0:00:00.667,0:00:02.770
So, well, I do applied math,

0:00:02.770,0:00:04.294
and this is a peculiar problem

0:00:04.294,0:00:06.467
for anyone who does applied math, is that

0:00:06.467,0:00:08.400
we are like management consultants.

0:00:08.400,0:00:10.346
No one knows what the hell we do.

0:00:10.346,0:00:12.620
So I am going to give you some -- attempt today

0:00:12.620,0:00:14.913
to try and explain to you what I do.

0:00:14.913,0:00:18.234
So, dancing is one of the most human of activities.

0:00:18.234,0:00:21.916
We delight at ballet virtuosos and tap dancers

0:00:21.916,0:00:23.064
you will see later on.

0:00:23.064,0:00:25.754
Now, ballet requires an extraordinary level of expertise

0:00:25.754,0:00:28.668
and a high level of skill,

0:00:28.668,0:00:31.199
and probably a level of initial suitability

0:00:31.199,0:00:33.046
that may well have a genetic component to it.

0:00:33.046,0:00:36.439
Now, sadly, neurological disorders such as Parkinson's disease

0:00:36.439,0:00:38.526
gradually destroy this extraordinary ability,

0:00:38.526,0:00:40.849
as it is doing to my friend Jan Stripling, who was

0:00:40.849,0:00:43.816
a virtuoso ballet dancer in his time.

0:00:43.816,0:00:46.870
So great progress and treatment has been made over the years.

0:00:46.870,0:00:49.814
However, there are 6.3 million people worldwide

0:00:49.814,0:00:53.262
who have the disease, and they have to live with

0:00:53.262,0:00:55.830
incurable weakness, tremor, rigidity

0:00:55.830,0:00:57.687
and the other symptoms that go along with the disease,

0:00:57.687,0:01:00.070
so what we need are objective tools

0:01:00.070,0:01:03.127
to detect the disease before it's too late.

0:01:03.127,0:01:05.681
We need to be able to measure progression objectively,

0:01:05.681,0:01:08.854
and ultimately, the only way we're going to know

0:01:08.854,0:01:11.046
when we actually have a cure is when we have

0:01:11.046,0:01:14.444
an objective measure that can answer that for sure.

0:01:14.444,0:01:17.294
But frustratingly, with Parkinson's disease

0:01:17.294,0:01:19.647
and other movement disorders, there are no biomarkers,

0:01:19.647,0:01:21.876
so there's no simple blood test that you can do,

0:01:21.876,0:01:23.678
and the best that we have is like

0:01:23.678,0:01:25.919
this 20-minute neurologist test.

0:01:25.919,0:01:28.377
You have to go to the clinic to do it. It's very, very costly,

0:01:28.377,0:01:31.134
and that means that, outside the clinical trials,

0:01:31.134,0:01:33.862
it's just never done. It's never done.

0:01:33.862,0:01:36.939
But what if patients could do this test at home?

0:01:36.939,0:01:39.037
Now, that would actually save on a difficult trip to the clinic,

0:01:39.037,0:01:43.291
and what if patients could do that test themselves, right?

0:01:43.291,0:01:45.211
No expensive staff time required.

0:01:45.211,0:01:46.629
Takes about $300, by the way,

0:01:46.629,0:01:48.622
in the neurologist's clinic to do it.

0:01:48.622,0:01:51.303
So what I want to propose to you as an unconventional way

0:01:51.303,0:01:52.817
in which we can try to achieve this,

0:01:52.817,0:01:54.625
because, you see, in one sense, at least,

0:01:54.625,0:01:57.881
we are all virtuosos like my friend Jan Stripling.

0:01:57.881,0:02:01.636
So here we have a video of the vibrating vocal folds.

0:02:01.636,0:02:04.865
Now, this is healthy and this is somebody making speech sounds,

0:02:04.865,0:02:08.329
and we can think of ourselves as vocal ballet dancers,

0:02:08.329,0:02:10.543
because we have to coordinate all of these vocal organs

0:02:10.543,0:02:12.838
when we make sounds, and we all actually

0:02:12.838,0:02:15.134
have the genes for it. FoxP2, for example.

0:02:15.134,0:02:17.847
And like ballet, it takes an extraordinary level of training.

0:02:17.847,0:02:20.432
I mean, just think how long it takes a child to learn to speak.

0:02:20.432,0:02:22.814
From the sound, we can actually track

0:02:22.814,0:02:25.095
the vocal fold position as it vibrates,

0:02:25.095,0:02:27.638
and just as the limbs are affected in Parkinson's,

0:02:27.638,0:02:30.419
so too are the vocal organs.

0:02:30.419,0:02:32.299
So on the bottom trace, you can see an example of

0:02:32.299,0:02:33.997
irregular vocal fold tremor.

0:02:33.997,0:02:35.165
We see all the same symptoms.

0:02:35.165,0:02:38.095
We see vocal tremor, weakness and rigidity.

0:02:38.095,0:02:40.199
The speech actually becomes quieter and more breathy

0:02:40.199,0:02:42.432
after a while, and that's one of the example symptoms of it.

0:02:42.432,0:02:45.279
So these vocal effects can actually be quite subtle,

0:02:45.279,0:02:48.495
in some cases, but with any digital microphone,

0:02:48.495,0:02:51.040
and using precision voice analysis software

0:02:51.040,0:02:53.449
in combination with the latest in machine learning,

0:02:53.449,0:02:55.027
which is very advanced by now,

0:02:55.027,0:02:57.913
we can now quantify exactly where somebody lies

0:02:57.913,0:03:00.794
on a continuum between health and disease

0:03:00.794,0:03:03.390
using voice signals alone.

0:03:03.390,0:03:05.704
So these voice-based tests, how do they stack up against

0:03:05.704,0:03:07.854
expert clinical tests? We'll, they're both non-invasive.

0:03:07.854,0:03:11.836
The neurologist's test is non-invasive. They both use existing infrastructure.

0:03:11.836,0:03:14.840
You don't have to design a whole new set of hospitals to do it.

0:03:14.840,0:03:17.142
And they're both accurate. Okay, but in addition,

0:03:17.142,0:03:20.469
voice-based tests are non-expert.

0:03:20.469,0:03:22.461
That means they can be self-administered.

0:03:22.461,0:03:25.041
They're high-speed, take about 30 seconds at most.

0:03:25.041,0:03:27.335
They're ultra-low cost, and we all know what happens.

0:03:27.335,0:03:29.775
When something becomes ultra-low cost,

0:03:29.775,0:03:32.071
it becomes massively scalable.

0:03:32.071,0:03:35.746
So here are some amazing goals that I think we can deal with now.

0:03:35.746,0:03:38.172
We can reduce logistical difficulties with patients.

0:03:38.172,0:03:40.484
No need to go to the clinic for a routine checkup.

0:03:40.484,0:03:42.804
We can do high-frequency monitoring to get objective data.

0:03:42.804,0:03:46.909
We can perform low-cost mass recruitment for clinical trials,

0:03:46.909,0:03:49.024
and we can make population-scale screening

0:03:49.024,0:03:50.620
feasible for the first time.

0:03:50.620,0:03:52.822
We have the opportunity to start to search

0:03:52.822,0:03:56.363
for the early biomarkers of the disease before it's too late.

0:03:56.363,0:03:59.121
So, taking the first steps towards this today,

0:03:59.121,0:04:01.247
we're launching the Parkinson's Voice Initiative.

0:04:01.247,0:04:03.479
With Aculab and PatientsLikeMe, we're aiming

0:04:03.479,0:04:05.407
to record a very large number of voices worldwide

0:04:05.407,0:04:08.547
to collect enough data to start to tackle these four goals.

0:04:08.547,0:04:10.247
We have local numbers accessible to three quarters

0:04:10.247,0:04:11.857
of a billion people on the planet.

0:04:11.857,0:04:14.934
Anyone healthy or with Parkinson's can call in, cheaply,

0:04:14.934,0:04:17.073
and leave recordings, a few cents each,

0:04:17.073,0:04:19.263
and I'm really happy to announce that we've already hit

0:04:19.263,0:04:22.806
six percent of our target just in eight hours.

0:04:22.806,0:04:26.557
Thank you. (Applause)

0:04:26.557,0:04:32.877
(Applause)

0:04:32.877,0:04:36.452
Tom Rielly: So Max, by taking all these samples of,

0:04:36.452,0:04:39.228
let's say, 10,000 people,

0:04:39.228,0:04:42.082
you'll be able to tell who's healthy and who's not?

0:04:42.082,0:04:43.767
What are you going to get out of those samples?

0:04:43.767,0:04:45.597
Max Little: Yeah. Yeah. So what will happen is that,

0:04:45.597,0:04:47.254
during the call you have to indicate whether or not

0:04:47.254,0:04:48.521
you have the disease or not, you see. TR: Right.

0:04:48.521,0:04:51.028
ML: You see, some people may not do it. They may not get through it.

0:04:51.028,0:04:53.745
But we'll get a very large sample of data that is collected

0:04:53.745,0:04:57.153
from all different circumstances, and it's getting it

0:04:57.153,0:04:59.058
in different circumstances that matter because then

0:04:59.058,0:05:02.442
we are looking at ironing out the confounding factors,

0:05:02.442,0:05:04.603
and looking for the actual markers of the disease.

0:05:04.603,0:05:07.100
TR: So you're 86 percent accurate right now?

0:05:07.100,0:05:08.294
ML: It's much better than that.

0:05:08.294,0:05:10.014
Actually, my student Thanasis, I have to plug him,

0:05:10.014,0:05:11.884
because he's done some fantastic work,

0:05:11.884,0:05:15.654
and now he has proved that it works over the mobile telephone network as well,

0:05:15.654,0:05:19.044
which enables this project, and we're getting 99 percent accuracy.

0:05:19.044,0:05:20.620
TR: Ninety-nine. Well, that's an improvement.

0:05:20.620,0:05:22.821
So what that means is that people will be able to —

0:05:22.821,0:05:24.673
ML: (Laughs)

0:05:24.673,0:05:26.579
TR: People will be able to call in from their mobile phones

0:05:26.579,0:05:29.651
and do this test, and people with Parkinson's could call in,

0:05:29.651,0:05:32.521
record their voice, and then their doctor can check up

0:05:32.521,0:05:35.202
on their progress, see where they're doing in this course of the disease.

0:05:35.202,0:05:36.172
ML: Absolutely.

0:05:36.172,0:05:37.915
TR: Thanks so much. Max Little, everybody.

0:05:37.915,0:05:43.072
ML: Thanks, Tom. (Applause)