Key Driving Forces in Health
Informatics
W. Ed Hammond. Ph.D., FACMI, FAIMBE, FIMIA, FHL7
Director, Duke Center for Health Informatics
Director, Applied Informatics Research, DHTS
Associate Director, Biomedical Informatics Core, DTMI
Professor, Department of Community and Family Medicine
Professor Emeritus, Department of Biomedical Engineering
Adjunct Professor, Fuqua School of Business
Duke University
Chair Emeritus and Secretary, HL7
Information is what our world
runs on. Man has evolved from
the food-gatherer to the
information gatherer. Information
pervades science from top to
bottom, transforming every
branch of knowledge.
Marshall McLuhan, 1967
Our age is the Information Age –
younger than some of us. Claude
Shannon created that age in 1948
when he made information a science;
he gave it a definition; he gave it units
of measurement – the bit. Shannon
estimated the digital information in a
text, a photograph, a record, and even
in the Library of Congress.
What are the Driving Forces?
•
•
•
•
•
•
•
•
Meaningful Use
Big Data
Health Analytics
Learning Health System
Genomics
Translational Medicine
Interoperability
Shared Data
Shared Data
Patient Reported
Outcomes
Aggregated
Patient-centric
Interoperability
Health
Analytics
Big Data
Genomic
Learning Health
System
Biomarkers
Clinical
EHR
Clinical
Trials
GIS
Environmental
Meaningful Use
Translational
Medicine
Images
Meaningful Use
• In 2008, the National Priorities Partnership, convened by the
National Quality Forum (NQF), released a report entitled
“National Priorities and Goals” which identified a set of
national priorities to help focus performance improvement
efforts.
• Among these priorities were patient engagement, reduction of
racial disparities, improved safety, increased efficiency,
coordination of care, and improved population health.
• These priorities have been used to create the framework for
“Meaningful Use” of an electronic health record system
• The ultimate goal of Meaningful Use of an Electronic Health
Record is to enable significant and measurable improvements
in population health through a transformed health care delivery
system
Implementation Incentives
• Entering orders, medications, etc. in CPOE
• Maintaining problem lists in ICD9-CM/ICD10-CM or
SNOMED CT® coding
• Maintain active medication list and electronic
prescribing
• Recording vital signs, smoking status
• Receive and display lab results encoded with LOINC
codes
• Generate patient lists based on specific conditions
and generate patient reminders
Implementation Incentives
• Provide patients with electronic copy and electronic
access to their record and discharge instructions
• Generate a clinical summary for each visit
• Exchange clinical data with other providers
• Protect the information, encrypt it and record
disclosures
Learning Health System
• Digital health data is essential for a continuous learning
health system.
• Necessary to coordinate and monitor patient care,
analyze and improve systems of care, conduct research
to develop new products and approaches, assess the
effectiveness of medical interventions, and advance
population health.
• Goal of IOM project is by year 2020, 90% of clinical
decisions will be supported by accurate, timely,
complete, reliable, trusted, and will reflect the best
evidence available,
Big data
•
EHR growth
–
•
•
•
•
•
from bytes to kilobytes to megabytes to gigabytes to
terabytes to petabytes to exabytes to zettabytes to
yettabytes to …
Less is more
Filters that are purpose and event driven
New methods of presentation
Overload and fatigue
Awareness
1
1950s – 600 Megabytes
In the 1950s, insurance companies had the
biggest data hoards as they collected and stored
data about policy holders. John Hancock Mutual
Life Insurance Company was one of the pioneers
in digitizing customer information, storing data
from two million life-insurance policies on a Univac
computing system acquired in 1955.
Source: Wall Street Journal March 2013
11
1960s – 807 Megabytes
In the 1960s, American Airlines developed Sabre,
a flight reservation system built around one of the
largest IBM computing systems available. Sabre
was one of the first on-line computing systems,
allowing the airline to keep track of an immense
matrix of reservations, flight schedules and seat
inventories.
Source: Wall Street Journal March 2013
12
1970s – 80 Gigabytes
FedEx’s Cosmos system, introduced in the
q970s , allowed the company to scan and
track its huge volume of packages being
shipped around the world.
Source: Wall Street Journal March 2013
13
1980s – 450 Gigabytes
In the 1980s, banks were at the forefront
of data growth, with ATMs coming into
vogue and banks focusing on collecting
and analyzing transaction data for all
their businesses. An example is
CitiCorp’s NAIB.
Source: Wall Street Journal March 2013
14
1990s – 180 Terabytes
Wal-Mart in this decade became the largest
American bricks and mortar retail operation, and, it
is believed, had the biggest commercial data
warehouse in the world. To put this number in
perspective, experts have estimated that
Amazon.com in the late 1990s had single-digit
terabytes of stored data.
Source: Wall Street Journal March 2013
15
2000s – 25 Petabytes
The explosion of the Internet in the ‘90s set the
stage for Web-based companies such as Google
to emerge in the following decade as the global
leaders in big data.
Source: Wall Street Journal March 2013
16
2010s – 100 Petabytes
Is Facebook’s data hoard bigger than
Google’s today? Some say yes, some say
no. According to Facebook, its user content
makes up more than 100 petabytes of stored
photos and video. And analyzing that data
generates about 500 terabytes of new
information every day – more than 2 ½ times
the size of a ‘90s Wal-Mart data cache.
Source: Wall Street Journal March 2013
17
120,000,000,000,000,000
100,000,000,000,000,000
The Growth of Big Data
80,000,000,000,000,000
60,000,000,000,000,000
Global data is
growing at rate of
59% per year
40,000,000,000,000,000
20,000,000,000,000,000
0
1940
1950
18
1960
1970
1980
1990
2000
2010
2020
2004 to 2012:
BI, Analytics, and Big Data
Analytics
Big Data
Google Trends K Search
Business
Intelligence
Challenges of Big Data
– Volume – traditional computers with relational data bases are
not capable of handling data of significant volume – Exabyte’s
100s of terabytes
– Velocity – data flowing into organizations is coming in very fast,
users are demanding live streams of data
– Variety – there are many kinds of data now including photos
video, audio, 3d models, simulations, location data
Topics in Analytics
• Statistical
Programming
• Data Mining
• Advanced Modeling
• Linear Regression
• Logistic Regression
• ANOVA
• Time Series
• Forecasting
• Survival Analysis
•
•
•
•
•
•
•
•
•
Text Mining
Financial Analytics
Risk Analytics
Customer Analytics
Fraud Detection
Geospatial Analytics
Linear Programming
Optimization
Data Visualization
Health Intelligence Team: Future
Possibilities
Predictive
Modeling
• Health system utilization
• Algorithms identifying high-risk patients
• Utilizing clinical and non-clinical data
Patient
Engagement
• Personalized dashboards
• Individualized care plans
• Mobile health applications
Optimization
• Integrating financial and clinical data
• Advanced cost/benefit analyses
• Analyzing and refining operations
Predictive Modeling Use Cases
Use Case
Readmission or death with 29 variables in
EHR
30-day readmission risk using hospitalization
history
MRSA in inpatient setting
Undiagnosed hypertension in primary care
setting
Pediatric “early warning scores”
Classification of Emergent and Non-Emergent
ED Use
Geriatric patient fall risk
Internet-based influenza surveillance system
Others in Literature:
Health System
Center for Clinical Innovation, Parkland
Health & Hospital System
Mount Sinai Medical Center
Center for Clinical Research & Informatics,
NorthShore University Health System
Center for Clinical Research & Informatics,
NorthShore University Health System
Cincinnati Children’s Hospital Medical Center
Beth Israel Deaconess Medical Center,
Harvard Medical School
Institute for Medical Informatics, University of
Braunschweig, Germany
Johns Hopkins, George Washington
University
Length of Stay
Medication Adherence
No Show Rates
From molecules to population
Molecular
Biology
Clinical
Research
Patient
Care
Public
Health
Population
Health
Translational Medicine T1 – T4
Individual, Family, Community, Societies
Site of Care: Intensive care, inpatient, ambulatory, intensive care,
emergency department, long term care, home care
Clinical Specialties
Global
25
Genetics
• Gene mutation will identify many treatable genes
such as Hirschsprung’s disease, muscular
dystrophy, and cystic fibrosis
• Drug treatments are already influenced by
genomic information
– The anticoagulant drug warfarin has a narrow
therapeutic window - too high a dose, patient can
bleed to death; too low a dose, clots remain unclotted.
Genetic information [certain versions of two genes
CYP2C9 and VKORC1] are highly predictive of rate of
metabolizing warfarin.
Clinical Research Informatics
• Registries, controlled data exchange among
researchers, provenance, data capture tools such
as REDCap, query tools such as i2b2, analytics
tools, management of clinical trials
• Patient care – continuing use of patient care data,
shared collection of data, cohort identification,
discovery of target drug requirements, early
translation of research into routine patient care
• Population Health – understanding of prevalence of
disease, identification of new drug targets
27
Patient Care Informatics
• Meaningful use of data, EHR systems, data exchange,
decision support, supporting safe and high quality health
care effectively and efficiently at best cost,
interoperability among all sites of care, embracing
preventive and personalized care, clinical data
warehouses, creation of new knowledge, supporting
patient-centric EHRs
• Adverse events fed back to clinical research and ‘omics
research
28
Public Health Informatics
• Acquisition of data and structured reports, tracking of
disease outbreaks, control of epidemics, insuring
appropriate immunization and vaccinations, analytics,
infectious disease control, disaster management
• Patient care – source of data from health surveillance
and disease management
29
EHR
• New types of data
–
–
–
–
–
–
–
Genomic
Image
Waveform
Patient-generated
Video
Environmental
Behavioral
• Within the next 5 years, over 50% of the data
contained in a patient’s EHR will be non-clinical.
30
Predictions for the next decade
• We will discover a genetic linkage to most diseases
• Early discovery of disease-causing mutations will
influence care with better outcomes
• Cost of DNA sequencing will drop to $500 or less
• Children will have their DNA sequenced at birth
• Most adults will have their DNA sequenced
• Genomic data coupled with personal health data will
have a strong impact on the pharmaceutical industry.
Drugs will be more targeted.
• Health care model will change as will the role of the
provider
31
Challenges
• Acceptance that cooperation and sharing is a win-win
situation
• Single ontology that links all domains; that is an open
process; that uses a common process in which expertise
is the dominant factor; identifies stewardship of each
ontological term
• Ontology removes all ambiguity in associated attributes;
ontology matches terms that are used in the process of
research and care
32
Challenges
• Solves privacy issues; recognizes that personal
control of data may harm creation of new
knowledge and seamlessly connecting the
contributing domains for the most effective care
• Identification and implementation of standards for
data and data exchange
• Controlled and purposeful exchange of data
• Quality of data is insured through process,
algorithms, and certainty factor
33
So, what can we do?
• Facilitate appropriate use of technology for
health care (Meaningful Use)
• Improve the ability to make appropriate
decisions about diagnoses and treatments (MU)
• Improve the knowledge base for health care
• Eliminate inefficiencies in the process of
administration and in health care delivery
• Create new models for delivery, administration,
and reimbursement of care
3
More of what can we do
• Evaluate the effectiveness of testing and treatment
(Comparative Effectiveness Research)
• Improve techniques for the management and use of data
(MU)
• Educate people in the need for and use of informatics
(HITECH, incentive funding)
• Ethics, privacy, security (HIPAA)
3
What is GIS?
 A geographic information system (GIS)
integrates hardware, software, and data for
capturing, managing, analyzing, and
displaying all forms of geographically
referenced information.
 GIS allows us to view, understand,
question, interpret, and visualize data in
many ways that reveal relationships,
patterns, and trends in the form of maps,
reports, and charts.
 Ultimately helps answer questions and
solve problems by allowing us to look at our
data in a way that is quickly understood and
easily shared.
Source: Sohayla Pruitt
GIS Core Capabilities:
Visualization
Source: Sohayla Pruitt
What I did not talk about
•
•
•
•
•
•
•
•
Mobile devices
Data Elements
Phenotypes
Personalized Medicine
Consumer Engagement
Social Networking
3D Printing
Virtual Reality
What has technology added?
• Moore’s Law – faster, bigger, cheaper
computers and communications than ever
• Wireless, global communication that instantly
connects the world
• Small, mobile devices that are ubiquitous
• More knowledge than we can use
• The ability to aggregate all data about a person
from all sources
• A rich mixture of media to enhance our
understanding of disease and its cure
Conclusion / Summary
• The pace of technology has been paced by Moore’s law:
roughly, computational power doubles approximately
every two years
• Use of technology – informatics – has not kept pace. The
future of health care depends on our getting ahead of the
curve
• That step demands a step change – revolution, not
evolution!
40