Putting IBM Watson to Work.. Saxena

© 2012 IBM Corporation

IBM Watson Making a Market, Making a Difference Manoj Saxena General Manager, IBM Watson Solu:ons

© 2012 IBM Corporation 2

Result of IBM Research “Grand Challenge”

On February 14, 2011, IBM Watson made history


Brief history of IBM Watson

R&D

Demonstra1on

Commercializa1on

Cross-‐industry Applica1ons

IBM Research Project

(2006 – )

Jeopardy! Grand Challenge

(Feb 2011)

Watson for

Healthcare (Aug 2011 –)

Watson Industry Solu1ons (2012 – )

Watson for Financial Services

(Mar 2012 – )

Expansion

New Division


+ +

4

Intelligent Instrumented Interconnected

Stockholm leverages GPS data to predict traffic –

reducing conges:ons and

emissions.

Over 10 billion CPUs were produced in

2008, up 1000% in 8 years.

2 billion people were on the

Web by 2011 ... with a trillion connected objects.

The world is geTng smarter


Businesses are “dying of thirst in an ocean of data”

1 in 2 business leaders

don’t have access to data they need

83% of CIOs cited BI and

analy:cs as part of their visionary plan

2.2X more likely that top performers use business analy:cs

80% of the world’s data today is unstructured

90% of the world’s data was created in the last two years

20% is the amount of available data

tradi:onal systems leverages


So why is it so hard for computers to understand humans?

  Noses that run and feet that smell?   How can a house can burn up as it burns down?   Does CPD represent a complex comorbidity of lung cancer?   What mix of zero-coupon, non-callable, A+ munis fit my risk portfolio?

Person Organization

L. Gerstner IBM

J. Welch GE

W. Gates Microsoft

“If leadership is an art then surely Jack Welch has proved himself a master painter during his tenure at GE.”

Welch ran this?


99% 60% 10%

IBM Watson brings together transforma:onal technologies to drive op:mized outcomes

Understands natural language and human speech

Adapts and Learns from user responses

Generates and evaluates

hypothesis for beZer outcomes

3

2

1

…built on a massively parallel architecture op4mized for POWER7


Answer Scoring

Models

Responses with Confidence

Inquiry

Evidence Sources

Models

Models

Models

Models

Models Primary Search

Candidate Answer Generation

Hypothesis Generation

Hypothesis and Evidence Scoring

Final Confidence Merging & Ranking Synthesis

Answer Sources

Inquiry/Topic Analysis

Evidence Retrieval

Deep Evidence Scoring

Learned Models help combine and weigh the Evidence

Hypothesis Generation

Hypothesis and Evidence Scoring

Inquiry Decomposition

How Watson Works: DeepQA Architecture

1000’s of Pieces of Evidence

Multiple Interpretations of a question

100,000’s Scores from many Deep Analysis Algorithms

100’s sources

100’s Possible Answers

Balance & Combine


Moving beyond Jeopardy! is a non-‐trivial challenge

Watson at Play Watson at Work

1 User

Max. input was two sentences

5+ days to retrain

Evidence not present

Text-‐only input

Q&A model

Basic security

10s of thousands concurrent users

Pages of input (e.g. medical record)

Dynamic content inges:on

Suppor:ng evidence integral

Text, tables and images as input

Both Q&A + Conversa:on model

High security (e.g. HIPAA)


Healthcare industry is beset with some of the most complex informa:on challenges we collec:vely face

Medical informa:on is doubling every 5 years, much of which is unstructured 81% of physicians report spending 5 hours or less per month reading medical journals

“Medicine has become too complex. Only about 20% of the knowledge clinicians use today is evidence-‐base.” Steven Shapiro, Chief Medical & Scien1fic Officer, UPMC


Sym

ptom

s

UTI

Diabetes

Influenza

Hypokalemia

Renal Failure

no abdominal pain no back pain no cough no diarrhea

(Thyroid Autoimmune)

Esophagitis

pravastatin Alendronate

levothyroxine hydroxychloroquine

Diagnosis Models

frequent UTI

cutaneous lupus

hyperlipidemia osteoporosis

hypothyroidism

Confidence difficulty swallowing

dizziness

anorexia

fever dry mouth thirst

frequent urination

Fam

ily

His

tory

Graves’ Disease

Oral cancer Bladder cancer Hemochromatosis Purpura

Patie

nt

His

tory

M

edic

atio

ns

Find

ings

supine 120/80 mm HG

urine dipstick: leukocyte esterase

urine culture: E. Coli heart rate: 88 bpm

Symptoms A 58-year-old woman complains of

dizziness, anorexia, dry mouth, increased thirst, and frequent

urination. She had also had a fever. She reported no pain in her abdomen,

back, and no cough, or diarrhea.

A 58-year-old woman presented to her primary care physician after several days

of dizziness, anorexia, dry mouth, increased thirst, and frequent urination.

She had also had a fever and reported that food would “get stuck” when she was

swallowing. She reported no pain in her abdomen, back, or flank and no cough,

shortness of breath, diarrhea, or dysuria

Family History

Her family history included oral and bladder cancer in her mother, Graves' disease in two sisters,

hemochromatosis in one sister, and idiopathic thrombocytopenic

purpura in one sister

Patient History

Her history was notable for cutaneous lupus, hyperlipidemia, osteoporosis,

frequent urinary tract infections, a left oophorectomy for a benign cyst, and primary hypothyroidism, diagnosed a

year earlier

Her medications were levothyroxine, hydroxychloroquine, pravastatin, and

alendronate.

Medications Findings A urine dipstick was positive for

leukocyte esterase and nitrites. The patient given a prescription fo

ciprofloxacin for a urinary tract infection. 3 days later, patient

reported weakness and dizziness. Her supine blood pressure was

120/80 mm Hg, and pulse was 88.

•  Extract Symptoms from record •  Use paraphrasings mined from text to handle

alternate phrasings and variants •  Perform broad search for possible diagnoses •  Score Confidence in each diagnosis based on

evidence so far

•  Identify negative Symptoms •  Reason with mined relations to explain away

symptoms (thirst is consistent w/ UTI)

•  Extract Family History •  Use Medical Taxonomies to generalize medical

conditions to the granularity used by the models

•  Extract Patient History •  Extract Medications •  Use database of drug side-effects •  Together, multiple diagnoses may best explain

symptoms •  Extract Findings: Confirms that UTI was present

Most Confident Diagnosis: Diabetes

Most Confident Diagnosis: UTI Most Confident Diagnosis: Esophagitis

Most Confident Diagnosis: Influenza

PuTng the pieces together at point of impact can be life changing


1 in 3 individuals will die

from cancer

3X rate cancer cost climbs vs.

std. health costs or 15-‐18% / yr.

20% of cancer cases receive the wrong diagnosis ini:ally with some as high as 44%

$263.8B overall costs of cancer

in the US in 2010

Cancer is an insidious disease and the second highest cause of death

✔ ✔

✔

Source: American Cancer Society, National Health Institute

X

$$$$$$$$$$$$ $$$$$$$$$$$$ $$$$$$$$$$$$ $$$$$$$$$$$$ ✔ ✔

✔ ✔

IBM + +

Working Together to Beat Cancer

IBM + + Working Together to Beat Cancer


IBM Watson goes to work in healthcare

1. Accelerate Time to Clinical Insights Support researchers and clinicians in discovery of new cancer therapies

2. Improve Decisions and Outcomes

Assist physicians and care providers with evidence based diagnosis and treatment

Genomic Researcher Analy:cs

Expert

Therapy Designer

Care Provider

Pa:ent

Oncologist

MEDICAL RESEARCH MEDICAL PRACTICE & PAYMENTS

Ultimate Goal: Become the Most Essential Company


IBM Watson Oncology Advisor

IBM Confidential: References to potential future products are subject to the Important Disclaimer provided earlier in the presentation

Demonstra:on of Watson Cancer Care Solu:on


Decide

• Ingest and analyze domain sources, info models • Generate evidence based decisions with confidence • Learn with new outcomes and ac:ons • e.g. -‐ Next genera:on Apps Probabilis:c Apps

Ask

•  Leverage vast amounts of data •  Ask ques:ons for greater insights •  Natural language inquiries •  e.g. -‐ Next genera:on Chat

Watson enables three classes of cogni:ve solu:ons

Discover • Find the ra:onale for given answers • Prompt for inputs to yield improved responses • Inspire considera:ons of new ideas • e.g. -‐ Next genera:on Search Discovery


Imagine if…

… call center agents could find better answers to customer questions 50% faster. That’s exactly what a major provider of financial management software did.

ASK

“Contact centers of the future will improve precision and personalization, transforming centers from a cost orientation to a strategic assets.”

- Leading Telco Supplier


Imagine if…

. . . new insights from medical research find their way to patient treatment programs in months instead of years? That’s exactly what a global leader in cancer care is doing today.

DISCOVER

“Watson will be an invaluable resource for our physicians and will dramatically enhance the quality and effectiveness of medical care.”

- Dr Sam Nussbaum, Chief Medical Officer, WellPoint


Imagine if…

DECIDE

. . . the 1.5M people diagnosed with cancer in the US last year had a better prognosis? That’s exactly what a major health plan provider is working to accomplish. “Watson can aggregate information and give probabilities that will enable (experts) to zero in on the most likely diagnosis.”

- Dr. Steven Nissen, Cleveland Clinic


How it Works: Watson Technology & Infrastructure

Watson for Healthcare

Watson For Financial Svcs.

Watson for Client Engagement

Watson for Industry

Advisor Solu1ons Advisor Solu1ons Advisor Solu1ons

U1liza1

on

Oncology

Cardiac

Diab

etes

Bank

ing

Ins1tu1o

nal

Re1rem

ent

Ins1tu1o

n

Call Ce

nter

Help Desk

Know

ledge

Technical

Model Train Learn Source

Workload Optimized Systems

Analytics Mobile NLP & Machine Learning

Big Data Cloud

ASK Services DECISION Services DISCOVER Services


How it Works: Cloud Delivery and Outcome Based Pricing

Dynamic Capacity Automate and control service provisioning

Flexible Consump1on Support alterna:ve delivery and value

pricing models

Time to Value Enable incremental automa:on and business agility

Hybrid Delivery

Extend & integrate on-‐premise solu:on with cloud offering


Getting Started with Watson GeTng Started with Watson

1. Discovery Workshops (4-‐6 Weeks)

2. Pilots (6-‐12 Months)

3. Scale Out (Ongoing)

Op:on A:

Deploy ini:al pilot • Build • Teach • Run

Step 1:

• Iden:fy Candidate Use Cases

• Assess Content Availability

• Model Business Value

Op:on B:

Apply Ready for Watson Progression Paths

• Smarter Analy:cs • Big Data • Industry Solu:ons

Step 3:

• Add Users • Add Geographies • Add Content • Expand Processes

22 IBM Confidential © 2012 International Business Machines Corporation

System Intelligence

1900 1950 2011

1

1

Watson is ushering in a new era of compu:ng . . .

. . .enabling new opportuni:es and outcomes

Tabula:on

Programma:c

Cogni1ve

Punch cards Time card readers

Search Determinis:c Enterprise data Machine language Simple outputs

Discovery Probabilis:c Big Data Natural language Intelligent op:ons


Thank you.

Putting IBM Watson to Work.. Saxena

Technology

Transcript of Putting IBM Watson to Work.. Saxena