Wouldn’t it be great if
we caught this problem before the heart attacks,
before the bypass surgeries, before the 35 stents
he had to be on? I think there is
definitely potential there to be able to change this. As you may know,
we are not doing that many bypass
surgeries nowadays, because we know that we can
do it better with medications. Even stents, we are not
putting that many in, because we know that, with
medication, the outcomes are equal or better. In fact, there were 800,000
stents placed last year in the United States that
were totally useless, that were not needed. Autism is a big area also. I feel there is a lot
of potential there. It’s a neurodevelopmental
disorder. By the way, I know a
little bit about it, because I have a
daughter with autism. So I kind of studied
that a little bit. Neurodevelopmental disorder,
kids are born with it. And they have it for
the rest of their lives. It’s a very expensive
disorder, because it costs more than a
few million dollars to take care of one
child, or one adult. And the incidence keeps growing. 1 in 59 was the
last one I think. And it keeps growing. And that’s what CDC says. And nobody knows why
this is happening. We don’t have any clue. I mean, genes are
probably involved, but that’s only a
small percentage. We don’t know if there is
something in the environment, if there’s something that we’re
eating, that we are wearing, that we’re inhaling. But this problem keeps growing. Right now, it’s
so big that it has overshadowed all the
developmental disorders, all the others. So this is bigger
than anything else. And this is one of the
most expensive disorders for the healthcare system. Another problem, opioid crisis– over the past, I
think, 20 years, so many people died of opioid
abuse during this crisis. And we don’t have a
solution for this. Well, we know that,
when we start people on opioid medications,
if they’re on opioid medicine for
more than three days, there’s a 20% chance that
they could get hooked on that medicine for
the rest of their lives. Imagine that. And we all must have
taken at some point, after surgery or whatever. And there was a recent study in
JAMA, Andrew Chang and others, they looked at really, opioids,
are they really that great? No. The pain control was the same
between opioids and nonopioids. If you give Tylenol
and Motrin, that’s the same, even if you
have a broken ankle. So why are we doing this? How did this crisis
happen, for so many years, for so many lives lost? It’s like the whole city
of Seattle, gone, vanished, because of this
crisis, because we don’t know the outcomes
of what we are doing. We don’t know. We haven’t studied it. Maybe we should. Maybe we should look at, what
are the chances of these people that we are starting them
on opioids that are going to end up dying in the future? Is it really helping their pain? Or is it just building
addicts out if it? So that’s a big, huge problem,
where data science, I think, can help. So to conclude my talk,
I’ll tell you a story. About 1,000 years
ago, there was a town. And it’s a nice peaceful town. Farmers living there, families. And there was a king. And he was a nice king. He took care of those people. One time, what happened
was there was a dragon that came to that town. And it started attacking
the chicken coops and eating their chicken,
because it was a small baby dragon. And so it was coming and
destroying and breaking things. And this is the kind of
dragon that spits fire. So it’ll set fire
on those things. And so it destroys
the property around. And they know that it
lives in the swamp right outside the town. And so the farmers, they came
together at the council meeting and said, we’ve got to
do something about this. And so one of the
farmers said, hey, what if we give the dragon
the chicken, every day, we feed him, we take them to
his place, and put it there, so that he doesn’t come here
and burn these things up? Yeah. That’s a great idea. And so they start
taking chicken every day and tying him up near the swamp. And the dragon
comes out at night. He eats and then goes
back into the swamp. Soon they ran out of chicken. Then, they started
putting sheep, goats. They ran out of those goats,
and cows, and everything. So now, they said, OK,
well, we have our kids. Let’s give our
children to the dragon. And so they started
tying these kids. And so the dragon keeps eating. And the dragon is getting
bigger and bigger. It wants more. And then, all the
children are gone. And now, it’s just the adults,
except the king’s daughter, the princess. She was a young teenager. And the farmers said,
well, it’s your turn, king. And the king was very
sad, because that was his only daughter. He had great hopes for her. But he gave his word that
they will feed the children. So they tied her
to a tree there. But then, before that
happened, the king announced, if anybody can
slay that dragon, I will give my daughter
to you in a wedding. And you will have
half of my kingdom. And so one shepherd boy came. He was another teenager. He said, well, I’ll
kill the dragon. Well, this guy
was not a soldier. He didn’t have any weapons. All these guys had
knives and swords and they couldn’t do anything. But he said, I will do it. And so what he did
was he went there, where the King’s
daughter was tied. He took the rope. And he hid behind a bush,
waited for the dragon. And then, he made a
noose with the rope, threw it around its
neck, tightened it, and so that the dragon
cannot breathe fire, or can’t even breathe. And he dragged the
dragon into the town and then, eventually, kills it. And that is St. George,
for those of you, you must have seen
some of these pictures. He became St. George. And he married and everybody
was happy after that. So the moral of the story is
that, we think about weapons, we think about things
to attack a problem. This shepherd boy, he
thought in a different way. He didn’t think about guns, and
knives, and swords, and spears. He thought about this rope. And we all have
this rope with us. We just have to use it. So let’s be smart
and save some lives and save some
doctors from grief. That’s my talk for today. And I welcome any questions. Thank you for waiting
here, patiently. I invite you to join
me in this fight. You can connect
with me on LinkedIn. That’s the easiest way. And I’m open to
questions right now. SPEAKER 1: Thank you so much. So I do have a
couple of questions on our online submission
for those folks who are on the Livestream,
since the people on Livestream are not able to speak
directly into the room. So I wanted to relay
a couple of these. The first is you said that
50% of what physicians do you see as potentially
being unnecessary. The question is do you
believe that some part of this comes from living in an overly
litigious culture, where physicians become
frightened of litigation and that is why the unnecessary
measures are being taken? Or do you see it coming
from some other source? They said, for example, bloated
bureaucracies, et cetera. ULI CHETTIPALLY:
So as I showed you, those three dragons,
litigation is one of those. Documentation, saying that
I did this, this, this, because I need to
get reimbursed. And the regulatory
framework, which causes us also to do more work. But there is a big
gap in our knowledge. And so a lot of it
is dogma, I feel. Oh, my professor taught
me to do it this this way. And this is what
the textbook say. So this is what I’m going to do,
not realizing that a lot of it may or may not be true. Because there may not be
scientific basis for whatever they are doing. I’ll give you a good example. You know, ankle sprain, we’ve
been taught, I don’t know, for 50 years or so that
it’s the RISE protocol, or RISE treatment, which is– now, I’m trying to
remember what RICE is– so rest, ice, C for
compression, E for elevation. So you have to rest your ankle. You have to ice it. You have to put a bandage
around it, compress it. And keep it elevated. I don’t know. Last time, I sprained my ankle. I just walked. I just did nothing. And it got better. Really? So we don’t know. Is rest better or
walking is better? Is putting a compression better? All those things, is ice
better or heat better? We have no idea, because
there’s no data or research that supports what
we’re trying to do. So like that, there
are several things that happen in healthcare
that physicians don’t know and nobody actually studied. So most of the research
funding doesn’t go to those kinds
of simple things. There has to be a fancy drug
or a fancy surgery or a device to get these researchers
interested or the funding agencies interested. And so the cost keeps
going up, because you get these new drugs. And so nobody’s
actually following up. Even when you have expensive
drugs or expensive devices, we are not following them up to
see, are they really working? Or is there something better? And so there’s no way to know,
all these drugs on the market, if they’re really effective. I know they did
a lot of research before they got approval. But once they get approval and
once they are in the market and they’re selling
those, we don’t know if that works or not. The research that happens is on
a few thousand patients here. But then millions of
patients are using those. But we not tracking all the data
on those millions of patients. So that’s where
the problem comes. SPEAKER 1: That is fascinating. We have a couple more that are
coming in, popping up here. One is from Thomas Kwan, in
another part of Mountain View. And his question is, with Kaiser
starting a new medical school next year, how do you plan
to influence the curriculum to train physicians in AI or
other Digital Era technologies? ULI CHETTIPALLY: Sure. So for those of you who
don’t know what the business model of Kaiser Permanente
is, Kaiser Permanente is one of the original
value-based care models that was started 70 years ago. It so happens that now it
has become the main thing, because people are
looking up to it. And in a value-based
care model, the idea is to keep the patients
healthy, or prevent disease. So that you get a fixed amount
of money and so, if you keep the patients healthy, that
means you make money when the patients are healthy, . which also means that we
have to go upstream and look at their vaccinations, their
colonoscopies, the mammograms, and all those things. How do we prevent cancer
from happening there, so that we can tackle it here? That in itself, now it is
becoming more and more. The rest of the
country is slowly trying to follow that model. Probably about 30%, 40% of
the healthcare companies are going to that model. And Medicare, the main
government funding agency for healthcare delivery,
is requiring more of these companies to do that. So yes. And what we found was
that, one of the big things is that in the traditional
medical schools, they are all geared
towards fee for service. That has been the tradition. So that continues. That’s one of the
reasons why Kaiser is planning to start
their own medical school. And hopefully, all this
training will be part of that. How do you prevent? How do you go upstream and
prevent problems later on? Right now, there
is enough data that shows that if you’re a Kaiser
Permanente– by the way, I’m not plugging in
Kaiser Permanente, but this is just
a scientific fact. If you’re a Kaiser
Permanente member, you have 50% less chance
of having a stroke or 30% less chance of having a
heart attack, just FYI. SPEAKER 1: Just FYI. That’s good for those of us
who are near Kaiser facilities here. Yes, we have one question
from the audience. One second. AUDIENCE: I have a question
about the decision tool that you’re building,
specifically around the fact that it takes a
lot of information. So it both looks at unsuccessful
cases as well successful cases to help determine what
the right path forward is. Looking very far
out, if we continue to use that successful case
and we continue to follow it, do we lose some of that
training information that’s important for that
comparative assessment? And then, kind of
thinking beyond that and tying in with
the Framingham Study, like the longitudinal effect. So maybe we would’ve
been told that opioids is really effective in
treating pain medicine. But we wouldn’t have
necessarily seen what happens five
years down the line and how some of the
longitudinal effects are being incorporated
into the decision tool. ULI CHETTIPALLY: That’s
a very good question. So this actually happened. So we were building this
tool for a condition called pulmonary embolism. Pulmonary embolism is where
you get a clot in your lung. It’s like a lung attack. And you can’t breathe. Some people die, because
it cuts off the circulation to the lungs and the brain. The brain doesn’t get oxygen. So about 95% of these patients
get admitted to the hospital, as soon as they are
diagnosed in the emergency. But we know that only a very
small percentage of patients will have bad outcomes. And so how do you
identify the ones that will have good
outcomes and not keep those patients in the
hospital, the low-risk ones? Because when you
stay in the hospital, you are at risk
for other things, hospital-acquired
infections and stuff. Also, it’s an inconvenience
for the patient. And it’s a cost
for the hospital. So we did a study
and said, OK, let’s build this tool which can
identify the low-risk patients. And we implemented it
in 10 medical centers, or 10 emergency departments. And the other 11 emergency
departments were controls. And we followed these
patients for more than a year. And we found that where
this tool was available, it was very successful
in decreasing the admission of unnecessary
low-risk patients. What we also found was
that– by the way, this tool we pulled it from
the literature. And this literature comes
from Canada, Australia. And then, what we found, once
we studied these patients, was that this tool was actually
overestimating the risk. Our patients are much healthier,
are at much lower risk. And so, version
two, what we did was we changed the dials
on that, so that it’ll estimate the risk correctly. And so to answer your question,
will we be losing data? Maybe. But maybe if we
track everybody, we won’t lose any data, even
bad outcomes, good outcomes. And then, we have
to constantly make the tool better and better. So each version becomes
better as the population gets healthier. And so in the long
run, it’ll work great, because the
population gets better and the tool gets better
at estimating the risk. But yeah, I understand, that
it does change our knowledge. And it does, hopefully,
for the good. SPEAKER 1: If everyone
has time to stick around for a couple more
questions, I’ve got a few more that
have popped up here. A quick one from Vivian Lee
in Cambridge, Massachusetts. She says thank you for the great
description of the chest pain AI application in
the Kaiser ERs. Do you have other
examples you can share? ULI CHETTIPALLY: Sure. We did one study
for head injury. A lot of kids that come
in with head injury, they fall and hit their heads. When they come in, one
of the tests we order is a CT scan of the head. And we know, again,
from studies, that if you do a CT
scan on kids, it’s bad, because it is radiation
at a young age. And it predisposes that person
to be at risk for cancer in the future. And so how do we decrease the
CT scans, especially in kids? The risk is higher. And so we designed
a study where, looking at the kid’s
features, how they present, can you identify the low-risk
kids that don’t need a CT scan? And we did the same thing
for belly pain in kids. What are the chances of this
kid having appendicitis, which is the most dreaded
diagnosis for these kids, because you got to
go through surgery. But then, radiating
them on their belly is probably the highest
radiation a kid can get, because this is a large
area in a small child. And so that study is
going on right now. And what we’re
trying to do is we’re trying to decrease the
number of radiations, without missing
appendicitis cases. That’s a great one. And then, so like
I said, surgery is the treatment
for appendicitis. But in the future,
maybe it’s antibiotics. Maybe it’s changing
in your diet. Maybe all those things we don’t
know, because we never thought about it or studied it. So there’s great information
coming out of that. SPEAKER 1: OK. Final question, and then I
think that we need to wrap up. Although we do have
a couple more here. There was one
anonymised question that came in asking about
single-payer systems, like the European
economies that you noted, which spend less but have
higher quality of care. These systems are often
accused of being overly costly for the country. Do you think that a
single-payer model could work in a country
like the United States? A nice easy question for you. ULI CHETTIPALLY:
That’s beyond my scope. I’m not a great business person. Economists may be able to
answer that question better. SPEAKER 1: Well,
Dr. Chettipally, thank you so much
for being here. We’re delighted to have you. And we very much look
forward to both the book coming out and hearing more
about what you’re doing with AI and healthcare in the future. So thank you so much. ULI CHETTIPALLY: Yes. Thank you. [APPLAUSE]

3 Comments

  1. Victoria Bell says:

    You lost me with the excessive introduction.

  2. my BulliPage says:

    Namaste, Dr.Chettipally garu. Thanks for an insight into present and future of medication. It has become quite a ritual to apply same diagnosis and treatment to all patients came with the same health complaints. It is still happening in India. It is heartening to note (and funny too to some extent) that Doctors hardly pay any interest in what patient is intend to say abou his own feelings/knowledge about his ailing body, how cruel and pity. Most of the patients also feel that it is their Doctors’ responsibility and duty to take care and control the patients’ ailments and just because of that the Doctors are paid. Well, I don’t know about rest of the world. Young medicaos are very much fascinated and fancied to use more and more mechanics (as you rightly said “punish the machines”) and depending on mechanical reports for diagnosis and treatment leaving/ignoring other patient specific unique disorders and ailments. Well, I don’t think AI can help in diagnosing patient specific disorders instead of speeding up generalised and “known” disorders which were fed up with ‘it’, I think you got what I mean it. What we all welcome is spending on prevention rather than spending on treatment as you rightly said. And I believe that spending on treatment is Doctor induced (more or less) and spending on prevention requires patient’s awareness of his/her own body and it’s requirements and reactions while we grow up in different ages, weather conditions, food intaking, emotion control, sanitary care, and some remote economical, social & geographical conditions which (all, I mean) can not be generalised. Hence, prevention starts with patient. Thank You for invoking interest.

  3. maninspired says:

    The history is interesting, but I'm interested in the headline. In the future, can you please at least share your thesis upfront. Thanks.

Leave a Reply

Your email address will not be published. Required fields are marked *