Pathology Informatics and the Future of Medicine

1
Pathology Informaticsand the Future of Medicine

• Practical Informatics Course
• APIII 2007

2
What is Pathology Informatics?

• The term has been defined largely by its usage,
  and its meaning is evolving (expanding)
• For many, informatics the use of computers (or
  any and all technology) in the practice of a
  discipline
• Shifts focus to the tools themselves rather than
  how the tools are used
• e.g., laboratory medicine is not simply the use
  of instruments to analyze blood
• e.g., molecular biology is not the use of
  bacteria in biology
• Dilutes the intent of defining an academic
  discipline

3
Informatics vs Computer Science ?

• Computer Science, among many other things,
  addresses how data is stored and organized in a
  computer.
• For the most part, however, computer scientists
  focus on the storage of the data itself, often
  without consideration for what the data
  represents in terms of real-life concepts
• Knowledge of the real-life concepts represented
  by the data, including the inherent and implicit
  relationships these concepts impose, is critical
  to proper management of that information. This
  domain layer is an important part of what
  separates informatics from computer science

4
141
1
8
5
73
21
13
56
92
5
Y
C
H
I
P
V
F
S
K
W
N
6
Informatics

• The study of how data is acquired, stored,
  processed, retrieved, analyzed, and presented, in
  such a way as to turn data into information
• Computer is not part of the definition
  discipline could have existed a century ago
• Information explosion, plus demands to make
  greater use of that information
• Manual management becomes inefficient and
  impractical
• Computer becomes an indispensable tool for the
  discipline of informatics

7
Informatics is a Process
8
Overlapping Scope of Informatics
9
Pathology Informatics

• The use of electronic information systems to
  facilitate, promote, and advance the practice of
  pathology
• Practice of pathology patient care, physician
  education, understanding disease
• Technology in and of itself does not count - it
  must convey information
• However, understanding the tools (computers,
  cameras, microscopes) is certainly an important
  part of the successful use of informatics

10
The Role of the Pathologist in Patient Care

• Pathologists do not simply attach names to pieces
  of tissue or report numbers measured by a
  machine we provide an assessment which directs
  the clinical care of the patient
• The final interpretation is based not solely
  upon the tissue and/or the results of analysis,
  but rather is tempered by the clinical setting
• We choose the wording of our interpretive reports
  to trigger the appropriate clinical response
• Tools
• The histologic image is one of the richest data
  streams in medicine, and the ability of the
  pathologist to synthesize and interpret this
  information is unlikely to be duplicated
  electronically for some time
• Other techniques immunohistochemistry, flow
  cytometry, gene rearrangement studies, molecular
  tests

11
PathologyToday
Electronic Data Layer

• Information systems used predominantly to manage
  practices and workflow
• Limited use to access external data sources
• No direct feedback from patient outcome

Dx
Rx
12
Forces Reshaping the Practice of Medicine

• Information Availability
• Massive availability and exchange of data
  on-line
• Increased expectation for all physicians to be
  up-to-date
• Outcomes-based treatment decisions
• Consumerism
• Information access (hidden medical knowledge
  revealed)
• Personal control of care
• Expectations for Consistency, Quality, and Value

• New, Complex Data Streams
• Human Genome
• Genetics
• Molecular Markers
• Diagnostic
• Prognostic
• Gene Expression Arrays
• Proteomics
• Pharmacogenomics

13
Controlling the Flow of Information

• Vast amounts of data need to be assimilated,
  filtered, and presented in such a way as to
  convert it into usable information
• Informatics tools will be needed to assist in the
  data integration process
• The decision making role in medicine will go to
  whoever can most effectively mobilize,
  capitalize, and leverage the informatics
  resources necessary to make this a reality
• Two views of the future

14
Pathologist as Data Provider
Electronic Data Layer

• Pathologist limits evaluation to tissue or fluids
  submitted
• Pathology output is a data element in treating
  clinicians decision support system
• As treating clinicians draw on other data
  sources, relative contribution of pathologist
  decreases

Electronic Data Layer
Decision Support
Dx Rx
15
Pathologist as Diagnostic Specialist
Electronic Data Layer

• Pathology moves aggressively to develop and adopt
  informatics tools to leverage data from emerging
  technologies
• Pathologist remains a diagnostic consultant,
  providing outcomes based treatment
  recommendations to treating clinicians

Decision Support
Dx
Rx
16
Choosing our Future

• Pathologist as a Data Provider
• Easier steady-as-she-goes result
• Pathology is protected from the new diagnostic
  modalities
• As new diagnostic modalities proliferate, the
  relative contribution of the pathology decreases
• Pathologist as the Diagnostic Specialist
• Pathologists, largely freed from the details of
  day-to-day patient care issues, are better
  positioned to fulfill the role of data
  integration
• Hospital based pathology departments have
  traditionally enjoyed better access to hospital
  capital budget funds than direct patient care
  disciplines
• More in keeping with the current roles of the
  respective players
• More in line with the interests and personalities
  for which the two disciplines have traditionally
  selected
• Pathologists and pathology departments must move
  aggressively to develop the skills and the
  informatics infrastructure necessary to leverage
  the information explosion

17
Building our Future

• Create informatics programs in pathology and/or
  laboratory medicine departments
• Get involved in Electronic Medical Record
  decisions at your institution
• New information systems structured for outcomes
  research
• Pathology Informatics and biomedical research
• Basic informatics research

18
What is the Future of Pathology Informatics?

• Probably NOT in NEW technology
• We already have more technology than we know what
  to do with
• Some exceptions (speech recognition, virtual
  microscope slides, etc.)
• These are simply different ways of doing what we
  already do
• Management of existing technology
• Requires an understanding of capabilities and
  limitations of the technology
• Requires a thorough understanding of the
  day-to-day practice of pathology and the barriers
  to adopting new practices
• Requires a vision of what the role of the
  pathologist will be in 10-20 years
• Should be a need-driven process
• Should arise within, or at least be integrally
  associated with, pathology / laboratory medicine
  departments
• Hospital / Institutional IT groups have different
  goals / priorities which may overlap, but often
  will not (different geographic coverage areas)

19
Role of the Pathologist in Managing and Using
Clinical Information Systems

• Whatever information system management structure
  exists at your future institutions, pathologists
  will have some role in decision making
• Selection of information system
• Selection of features to implement
• Integration of system into departmental workflow
• Management of information system
• Development/deployment of add-on technology

20
Developing a Pathology Informatics Program

• Goal
• Scope
• Main clinical system(s), desktop support,
  infrastructure, R D
• Philosophy
• New Technology vs Application of existing
  technology
• In-house development vs off-the-shelf
• Identify Resources
• People, capital investments, on-going support
  costs
• Relationship to Institutional IT Programs
• Role of Pathologist(s)

21
Pathology Informatics Philosophies
Development of New Technology (Basic Informatics)
Application of Existing Technology (Translational
Informatics)

• Goal Proof-of-Concept
• Fits traditional research model
• Externally Fundable
• Publishable
• Promotable
• Cutting-Edge work is expensive (5-20x)
• Brings money into institution
• Pathologist is a consultant

• Goal Usable solution
• Clashes with traditional model
• Generally internally funded
• May be institution specific - soft publications
• Not viewed as scholarly
• Costs are more predictable
• In short term, costs institution money
• Pathologist should set direction

22
Scope of Pathology Informatics Programs
23
Pathology and Computerized Patient Records

• Electronic Medical Records are coming
• Natural extension of spread of information
  management through healthcare
• Why should pathology be integral to the selection
  and deployment of these systems?
• Pathology is the largest volume data generator
  for input into EMRs
• Longest history with longitudinal electronic
  patient records
• Pathologists must maintain stewardship for their
  data in EMRs
• Accuracy
• Display formats - the appearance of the data in
  the EMR is replacing the printed report as the
  primary product of the pathology laboratory

24
New Systems for Outcomes Research

• Importance of outcome information
• Pathology is the study of disease, not just the
  histology/chemistry of disease
• Patient outcome is a crucial part of
  understanding the disease
• Current anatomic and clinical pathology
  information systems are not ideal for outcomes
  research
• Current systems are specimen based rather than
  patient based
• Interpretive data stored as free text discrete
  data elements often not readily available
• Tumor size, LN status
• Hospital Tumor Registries
• Starting point - model for the type of
  information systems needed
• Manual curation of data
• Pathology departments should be more actively
  associated
• More accurate abstracting of data
• Multi-institutional systems

25
Enabling Biomedical Research

• The data stored within pathology information
  systems is incredibly valuable for clinical
  research
• When combined with archived human tissue, becomes
  the single most valuable commodity for basic
  biomedical research
• Issues
• Tissue ownership
• Potential future diagnostic value of tissue -
  what is excess
• Reclaiming results from research studies to use
  in decision support
• Embrace new diagnostic tests as they become
  available
• Pathology has the experience necessary to
  critically evaluate the true diagnostic/prognostic
  value of these tests

26
Basic Informatics Research

• Information extraction from text
• Natural language processing
• Mobilize data stored in pathology archives, as
  well as discharge summaries and operative notes
• Issues negation, synonyms, term evolution
• Information extraction from images
• Humans are incredibly good at pattern recognition
• Human eye is bad at absolute density assessment -
  quantitation
• Electronic cytotechnologists
• Computer assisted diagnosis
• Potential role for whole slide imaging

27
Absolute Intensity Assessment
28
Data Display / Presentation Tools

• New diagnostic tests are highly parallel,
  producing as output hundreds to thousands of data
  points
• People, in general, are not very skilled at
  assimilating long tables of numbers and drawing
  inferences and/or conclusions
• People in general, and pathologists in
  particular, are very good at gleaning information
  from graphical or other similar displays
• Tools need to be developed to display the results
  of these new tests in a more visually efficient
  manner

29
Additional areas of Basic Informatics Research

• Knowledge Management Tools
• Seamless access into on-line databases
• Medical equivalents of google
• Decision support systems
• Not necessarily expert systems
• Computer assisted diagnosis
• Role in increasing utility of whole-slide imaging
• Modeling of biological systems
• Better prediction of tumor biology based on
  particular collection of mutations
• Customized therapies and personalized medicine
• Pharmacogenomics

30
Pathology Informatics - A Subspecialty?

• Arguments for
• Skill set and knowledge base beyond basic
  pathology
• Departments/practices benefit from individuals
  with specialty training in informatics
• Improves recognition and validity of discipline
• Makes a more credible argument for ownership or
  at least a role in institutional information
  system decisions
• Arguments against
• With little agreement about the scope, what would
  be the curriculum?
• Hematopathology model vs Immunohistochemistry
  model

31
The Future of Pathology
Data Provider or Diagnostic Specialist ??

• To assure our future, we must
• Move quickly to adopt new diagnostic and
  prognostic modalities
• Be willing to rethink and adjust how and why we
  store data electronically
• Work with software vendors to steer software
  development toward the development of systems
  which will meet the long term goal of outcomes
  analysis
• Invest the resources needed to develop, pilot,
  and implement such systems
• Accept a quality assurance role for data
  integrity to secure the future utility of the
  data in our possession
• Recognize and support the role Informatics will
  have in shaping that future
• Develop intradepartmental informatics
  initiatives, guided and directed by pathologists
• Recognize that informatics is not simply a
  technical activity best relegated to others
• Leverage our existing and growing informatics
  infrastructure to explore new options for data
  storage, analysis, interpretation, and
  presentation

32
The Future of Pathology

• The best way to predict the future is to make
  the future

The future of pathology as the diagnostic
specialty is both desirable and attainable, but
it is not being reserved for us. Pathology
departments must take an active role in shaping
their own future, or someone elses vision of the
future will be forced upon us.
33
Practical Informatics Course

• Pathology Informatics and the Future of Medicine
  (Sinard)
• Desktop Computer Hardware/Software (Parwani)
• Digital Imaging Basics (Sinard)
• Break
• Telepathology and Whole Slide Imaging (Parwani)
• Electronic Data Exchange (Sinard)
• Data Standards and Coding (Crowley)