WEBVTT

00:00:00.336 --> 00:00:02.569
Motor racing is a funny old business.

00:00:02.593 --> 00:00:04.886
We make a new car every year,

00:00:04.910 --> 00:00:07.074
and then we spend the rest of the season

00:00:07.098 --> 00:00:09.850
trying to understand
what it is we've built

00:00:09.874 --> 00:00:13.071
to make it better, to make it faster.

00:00:13.095 --> 00:00:16.346
And then the next year, we start again.

00:00:16.370 --> 00:00:20.584
Now, the car you see in front
of you is quite complicated.

00:00:20.608 --> 00:00:24.203
The chassis is made
up of about 11,000 components,

00:00:24.227 --> 00:00:26.671
the engine another 6,000,

00:00:26.695 --> 00:00:29.764
the electronics
about eight and a half thousand.

00:00:29.788 --> 00:00:34.165
So there's about 25,000 things
there that can go wrong.

00:00:34.189 --> 00:00:38.991
So motor racing is very much
about attention to detail.

00:00:39.015 --> 00:00:42.254
The other thing about Formula
1 in particular

00:00:42.278 --> 00:00:44.378
is we're always changing the car.

00:00:44.402 --> 00:00:46.658
We're always trying to make it faster.

00:00:46.682 --> 00:00:49.642
So every two weeks, we will be making

00:00:49.666 --> 00:00:53.842
about 5,000 new components
to fit to the car.

00:00:53.866 --> 00:00:56.020
Five to 10 percent of the race car

00:00:56.044 --> 00:00:59.772
will be different
every two weeks of the year.

00:00:59.796 --> 00:01:02.081
So how do we do that?

00:01:02.105 --> 00:01:05.825
Well, we start our life
with the racing car.

00:01:05.849 --> 00:01:09.816
We have a lot of sensors
on the car to measure things.

00:01:09.840 --> 00:01:11.698
On the race car in front of you here

00:01:11.722 --> 00:01:14.857
there are about 120 sensors
when it goes into a race.

00:01:14.881 --> 00:01:18.509
It's measuring all sorts
of things around the car.

00:01:18.533 --> 00:01:20.561
That data is logged. We're logging about

00:01:20.585 --> 00:01:24.265
500 different parameters
within the data systems,

00:01:24.289 --> 00:01:27.930
about 13,000 health parameters and events

00:01:27.954 --> 00:01:32.495
to say when things are not
working the way they should do,

00:01:32.519 --> 00:01:35.320
and we're sending that data
back to the garage

00:01:35.344 --> 00:01:40.299
using telemetry at a rate
of two to four megabits per second.

00:01:40.323 --> 00:01:43.426
So during a two-hour race,
each car will be sending

00:01:43.450 --> 00:01:45.701
750 million numbers.

00:01:45.725 --> 00:01:48.844
That's twice as many numbers
as words that each of us

00:01:48.868 --> 00:01:50.475
speaks in a lifetime.

00:01:50.499 --> 00:01:53.093
It's a huge amount of data.

00:01:53.117 --> 00:01:55.738
But it's not enough just
to have data and measure it.

00:01:55.762 --> 00:01:57.896
You need to be able to do
something with it.

00:01:57.920 --> 00:02:00.290
So we've spent a lot of time and effort

00:02:00.314 --> 00:02:02.159
in turning the data into stories

00:02:02.183 --> 00:02:05.264
to be able to tell,
what's the state of the engine,

00:02:05.288 --> 00:02:07.536
how are the tires degrading,

00:02:07.560 --> 00:02:11.284
what's the situation
with fuel consumption?

00:02:11.308 --> 00:02:14.072
So all of this is taking data

00:02:14.096 --> 00:02:17.874
and turning it into knowledge
that we can act upon.

00:02:17.898 --> 00:02:20.512
Okay, so let's have a look
at a little bit of data.

00:02:20.536 --> 00:02:22.542
Let's pick a bit of data from

00:02:22.566 --> 00:02:25.621
another three-month-old patient.

00:02:25.645 --> 00:02:29.792
This is a child, and what you're
seeing here is real data,

00:02:29.816 --> 00:02:31.769
and on the far right-hand side,

00:02:31.793 --> 00:02:34.555
where everything starts getting
a little bit catastrophic,

00:02:34.579 --> 00:02:37.940
that is the patient going
into cardiac arrest.

00:02:37.964 --> 00:02:41.172
It was deemed to be
an unpredictable event.

00:02:41.196 --> 00:02:44.961
This was a heart attack
that no one could see coming.

00:02:44.985 --> 00:02:47.511
But when we look at the information there,

00:02:47.535 --> 00:02:49.860
we can see that things
are starting to become

00:02:49.884 --> 00:02:53.889
a little fuzzy about five minutes
or so before the cardiac arrest.

00:02:53.913 --> 00:02:55.926
We can see small changes

00:02:55.950 --> 00:02:58.309
in things like the heart rate moving.

00:02:58.333 --> 00:03:00.795
These were all undetected
by normal thresholds

00:03:00.819 --> 00:03:03.203
which would be applied to data.

00:03:03.227 --> 00:03:06.346
So the question is, why
couldn't we see it?

00:03:06.370 --> 00:03:08.927
Was this a predictable event?

00:03:08.951 --> 00:03:11.937
Can we look more
at the patterns in the data

00:03:11.961 --> 00:03:15.317
to be able to do things better?

00:03:15.341 --> 00:03:17.967
So this is a child,

00:03:17.991 --> 00:03:21.199
about the same age
as the racing car on stage,

00:03:21.223 --> 00:03:22.829
three months old.

00:03:22.853 --> 00:03:25.434
It's a patient with a heart problem.

00:03:25.458 --> 00:03:28.902
Now, when you look at some of the data
on the screen above,

00:03:28.926 --> 00:03:33.804
things like heart rate, pulse,
oxygen, respiration rates,

00:03:33.828 --> 00:03:36.880
they're all unusual for a normal child,

00:03:36.904 --> 00:03:39.522
but they're quite normal
for the child there,

00:03:39.546 --> 00:03:43.660
and so one of the challenges
you have in health care is,

00:03:43.684 --> 00:03:46.511
how can I look at the patient
in front of me,

00:03:46.535 --> 00:03:49.558
have something which is specific for her,

00:03:49.582 --> 00:03:52.346
and be able to detect when
things start to change,

00:03:52.370 --> 00:03:54.445
when things start to deteriorate?

00:03:54.469 --> 00:03:57.495
Because like a racing car, any patient,

00:03:57.519 --> 00:04:00.471
when things start to go
bad, you have a short time

00:04:00.495 --> 00:04:02.302
to make a difference.

00:04:02.326 --> 00:04:05.056
So what we did is we took a data system

00:04:05.080 --> 00:04:08.187
which we run every two weeks
of the year in Formula 1

00:04:08.211 --> 00:04:11.189
and we installed it
on the hospital computers

00:04:11.213 --> 00:04:13.479
at Birmingham Children's Hospital.

00:04:13.503 --> 00:04:15.918
We streamed data
from the bedside instruments

00:04:15.942 --> 00:04:18.475
in their pediatric intensive care

00:04:18.499 --> 00:04:21.930
so that we could both look
at the data in real time

00:04:21.954 --> 00:04:24.802
and, more importantly, to store the data

00:04:24.826 --> 00:04:27.859
so that we could start to learn from it.

00:04:27.883 --> 00:04:32.243
And then, we applied an application on top

00:04:32.267 --> 00:04:35.513
which would allow us to tease
out the patterns in the data

00:04:35.537 --> 00:04:38.469
in real time so we could
see what was happening,

00:04:38.493 --> 00:04:42.182
so we could determine when
things started to change.

00:04:42.206 --> 00:04:46.045
Now, in motor racing, we're
all a little bit ambitious,

00:04:46.069 --> 00:04:48.594
audacious, a little bit
arrogant sometimes,

00:04:48.618 --> 00:04:51.992
so we decided we would also
look at the children

00:04:52.016 --> 00:04:54.949
as they were being transported
to intensive care.

00:04:54.973 --> 00:04:57.497
Why should we wait
until they arrived in the hospital

00:04:57.521 --> 00:04:59.097
before we started to look?

00:04:59.121 --> 00:05:02.094
And so we installed a real-time link

00:05:02.118 --> 00:05:04.930
between the ambulance and the hospital,

00:05:04.954 --> 00:05:08.706
just using normal 3G
telephony to send that data

00:05:08.730 --> 00:05:11.193
so that the ambulance became an extra bed

00:05:11.217 --> 00:05:14.329
in intensive care.

00:05:14.353 --> 00:05:18.031
And then we started looking at the data.

00:05:18.055 --> 00:05:20.952
So the wiggly lines
at the top, all the colors,

00:05:20.976 --> 00:05:24.146
this is the normal sort of data
you would see on a monitor --

00:05:24.170 --> 00:05:27.918
heart rate, pulse,
oxygen within the blood,

00:05:27.942 --> 00:05:30.553
and respiration.

00:05:30.577 --> 00:05:33.306
The lines on the bottom,
the blue and the red,

00:05:33.330 --> 00:05:34.807
these are the interesting ones.

00:05:34.831 --> 00:05:37.875
The red line is showing
an automated version

00:05:37.899 --> 00:05:39.472
of the early warning score

00:05:39.496 --> 00:05:42.211
that Birmingham Children's Hospital
were already running.

00:05:42.235 --> 00:05:44.297
They'd been running that since 2008,

00:05:44.321 --> 00:05:46.553
and already have stopped cardiac arrests

00:05:46.577 --> 00:05:49.310
and distress within the hospital.

00:05:49.334 --> 00:05:51.742
The blue line is an indication

00:05:51.766 --> 00:05:54.242
of when patterns start to change,

00:05:54.266 --> 00:05:56.551
and immediately, before we even started

00:05:56.575 --> 00:05:58.259
putting in clinical interpretation,

00:05:58.283 --> 00:06:01.129
we can see that the data
is speaking to us.

00:06:01.153 --> 00:06:04.665
It's telling us that something
is going wrong.

00:06:04.689 --> 00:06:08.481
The plot with the red and the green blobs,

00:06:08.505 --> 00:06:11.286
this is plotting different components

00:06:11.310 --> 00:06:13.833
of the data against each other.

00:06:13.857 --> 00:06:17.673
The green is us learning
what is normal for that child.

00:06:17.697 --> 00:06:20.283
We call it the cloud of normality.

00:06:20.307 --> 00:06:22.524
And when things start to change,

00:06:22.548 --> 00:06:25.088
when conditions start to deteriorate,

00:06:25.112 --> 00:06:27.326
we move into the red line.

00:06:27.350 --> 00:06:28.983
There's no rocket science here.

00:06:29.007 --> 00:06:33.096
It is displaying data that exists
already in a different way,

00:06:33.120 --> 00:06:36.487
to amplify it, to provide
cues to the doctors,

00:06:36.511 --> 00:06:39.225
to the nurses, so they can
see what's happening.

00:06:39.249 --> 00:06:42.355
In the same way that a good racing driver

00:06:42.379 --> 00:06:46.399
relies on cues to decide
when to apply the brakes,

00:06:46.423 --> 00:06:47.875
when to turn into a corner,

00:06:47.899 --> 00:06:50.793
we need to help
our physicians and our nurses

00:06:50.817 --> 00:06:54.413
to see when things
are starting to go wrong.

00:06:54.437 --> 00:06:57.359
So we have a very ambitious program.

00:06:57.383 --> 00:07:02.095
We think that the race is on to do
something differently.

00:07:02.119 --> 00:07:04.999
We are thinking big.
It's the right thing to do.

00:07:05.023 --> 00:07:08.411
We have an approach which,
if it's successful,

00:07:08.435 --> 00:07:11.055
there's no reason why it
should stay within a hospital.

00:07:11.079 --> 00:07:12.783
It can go beyond the walls.

00:07:12.807 --> 00:07:14.854
With wireless connectivity these days,

00:07:14.878 --> 00:07:18.298
there is no reason why
patients, doctors and nurses

00:07:18.322 --> 00:07:20.469
always have to be in the same place

00:07:20.493 --> 00:07:22.462
at the same time.

00:07:22.486 --> 00:07:26.457
And meanwhile, we'll take
our little three-month-old baby,

00:07:26.481 --> 00:07:30.214
keep taking it to the track,
keeping it safe,

00:07:30.238 --> 00:07:32.547
and making it faster and better.

00:07:32.571 --> 00:07:33.952
Thank you very much.

00:07:33.976 --> 00:07:38.930
(Applause)