1
00:00:00,738 --> 00:00:02,952
I have a tendency to assume the worst,

2
00:00:03,757 --> 00:00:06,620
and once in a while,
this habit plays tricks on me.

3
00:00:06,644 --> 00:00:09,925
For example, if I feel
unexpected pain in my body

4
00:00:09,949 --> 00:00:13,456
that I have not experienced before
and that I cannot attribute,

5
00:00:13,480 --> 00:00:18,000
then all of a sudden, my mind
might turn a tense back into heart disease

6
00:00:18,024 --> 00:00:20,920
or calf muscle pain
into deep vein thrombosis.

7
00:00:21,699 --> 00:00:25,217
But so far, I haven't been diagnosed
with any deadly or incurable disease.

8
00:00:25,241 --> 00:00:27,653
Sometimes things just hurt
for no clear reason.

9
00:00:29,098 --> 00:00:31,318
But not everyone is as lucky as me.

10
00:00:31,970 --> 00:00:35,215
Every year, more than
50 million people die worldwide.

11
00:00:36,287 --> 00:00:39,820
Especially in high-income
economies like ours,

12
00:00:39,844 --> 00:00:44,599
a large fraction of deaths is caused
by slowly progressing diseases:

13
00:00:44,623 --> 00:00:48,931
heart disease, chronic lung disease,
cancer, Alzheimer's, diabetes,

14
00:00:48,955 --> 00:00:50,325
just to name a few.

15
00:00:51,300 --> 00:00:55,804
Now, humanity has made tremendous progress
in diagnosing and treating many of these.

16
00:00:56,859 --> 00:01:00,060
But we are at a stage
where further advancement in health

17
00:01:00,084 --> 00:01:03,374
cannot be achieved only
by developing new treatments.

18
00:01:03,398 --> 00:01:05,798
This becomes evident
when we look at one aspect

19
00:01:05,822 --> 00:01:08,188
that many of these
diseases have in common:

20
00:01:09,418 --> 00:01:12,246
the probability for successful treatment

21
00:01:12,270 --> 00:01:14,912
strongly depends on
when treatment is started.

22
00:01:15,370 --> 00:01:18,814
But a disease is typically only detected
once symptoms occur.

23
00:01:19,632 --> 00:01:24,025
The problem here is that, in fact,
many disease can remain asymptomatic,

24
00:01:24,049 --> 00:01:26,902
hence undetected,
for a long period of time.

25
00:01:27,902 --> 00:01:31,803
Because of this, there is
a persisting need for new ways

26
00:01:31,827 --> 00:01:33,775
of detecting disease at early stage,

27
00:01:33,799 --> 00:01:35,937
way before any symptoms occur.

28
00:01:37,135 --> 00:01:39,411
In health care, this is called screening.

29
00:01:39,435 --> 00:01:42,600
And as defined by
the World Health Organization,

30
00:01:42,624 --> 00:01:48,282
screening is "the presumptive
identification of unrecognized disease

31
00:01:48,306 --> 00:01:50,867
in an apparently healthy [person],

32
00:01:50,891 --> 00:01:54,680
by means of tests ... that can be applied
rapidly and easily ..."

33
00:01:55,429 --> 00:01:57,618
That's a long definition,
so let me repeat it:

34
00:01:58,237 --> 00:02:01,722
identification of unrecognized disease

35
00:02:01,746 --> 00:02:04,230
in an apparently healthy person

36
00:02:04,254 --> 00:02:08,268
by means of tests that can be applied
both rapidly and easily.

37
00:02:08,292 --> 00:02:11,671
And I want to put special emphasis
on the words "rapidly" and "easily,"

38
00:02:11,695 --> 00:02:13,945
because many of the existing
screening methods

39
00:02:13,969 --> 00:02:15,628
are exactly the opposite.

40
00:02:16,189 --> 00:02:18,888
And those of you
who have undergone colonoscopy

41
00:02:18,912 --> 00:02:22,111
as part of a screening program
for colorectal cancer

42
00:02:22,135 --> 00:02:23,482
will know what I mean.

43
00:02:25,209 --> 00:02:28,285
Obviously, there's a variety
of medical tools available

44
00:02:28,309 --> 00:02:29,838
to perform screening tests.

45
00:02:30,298 --> 00:02:33,984
This ranges from imaging techniques
such as radiography

46
00:02:34,008 --> 00:02:36,138
or magnetic resonance imaging

47
00:02:36,162 --> 00:02:38,166
to the analysis of blood or tissue.

48
00:02:38,190 --> 00:02:40,106
We have all had such tests.

49
00:02:41,344 --> 00:02:44,917
But there's one medium
that for long has been overlooked,

50
00:02:45,674 --> 00:02:48,416
a medium that is easily accessible,

51
00:02:48,440 --> 00:02:50,461
basically nondepletable,

52
00:02:50,485 --> 00:02:53,732
and it holds tremendous promise
for medical analysis.

53
00:02:54,994 --> 00:02:56,522
And that is our breath.

54
00:02:58,514 --> 00:03:01,895
Human breath is essentially
composed of five components:

55
00:03:02,649 --> 00:03:09,325
nitrogen, oxygen, carbon dioxide,
water and argon.

56
00:03:09,349 --> 00:03:12,748
But besides these five, there are
hundreds of other components

57
00:03:12,772 --> 00:03:15,155
that are present in very low quantity.

58
00:03:15,179 --> 00:03:17,732
These are called volatile
organic compounds,

59
00:03:17,756 --> 00:03:20,089
and we release hundreds,
even thousands of them,

60
00:03:20,113 --> 00:03:21,872
every time we exhale.

61
00:03:23,110 --> 00:03:26,479
The analysis of these volatile
organic compounds in our breath

62
00:03:26,503 --> 00:03:28,204
is called breath analysis.

63
00:03:29,545 --> 00:03:33,309
In fact, I believe that many of you
have already experienced breath analysis.

64
00:03:33,333 --> 00:03:36,934
Imagine: you're driving home
late at night,

65
00:03:36,958 --> 00:03:39,656
when suddenly, there's
a friendly police officer

66
00:03:39,680 --> 00:03:42,472
who asks you kindly but firmly

67
00:03:42,496 --> 00:03:45,899
to pull over and blow
into a device like this one.

68
00:03:47,486 --> 00:03:50,709
This is an alcohol breath tester

69
00:03:50,733 --> 00:03:54,440
that is used to measure
the ethanol concentration in your breath

70
00:03:54,464 --> 00:03:57,821
and determine whether driving
in your condition is a clever idea.

71
00:03:58,509 --> 00:04:00,456
Now, I'd say my driving was pretty good,

72
00:04:00,480 --> 00:04:01,700
but let me check.

73
00:04:02,788 --> 00:04:04,715
(Beep)

74
00:04:08,445 --> 00:04:11,405
Zero point zero, so nothing
to worry about, all fine.

75
00:04:11,429 --> 00:04:12,696
(Laughter)

76
00:04:14,298 --> 00:04:16,959
Now imagine a device like this one,

77
00:04:16,983 --> 00:04:19,927
that does not only measure
alcohol levels in your breath,

78
00:04:19,951 --> 00:04:23,767
but that detects diseases
like the ones I've shown you

79
00:04:23,791 --> 00:04:25,507
and potentially many more.

80
00:04:26,868 --> 00:04:30,666
The concept of correlating
the smell of a person's breath

81
00:04:30,690 --> 00:04:32,810
with certain medical conditions

82
00:04:32,834 --> 00:04:35,254
in fact, dates back to Ancient Greece.

83
00:04:35,905 --> 00:04:40,876
But only recently, research efforts
on breath analysis have skyrocketed,

84
00:04:40,900 --> 00:04:44,307
and what once was a dream
is now becoming reality.

85
00:04:45,307 --> 00:04:48,321
Let me pull up this list again
that I showed you earlier.

86
00:04:49,047 --> 00:04:51,813
For the majority of diseases listed here,

87
00:04:52,495 --> 00:04:54,688
there's substantial scientific evidence

88
00:04:54,712 --> 00:04:58,369
suggesting that the disease
could be detected by breath analysis.

89
00:04:59,917 --> 00:05:01,612
But how does it work, exactly?

90
00:05:02,421 --> 00:05:05,199
The essential part is a sensor device

91
00:05:05,223 --> 00:05:08,435
that detects the volatile
organic compounds in our breath.

92
00:05:09,364 --> 00:05:12,294
Simply put: when exposed
to a breath sample,

93
00:05:12,318 --> 00:05:15,002
the sensor outputs a complex signature

94
00:05:15,026 --> 00:05:19,006
that results from the mixture of volatile
organic compounds that we exhale.

95
00:05:20,141 --> 00:05:24,457
Now, this signature represents
a fingerprint of your metabolism,

96
00:05:24,481 --> 00:05:25,646
your microbiome

97
00:05:25,670 --> 00:05:29,273
and the biochemical processes
that occur in your body.

98
00:05:29,980 --> 00:05:31,573
If you have a disease,

99
00:05:31,597 --> 00:05:33,884
your organisms will change

100
00:05:33,908 --> 00:05:36,868
and so will the composition
of your exhaled breath.

101
00:05:37,479 --> 00:05:42,609
And then the only thing that is left to do
is to correlate a certain signature

102
00:05:42,633 --> 00:05:45,739
with the presence or absence
of certain medical conditions.

103
00:05:47,921 --> 00:05:51,156
The technology promises
several undeniable benefits.

104
00:05:51,719 --> 00:05:54,816
Firstly, the sensor can be miniaturized

105
00:05:54,840 --> 00:05:57,686
and integrated into small,
handheld devices

106
00:05:57,710 --> 00:05:59,548
like this alcohol breath tester.

107
00:06:00,437 --> 00:06:03,450
This would allow the test to be used
in many different settings

108
00:06:03,474 --> 00:06:05,248
and even at home,

109
00:06:05,272 --> 00:06:07,266
so that a visit at the doctor's office

110
00:06:07,290 --> 00:06:10,049
is not needed each time
a test shall be performed.

111
00:06:10,946 --> 00:06:14,326
Secondly, breath analysis is noninvasive

112
00:06:14,350 --> 00:06:17,997
and can be as simple as blowing
into an alcohol breath tester.

113
00:06:18,900 --> 00:06:23,894
Such simplicity and ease of use
would reduce patient burden

114
00:06:23,918 --> 00:06:27,361
and provide an incentive
for broad adoption of the technology.

115
00:06:27,973 --> 00:06:32,081
And thirdly, the technology is so flexible

116
00:06:32,105 --> 00:06:34,345
that the same device could be used

117
00:06:34,369 --> 00:06:37,509
to detect a broad range
of medical conditions.

118
00:06:38,224 --> 00:06:42,909
Breath analysis could be used to screen
for multiple diseases at the same time.

119
00:06:44,195 --> 00:06:47,526
Nowadays, each disease typically requires
a different medical tool

120
00:06:47,550 --> 00:06:49,006
to perform a screening test.

121
00:06:49,561 --> 00:06:52,645
But this means you can only find
what you're looking for.

122
00:06:54,637 --> 00:06:58,496
With all of these features,
breath analysis is predestined

123
00:06:58,520 --> 00:07:01,584
to deliver what many traditional
screening tests are lacking.

124
00:07:02,671 --> 00:07:04,310
And most importantly,

125
00:07:04,334 --> 00:07:07,684
all of these features should
eventually provide us

126
00:07:07,708 --> 00:07:10,214
with a platform for medical analysis

127
00:07:10,238 --> 00:07:13,943
that can operate at attractively
low cost per test.

128
00:07:15,713 --> 00:07:18,963
On the contrary, existing medical tools

129
00:07:18,987 --> 00:07:21,373
often lead to rather high cost per test.

130
00:07:21,920 --> 00:07:24,198
Then, in order to keep costs down,

131
00:07:24,222 --> 00:07:26,832
the number of tests
needs to be restricted,

132
00:07:27,562 --> 00:07:30,803
and this means a) that the tests
can only be performed

133
00:07:30,827 --> 00:07:35,511
on a narrow part of the population,
for example, the high-risk population;

134
00:07:35,535 --> 00:07:40,690
and b) that the number of tests per person
needs to be kept at a minimum.

135
00:07:41,459 --> 00:07:43,303
But wouldn't it actually be beneficial

136
00:07:43,327 --> 00:07:46,287
if the test was performed
on a larger group of people

137
00:07:46,311 --> 00:07:50,559
and more often and over a longer period
of time for each individual?

138
00:07:51,845 --> 00:07:55,706
Especially the latter would give access
to something very valuable

139
00:07:55,730 --> 00:07:57,997
that is called longitudinal data.

140
00:07:58,950 --> 00:08:03,052
Longitudinal data is a data set
that tracks the same patient

141
00:08:03,076 --> 00:08:05,572
over the course of many months or years.

142
00:08:07,318 --> 00:08:12,368
Nowadays, medical decisions
are often based on a limited data set,

143
00:08:12,392 --> 00:08:15,584
where only a glimpse
of a patient's medical history

144
00:08:15,608 --> 00:08:17,662
is available for decision-making.

145
00:08:18,210 --> 00:08:20,509
In such a case,

146
00:08:20,533 --> 00:08:23,326
abnormalities are typically detected

147
00:08:23,350 --> 00:08:25,916
by comparing a patient's health profile

148
00:08:25,940 --> 00:08:29,212
to the average health profile
of a reference population.

149
00:08:30,419 --> 00:08:34,337
Longitudinal data would
open up a new dimension

150
00:08:34,361 --> 00:08:37,229
and allow abnormalities to be detected

151
00:08:37,253 --> 00:08:39,950
based on a patient's own medical history.

152
00:08:40,862 --> 00:08:43,685
This will pave the way
for personalized treatment.

153
00:08:44,741 --> 00:08:46,305
Sounds pretty great, right?

154
00:08:47,194 --> 00:08:50,169
Now you will certainly have a question
that is something like,

155
00:08:50,193 --> 00:08:54,988
"If the technology is as great as he says,
then why aren't they using it today?"

156
00:08:55,012 --> 00:08:57,490
And the only answer I can give you is,

157
00:08:57,514 --> 00:09:00,094
not everything is as easy as it sounds.

158
00:09:00,983 --> 00:09:04,470
There are technical
challenges, for example.

159
00:09:04,494 --> 00:09:07,508
There's the need for
extremely reliable sensors

160
00:09:07,532 --> 00:09:10,800
that can detect mixtures
of volatile organic compounds

161
00:09:10,824 --> 00:09:12,917
with sufficient reproducibility.

162
00:09:13,629 --> 00:09:16,064
And another technical challenge is:

163
00:09:16,088 --> 00:09:21,074
How do you sample a person's breath
in a very defined manner

164
00:09:21,098 --> 00:09:23,446
so that the sampling process itself

165
00:09:23,470 --> 00:09:26,123
does not alter the result of the analysis?

166
00:09:27,187 --> 00:09:28,975
And there's the need for data.

167
00:09:29,556 --> 00:09:33,514
Breath analysis needs
to be validated in clinical trials,

168
00:09:33,538 --> 00:09:36,030
and enough data needs to be collected

169
00:09:36,054 --> 00:09:40,296
so that individual conditions
can be measured against baselines.

170
00:09:41,167 --> 00:09:43,640
Breath analysis can only succeed

171
00:09:43,664 --> 00:09:46,437
if a large enough data set
can be generated

172
00:09:46,461 --> 00:09:48,811
and made available for broad use.

173
00:09:51,050 --> 00:09:53,880
If breath analysis
holds up to its promises,

174
00:09:54,745 --> 00:09:57,271
this is a technology
that could truly aid us

175
00:09:57,295 --> 00:09:59,538
to transform our health care system,

176
00:10:00,410 --> 00:10:03,564
transform it from a reactive system

177
00:10:03,588 --> 00:10:06,750
where treatment is triggered
by symptoms of disease

178
00:10:06,774 --> 00:10:08,789
to a proactive system,

179
00:10:08,813 --> 00:10:12,027
where disease detection,
diagnosis and treatment

180
00:10:12,051 --> 00:10:13,935
can happen at early stage,

181
00:10:13,959 --> 00:10:16,058
way before any symptoms occur.

182
00:10:17,590 --> 00:10:20,976
Now this brings me to my last point,
and it's a fundamental one.

183
00:10:21,537 --> 00:10:23,526
What, exactly, is a disease?

184
00:10:24,590 --> 00:10:28,625
Imagine that breath analysis
can be commercialized as I describe it,

185
00:10:28,649 --> 00:10:31,255
and early detection becomes routine.

186
00:10:32,417 --> 00:10:34,721
A problem that remains
is, in fact, a problem

187
00:10:34,745 --> 00:10:37,733
that any screening activity has to face,

188
00:10:37,757 --> 00:10:39,578
because for many diseases,

189
00:10:39,602 --> 00:10:43,426
it is often impossible to predict
with sufficient certainty

190
00:10:43,450 --> 00:10:45,992
whether the disease
would ever cause any symptoms

191
00:10:46,016 --> 00:10:48,138
or put a person's life at risk.

192
00:10:48,990 --> 00:10:51,549
This is called overdiagnosis,

193
00:10:51,573 --> 00:10:53,211
and it leads to a dilemma.

194
00:10:53,927 --> 00:10:56,011
If a disease is identified,

195
00:10:56,035 --> 00:10:58,079
you could decide not to treat it,

196
00:10:58,103 --> 00:11:02,074
because there's a certain probability
that you would never suffer from it.

197
00:11:02,955 --> 00:11:04,454
But how much would you suffer

198
00:11:04,478 --> 00:11:07,865
just from knowing that you have
a potentially deadly disease?

199
00:11:07,889 --> 00:11:10,899
And wouldn't you actually regret
that the disease was detected

200
00:11:10,923 --> 00:11:12,080
in the first place?

201
00:11:13,281 --> 00:11:17,169
Your second option
is to undergo early treatment

202
00:11:17,193 --> 00:11:18,843
with the hope for curing it.

203
00:11:19,485 --> 00:11:22,314
But often, this would not come
without side effects.

204
00:11:25,623 --> 00:11:27,727
To be precise:

205
00:11:27,751 --> 00:11:30,325
the bigger problem is not overdiagnosis,

206
00:11:30,349 --> 00:11:31,802
it's overtreatment,

207
00:11:31,826 --> 00:11:34,841
because not every disease
has to be treated immediately

208
00:11:34,865 --> 00:11:36,925
just because a treatment is available.

209
00:11:39,441 --> 00:11:42,139
The increasing adoption
of routine screening

210
00:11:42,163 --> 00:11:43,912
will raise the question:

211
00:11:45,082 --> 00:11:49,173
What do we call a disease
that can rationalize treatment,

212
00:11:49,197 --> 00:11:53,785
and what is just an abnormality
that should not be a source of concern?

213
00:11:55,404 --> 00:12:00,336
My hopes are that routine screening
using breath analysis

214
00:12:00,360 --> 00:12:03,120
can provide enough data and insight

215
00:12:03,144 --> 00:12:07,690
so that at some point,
we'll be able to break this dilemma

216
00:12:07,714 --> 00:12:10,144
and predict with sufficient certainty

217
00:12:10,168 --> 00:12:12,858
whether and when to treat at early stage.

218
00:12:14,794 --> 00:12:19,513
Our breath and the mixture of volatile
organic compounds that we exhale

219
00:12:19,537 --> 00:12:24,207
hold tremendous amounts of information
on our physiological condition.

220
00:12:25,262 --> 00:12:28,602
With what we know today,
we have only scratched the surface.

221
00:12:29,993 --> 00:12:34,234
As we collect more and more data
and breath profiles across the population,

222
00:12:34,258 --> 00:12:38,790
including all varieties of gender,
age, origin and lifestyle,

223
00:12:38,814 --> 00:12:41,147
the power of breath analysis
should increase,

224
00:12:42,147 --> 00:12:47,725
and eventually, breath analysis
should provide us with a powerful tool

225
00:12:47,749 --> 00:12:52,215
not only to proactively detect
specific diseases,

226
00:12:52,239 --> 00:12:55,888
but to predict
and ultimately prevent them.

227
00:12:57,110 --> 00:12:59,728
And this should be enough motivation

228
00:12:59,752 --> 00:13:03,935
to embrace the opportunities
and challenges

229
00:13:03,959 --> 00:13:06,420
that breath analysis can provide,

230
00:13:06,444 --> 00:13:09,846
even for people that are not
part-time hypochondriacs like me.

231
00:13:10,532 --> 00:13:11,689
Thank you.

232
00:13:11,713 --> 00:13:15,730
(Applause)