ABOUT THE SPEAKER
Pratik Shah - Medical technologist
Dr. Pratik Shah creates novel intersections between engineering, medical imaging, machine learning and medicine.

Why you should listen

Dr. Shah's research program at the MIT Media Lab develops scalable and low-cost diagnostics and therapeutics. His ongoing research areas at MIT include: 1) artificial intelligence and machine learning methods for detection of cancer biomarkers using standard photographs vs. expensive medical images; 2) unorthodox artificial intelligence and machine learning algorithms to design optimal and faster clinical trials and to reduce adverse effects on patients; and 3) low-cost and open source imaging devices, paper diagnostics, algorithms and mobile phones to improve public health and generate real-world data.

Clinical studies with Pratik's medical technologies have revealed "missing sick" patients, who otherwise remain undiagnosed in conventional healthcare settings. Dr. Shah's graduate and postdoctoral research contributed to the discovery of a vaccine component to prevent pneumococcal (Streptococcus pneumoniae) diseases; the identification of new pathways, technologies and metabolites as antimicrobials to target gastrointestinal infections; and a nonprofit to deploy a low-cost water quality test for the developing world.

Past recognition for Dr. Shah includes the American Society for Microbiology's Raymond W. Sarber national award, the Harvard Medical School and Massachusetts General Hospitals ECOR Fund for Medical Discovery postdoctoral fellowship, the AAAS-Lemelson Invention Ambassador Award and a TED Fellowship. Pratik has been an invited discussion leader at Gordon Research Seminars; a speaker at Cold Spring Harbor Laboratories, Gordon Research Conferences and IEEE bioengineering conferences; and a peer reviewer for leading scientific publications and funding agencies. Pratik has a BS, MS, and a PhD in Microbiology and completed fellowship training at The Broad Institute of MIT and Harvard, Massachusetts General Hospital and Harvard Medical School.

More profile about the speaker
Pratik Shah | Speaker | TED.com
TEDGlobal 2017

Pratik Shah: How AI is making it easier to diagnose disease

Pratik Shah: Como a inteligência artificial está facilitando o diagnóstico de doenças

Filmed:
1,571,835 views

Os algoritmos de inteligência artificial atuais exigem dezenas de milhares de imagens médicas caras para detectar a doença de um paciente. E se pudéssemos reduzir drasticamente a quantidade de dados necessários para capacitar uma IA, fazendo diagnósticos de baixo custo e mais eficazes? O bolsista TED Pratik Shah está trabalhando em um sistema inteligente para fazer exatamente isso. Usando uma abordagem de IA pouco convencional, Shah desenvolveu uma tecnologia que requer apenas 50 imagens para desenvolver um algoritmo de trabalho e pode até usar fotos tiradas em telefones celulares de médicos para fornecer um diagnóstico. Saiba mais sobre como essa nova maneira de analisar informações médicas pode levar à detecção precoce de doenças potencialmente fatais e levar o diagnóstico auxiliado por IA a mais locais de atendimento médico em todo o mundo.
- Medical technologist
Dr. Pratik Shah creates novel intersections between engineering, medical imaging, machine learning and medicine. Full bio

Double-click the English transcript below to play the video.

00:13
Computer algorithms today
are performing incredible tasks
0
1280
3856
Os algoritmos de computador hoje
estão realizando tarefas incríveis
00:17
with high accuracies, at a massive scale,
using human-like intelligence.
1
5160
4736
com alta precisão, em larga escala,
usando inteligência semelhante à humana.
00:21
And this intelligence of computers
is often referred to as AI
2
9920
3936
Essa inteligência dos computadores
é muitas vezes apresentada como IA
00:25
or artificial intelligence.
3
13880
1856
ou inteligência artificial.
00:27
AI is poised to make an incredible impact
on our lives in the future.
4
15760
4200
A IA está pronta para causar um impacto
incrível em nossa vida no futuro.
00:32
Today, however,
we still face massive challenges
5
20880
3936
Hoje, no entanto,
ainda enfrentamos enormes desafios
00:36
in detecting and diagnosing
several life-threatening illnesses,
6
24840
3496
na detecção e no diagnóstico
de várias doenças potencialmente fatais,
00:40
such as infectious diseases and cancer.
7
28360
2360
como as doenças infecciosas e o câncer.
00:44
Thousands of patients every year
8
32000
2296
Milhares de pacientes, todos os anos,
00:46
lose their lives
due to liver and oral cancer.
9
34320
2800
perdem a vida devido ao câncer
de fígado e de boca.
00:49
Our best way to help these patients
10
37880
2696
Nossa melhor maneira
de ajudar esses pacientes
00:52
is to perform early detection
and diagnoses of these diseases.
11
40600
4320
é fazendo a detecção precoce
e o diagnóstico dessas doenças.
00:57
So how do we detect these diseases today,
and can artificial intelligence help?
12
45880
4160
Como podemos detectar essas doenças hoje,
e a inteligência artificial pode ajudar?
01:03
In patients who, unfortunately,
are suspected of these diseases,
13
51920
3656
Para pacientes que, infelizmente,
são suspeitos de terem essas doenças,
01:07
an expert physician first orders
14
55600
2656
um médico especialista pede primeiro
a realização de exames de imagem caros,
01:10
very expensive
medical imaging technologies
15
58280
2616
01:12
such as fluorescent imaging,
CTs, MRIs, to be performed.
16
60920
4096
tais como imagens fluorescentes,
tomografias, imagens de ressonâncias.
01:17
Once those images are collected,
17
65040
2296
Assim que as imagens são coletadas,
01:19
another expert physician then diagnoses
those images and talks to the patient.
18
67360
4520
outro médico especialista faz
o diagnóstico delas e fala com o paciente.
01:24
As you can see, this is
a very resource-intensive process,
19
72520
3456
Como podem ver, é um processo
de recursos muito dispendioso,
01:28
requiring both expert physicians,
expensive medical imaging technologies,
20
76000
4416
que exige médicos especialistas
e exames de imagem caros,
01:32
and is not considered practical
for the developing world.
21
80440
3096
e não é considerado prático
para os países em desenvolvimento
01:35
And in fact, in many
industrialized nations, as well.
22
83560
3360
nem, de fato, para muitos
países industrializados.
01:39
So, can we solve this problem
using artificial intelligence?
23
87760
2880
Podemos resolver esse problema
usando inteligência artificial?
01:43
Today, if I were to use traditional
artificial intelligence architectures
24
91840
4056
Hoje, se eu fosse utilizar arquiteturas
tradicionais de inteligência artificial
01:47
to solve this problem,
25
95920
1216
para resolver o problema,
eu solicitaria primeiro 10 mil,
01:49
I would require 10,000 --
26
97160
1456
01:50
I repeat, on an order of 10,000
of these very expensive medical images
27
98640
4016
repito, 10 mil dessas imagens
médicas muito caras.
01:54
first to be generated.
28
102680
1376
01:56
After that, I would then go
to an expert physician,
29
104080
2896
Depois disso, eu iria
a um médico especialista,
01:59
who would then analyze
those images for me.
30
107000
2496
que, então, analisaria
essas imagens para mim.
02:01
And using those two pieces of information,
31
109520
2096
Usando essas duas informações,
02:03
I can train a standard deep neural network
or a deep learning network
32
111640
3656
posso capacitar uma rede neural
ou de aprendizagem profunda padrão
02:07
to provide patient's diagnosis.
33
115320
2136
a fornecer o diagnóstico do paciente.
02:09
Similar to the first approach,
34
117480
1736
Semelhante à primeira abordagem,
02:11
traditional artificial
intelligence approaches
35
119240
2143
as abordagens tradicionais de inteligência
artificial sofrem do mesmo problema.
02:13
suffer from the same problem.
36
121407
1449
02:14
Large amounts of data, expert physicians
and expert medical imaging technologies.
37
122880
4560
Grandes quantidades de dados,
médicos especialistas
e tecnologias especializadas
de imagem médica,
02:20
So, can we invent more scalable, effective
38
128320
4296
Será que podemos inventar
arquiteturas de inteligência artificial
02:24
and more valuable artificial
intelligence architectures
39
132640
3296
mais valiosas, escaláveis e eficazes
02:27
to solve these very important
problems facing us today?
40
135960
3056
para resolver esses problemas
muito importantes que enfrentamos hoje?
02:31
And this is exactly
what my group at MIT Media Lab does.
41
139040
3296
É exatamente isso o que meu grupo
do MIT Media Lab faz.
02:34
We have invented a variety
of unorthodox AI architectures
42
142360
3856
Temos inventado uma variedade
de arquiteturas de IA pouco convencionais
02:38
to solve some of the most important
challenges facing us today
43
146240
3176
para resolver alguns dos desafios
mais importantes que enfrentamos hoje
02:41
in medical imaging and clinical trials.
44
149440
2200
em exames de imagem e ensaios clínicos.
No exemplo que compartilhei hoje
com vocês, tínhamos dois objetivos.
02:44
In the example I shared
with you today, we had two goals.
45
152480
3056
02:47
Our first goal was to reduce
the number of images
46
155560
2976
O primeiro era reduzir o número de imagens
02:50
required to train
artificial intelligence algorithms.
47
158560
3256
necessárias para capacitar
os algoritmos de inteligência artificial.
02:53
Our second goal -- we were more ambitious,
48
161840
2096
O segundo objetivo era mais ambicioso:
02:55
we wanted to reduce the use
of expensive medical imaging technologies
49
163960
3736
reduzir o uso de tecnologias caras
de imagem médica para examinar pacientes.
02:59
to screen patients.
50
167720
1216
03:00
So how did we do it?
51
168960
1200
Como fizemos isso?
03:02
For our first goal,
52
170920
1216
Para o primeiro objetivo, em vez
de começarmos com dezenas e milhares
03:04
instead of starting
with tens and thousands
53
172160
2056
03:06
of these very expensive medical images,
like traditional AI,
54
174240
3016
de imagens muito caras,
como a IA tradicional,
03:09
we started with a single medical image.
55
177280
2056
começamos com uma única imagem.
03:11
From this image, my team and I
figured out a very clever way
56
179360
3776
A partir dela, minha equipe e eu
descobrimos uma maneira muito inteligente
03:15
to extract billions
of information packets.
57
183160
2736
de extrair bilhões de pacotes
de informação.
03:17
These information packets
included colors, pixels, geometry
58
185920
3696
Esses pacotes incluíam
cores, pixels, geometria
03:21
and rendering of the disease
on the medical image.
59
189640
2536
e renderização da doença na imagem médica.
03:24
In a sense, we converted one image
into billions of training data points,
60
192200
4336
De certa forma, convertemos uma imagem
em bilhões de pontos de dados de formação,
03:28
massively reducing the amount of data
needed for training.
61
196560
3536
reduzindo bastante a quantidade
de dados necessários para a formação.
03:32
For our second goal,
62
200120
1216
Para o segundo objetivo,
03:33
to reduce the use of expensive medical
imaging technologies to screen patients,
63
201360
3856
reduzir o uso de tecnologias caras
de imagem médica para examinar pacientes,
03:37
we started with a standard,
white light photograph,
64
205240
2856
começamos com uma fotografia
padrão, de luz branca,
03:40
acquired either from a DSLR camera
or a mobile phone, for the patient.
65
208120
4336
obtida a partir de uma câmera DSLR
ou de um telefone celular para o paciente.
03:44
Then remember those
billions of information packets?
66
212480
2456
Lembram-se dos bilhões
de pacotes de informação?
Sobrepusemos os da imagem
médica a essa imagem,
03:46
We overlaid those from
the medical image onto this image,
67
214960
3536
03:50
creating something
that we call a composite image.
68
218520
2520
criando algo que chamamos
de imagem composta.
03:53
Much to our surprise,
we only required 50 --
69
221480
3296
Para nossa surpresa,
precisamos apenas de 50,
03:56
I repeat, only 50 --
70
224800
1336
repito, apenas 50
03:58
of these composite images to train
our algorithms to high efficiencies.
71
226160
3840
dessas imagens compostas para capacitar
nossos algoritmos para altos rendimentos.
Para resumir nossa abordagem,
04:02
To summarize our approach,
72
230680
1336
04:04
instead of using 10,000
very expensive medical images,
73
232040
3176
em vez de usarmos 10 mil
imagens médicas muito caras,
podemos capacitar os algoritmos de IA
de um modo pouco convencional,
04:07
we can now train the AI algorithms
in an unorthodox way,
74
235240
3016
04:10
using only 50 of these high-resolution,
but standard photographs,
75
238280
4256
usando apenas 50 dessas fotografias
de alta resolução, porém, padrão,
04:14
acquired from DSLR cameras
and mobile phones,
76
242560
2496
obtidas de câmeras DSLR
e telefones celulares,
04:17
and provide diagnosis.
77
245080
1536
e fornecer o diagnóstico.
04:18
More importantly,
78
246640
1216
Principalmente,
04:19
our algorithms can accept,
in the future and even right now,
79
247880
3136
nossos algoritmos podem aceitar,
no futuro e até neste momento,
04:23
some very simple, white light
photographs from the patient,
80
251040
2816
algumas fotos muito simples,
de luz branca, do paciente,
04:25
instead of expensive
medical imaging technologies.
81
253880
2440
em vez de tecnologias
de imagens médicas caras.
04:29
I believe that we are poised
to enter an era
82
257120
3096
Acredito que estamos prontos
para entrar em uma era
04:32
where artificial intelligence
83
260240
1936
em que a inteligência artificial
04:34
is going to make an incredible
impact on our future.
84
262200
2536
irá causar um impacto incrível
em nosso futuro.
04:36
And I think that thinking
about traditional AI,
85
264760
2456
Quando pensamos na IA tradicional,
04:39
which is data-rich but application-poor,
86
267240
2776
rica em dados, mas pobre em aplicativos,
04:42
we should also continue thinking
87
270040
1536
devemos também continuar pensando
04:43
about unorthodox artificial
intelligence architectures,
88
271600
3016
em arquiteturas pouco convencionais
de inteligência artificial,
04:46
which can accept small amounts of data
89
274640
1936
que aceitem pequenas
quantidades de dados
04:48
and solve some of the most important
problems facing us today,
90
276600
2936
e resolvam problemas importantes,
especialmente na assistência médica.
04:51
especially in health care.
91
279560
1256
04:52
Thank you very much.
92
280840
1216
Muito obrigado.
04:54
(Applause)
93
282080
3840
(Aplausos)
Translated by Maurício Kakuei Tanaka
Reviewed by Maricene Crus

▲Back to top

ABOUT THE SPEAKER
Pratik Shah - Medical technologist
Dr. Pratik Shah creates novel intersections between engineering, medical imaging, machine learning and medicine.

Why you should listen

Dr. Shah's research program at the MIT Media Lab develops scalable and low-cost diagnostics and therapeutics. His ongoing research areas at MIT include: 1) artificial intelligence and machine learning methods for detection of cancer biomarkers using standard photographs vs. expensive medical images; 2) unorthodox artificial intelligence and machine learning algorithms to design optimal and faster clinical trials and to reduce adverse effects on patients; and 3) low-cost and open source imaging devices, paper diagnostics, algorithms and mobile phones to improve public health and generate real-world data.

Clinical studies with Pratik's medical technologies have revealed "missing sick" patients, who otherwise remain undiagnosed in conventional healthcare settings. Dr. Shah's graduate and postdoctoral research contributed to the discovery of a vaccine component to prevent pneumococcal (Streptococcus pneumoniae) diseases; the identification of new pathways, technologies and metabolites as antimicrobials to target gastrointestinal infections; and a nonprofit to deploy a low-cost water quality test for the developing world.

Past recognition for Dr. Shah includes the American Society for Microbiology's Raymond W. Sarber national award, the Harvard Medical School and Massachusetts General Hospitals ECOR Fund for Medical Discovery postdoctoral fellowship, the AAAS-Lemelson Invention Ambassador Award and a TED Fellowship. Pratik has been an invited discussion leader at Gordon Research Seminars; a speaker at Cold Spring Harbor Laboratories, Gordon Research Conferences and IEEE bioengineering conferences; and a peer reviewer for leading scientific publications and funding agencies. Pratik has a BS, MS, and a PhD in Microbiology and completed fellowship training at The Broad Institute of MIT and Harvard, Massachusetts General Hospital and Harvard Medical School.

More profile about the speaker
Pratik Shah | Speaker | TED.com