IT as an Enabler for Translating Quality Research to Optimized Healthcare Delivery

IT as an Enabler for Translating Quality Research to 
Optimized Healthcare Delivery 
Luu Danh Anh Vu (Vu Luu) 
Country Manager for Technology Solutions 
IBM Vietnam 
(vuluu@vn.ibm.com) 
2010 2020 
Sensors 
& Devices 
Natural 
Language 
Enterprise 
Data 
Medical 
Images 
Images/ 
Multimedia 
You are here 
44 zettabytes 
Growing data volume and complexity demands a new approach. 
Tabulating 
Systems Era 
1900 – 1940s 
Programmable 
Systems Era 
1950s – Present 
Cognitive 
Computing Era 
2011 – 
3 
Cognitive systems expand the problems we can address 
Cognitive Systems 
• Leverage traditional data sources 
• Follow pre-defined rules (programs) 
• Provide the same output to all users 
• Are taught, not programmed. 
• Learn and improve based on experience 
• Interpret sensory and non-traditional data 
• Relate to each of us as individuals 
• Allow us to expand and scale our own thinking 
Programmatic Systems 
Generation and Delivery of Evidence and Insights 
Published 
Knowledge 
Knowledge-Driven Method Data-Driven Method 
Observational 
Data 
• Longitudinal records 
• Claims, Rx, Labs 
• Patient reported data 
• Scientific papers 
• Books 
• Guidelines 
Closing the translational knowledge gap Personalized Insights from institutional data 
From population averages  To insights for individual patient! 
Learns 
Decisions made by leading experts feed the 
engine. Watson learns & improves over time. 
Understands 
Watson can read & understand 
documents & data – both structured & 
unstructured – at a massive scale. 
Reasons 
Watson searches & analyzes 
data, returning evidence-based 
recommendations. 
WATSON: A COGNITIVE SYSTEM 
Why Now? The Healthcare Disruption 
$47 trillion 
Estimated global economic 
impact of chronic disease by 
20303 
75%+ 
Percentage of patients expected 
to use digital health services in 
the future5 
24 months 
Frequency at which 
healthcare data doubles1 
150+ exabytes 
Amount of healthcare 
data today2 
Sources: McKinsey&Company, Centers for Medicare and Medicaid Services, Centers for Disease Control and Prevention 
© 2016 International Business Machines Corporation 7 
6 Terabytes 
Per lifetime 
Volume, Variety, Velocity, 
Veracity 
60% 
Exogenous Factors 
1100 Terabytes 
Generated per lifetime 
0.4 Terabytes 
Per lifetime 
30% 
Genomics Factors 
10% 
Clinical Factors 
A vast amount of untapped data could have a great impact on 
our health — yet it exists outside medical systems 
Rethinking Oncology 
By 2025, overall demand 
for medical oncology 
services will grow 42%. 
The number of oncologists 
will likely grow by only 28%. 
9 
© 2016 International Business Machines Corporation 10 
WATSON ONCOLOGY HELPS MEDICAL 
ONCOLOGISTS AND THEIR CARE TEAMS 
ADDRESS THESE CHALLENGES 
61 y/o woman s/p 
mastectomy is here to 
discuss treatment options 
for a recently diagnosed 
4.2 cm grade 2 infiltrating 
ductal carcinoma 
Prioritized Treatment 
Options 
+ 
Evidence Profile 
Patient Case • Inclusion / exclusion 
criteria 
• Co morbidities 
• Contraindications 
• FDA risk factors 
• MSK preferred 
treatments 
• Other guidelines 
• Published literature - 
studies, reports, 
opinions from Text 
Books, Journals, 
Manuals, etc. 
Evidence 
Watson Oncology 
Key Case 
Attributes 
Candidate 
Treatment Options 
Supportin
g 
Evidence 
Extract key attributes 
from a patient’s case 
1 
Use those attributes to find 
candidate treatment 
options as determined by 
consulting NCCN 
Guidelines 
2 
Use Watson’s analytic 
algorithms to prioritize 
treatment options based on 
best evidence. 
4 
Guidelines 
Search a corpus of 
evidence data to find 
supporting evidence for 
each option 
3 
Medications 
Symptoms Diseases 
Modifiers 
Natural Language Processing 
We use Natural Language Processing and UMLS (Unified Medical Language 
System) CUIs (Concept Unique IDs) to recognize medical concepts 
Rethinking Genomic Medicine 
Next-generation 
sequencing (NGS) data 
streams range between 
1 and 10 terabytes. 
800 Billion base pairs 
of DNA to analyze 
one brain tumor 
13 
 Pubmed Abstracts (23M) 
Drugbank.ca 
Clinicaltrials.gov 
Geneontology.org 
COSMIC from Sanger 
Uniprot.org 
Ensembl.org 
NCI PID 
Clinvar 
Genenames.org (HUGO) 
Drugs@FDA 
dbNSNP 
TCGA 
NCI Thesaurus 
NCI Drug Info 
NCI Drug Dictionary 
Elsevier Gold Standards 
Select whole text journal articles 
WGA Content 
Molecular Profile Analysis 
Pathway Analysis 
Drug Analysis 
Watson Genomics Analytics 
16 partners 
Rethinking Clinical Trial Matching 
30% of sites for clinical 
trials fail in enrolling even 
a single patient. 
Approximately 3% of adult 
cancer patients participate in 
clinical trials. 
14 
16 
IBM Watson Health // ©2015 International Business Machines Corporation #Watson Health 
Watson Health - Vision 
WATSON DISCOVERY ADVISOR 
Watson solution: 
Making linkages that unlock insights 
Which accelerate breakthroughs 
in 
• Disease understanding 
• Drug discovery 
• Toxicity assessment (early safety) 
• Trial design 
• Comparative effectiveness 
• Pharmacovigilance (drug safety) 
16M+ patents 
23M+ abstracts 
100+ journals 
50+ books 
11,000+ drugs 
20,000+ genes 
12M+ chemical 
structures Watson Corpus 
Over 1TB of data 
Over 40m 
documents 
Over 100m entities 
and relationships 
Available External Data 
Chemical database 
Public genomics 
Medical textbooks 
Medline 
Other journals 
FDA 
drugs/labels 
Patents 
17 
© 2014 International Business Machines Corporation 
Business challenge: 
• Researchers can’t innovate fast enough to create truly breakthrough 
therapies 
• They struggle to anticipate the safety profile of new treatments and design 
trials that demonstrate efficacy and safety 
WATSON DISCOVERY ADVISOR: 
ACCELERATING BREAKTHROUGH INSIGHTS ACROSS LIFE SCIENCE FUNCTIONS 
• What new ways could we target this 
disease pathway? 
Let’s look at all the genes identified in 
every disease that are activated by this 
protein 
Lead & Drug Discovery 
• How can we quickly identify if this 
compound has a toxicity issue? 
Signals from internal toxicology reports 
and published studies suggest this 
compound may cause serious AEs 
Safety & Toxicity 
Assessment 
• Are there reasons for the early safety 
signals that we can quickly identify? 
AE reports suggest that our drug is 
often being taken with dairy foods 
when this side effect is being reported 
Pharmacovigilance 
• Does this drug have an effect on the 
pathway of another disease? 
There are several diseases where the 
same receptors that this compound 
binds to exist 
Drug Repurposing 
• What populations are likely to benefit 
most from this intervention? 
Looking at all known studies of similar 
compounds, this is how this treatment 
might perform in these populations 
Comparative Effectiveness / 
Clinical Trial Design 
• What do early studies of competitors 
reveal about their efficacy and safety? 
Animal models revealed early 
effectiveness and faster onset, 
differentiating from current products 
Competitive Intelligence 
Reducing CHF readmission to improve care 
Seton Healthcare strives to reduce the occurrence 
of high cost Congestive Heart Failure (CHF) 
readmissions by proactively identifying patients 
likely to be readmitted on an emergency basis. 
How? 
Utilizing natural language processing to extract key 
elements from unstructured History and Physical, 
Discharge Summaries, Echocardiogram Reports, and 
Consult Notes 
Featured on 
1. Jugular Venous Distention (JVD ↑) Indicator 
2. Paid by Medicaid Indicator 
3. Immunity Disorder Disease Indicator 
4. Cardiac Rehab Admit Diagnosis with CHF Indicator 
5. Lack of Emotional Support Indicator 
6. Self COPD Moderate Limit Health History Indicator 
7. With Genitourinary System and Endocrine Disorders 
8. Heart Failure History 
9. High BNP Indicator 
10. Low Hemoglobin Indicator 
11. Low Sodium Level Indicator 
12. Assisted Living 
13. High Cholesterol History 
21 
Top predictors of Readmission 
Phase 1 
Stream computing provides 
real time analytical insights 
and notifications to nurses 
for critical trends 
- Physiologic monitor data 
- Laboratory results 
- Radiology results - 
Patient in 
ICU
EMR will be pulled 
in using natural 
language 
processing and 
content analytics 
Watson will be 
able to identify 
risks along with 
streaming data to 
predict extended 
set of conditions 
Phase 2 
Treatment 
option based 
on ingested 
intensive 
care corpus 
and best 
practice 
guidelines Treatment decision 
support and exposure 
of data to research 
and look at patient 
similarities 
Phase 3 
The Mt Elizabeth Novena ICU real time cognitive solution can proactively 
alert and prevent life threatening complications 
© 2016 International Business Machines Corporation 24 
https://www-03.ibm.com/press/us/en/pressrelease/48764.wss 
Developing an image-guided informatic system 
to provide holistic summaries of patient 
conditions and evidence-based clinical decision 
support to radiologists. 
The system 
• integrates clinical and imaging data 
• filters out irrelevant images using multimodal 
analytics 
• highlights disease depicting regions 
(anomalies) 
• flags coincidental diagnosis 
• offers clinical decision support 
MEDICAL SIEVE - RESEARCH