Transcript Title

The Promise of
Pathology Informatics
James J. Cimino, M.D.
Department of Medical Informatics
Columbia University
Questions You Would
Like Answered
• How can we link patient record information
to pathology specimens to support clinical
research?
• If we can do this, what will be possible?
The Question I Think
You Need to Answer
• How can pathology information be
represented to support data reuse?
What is “Data Reuse”?
In contrast to the primary use of health data
(“show it to the doctor”)…
…any secondary use of the data for tasks
ranging from direct patient care, to financial
purposes, to research
My Goal for this Talk
• Expand upon the notion of data reuse
• Discuss representational issues needed
for reuse
• Propose a challenge
First Admission: August, 1983
In August, 1983, a 50 year old male presented to
the St. Vincent’s Hospital (NY) emergency room
with a scalp laceration due to a falling paint can.
The wound was cleaned and sutured, and the
patient was give a follow up appointment for
surgery clinic. Two weeks later, the patient was
seen at the scheduled clinic visit and was found
to have delayed healing of one portion of the
wound. After several weekly visits, the poorlyhealing area was excised and the wound was
closed. The patient had a good result and was
discharged from further follow up.
Second Admission - March, 1984
The patient was brought to the emergency room
for recent increasing lethargy. Laboratory
evaluation was remarkable only for a calcium of
17 mg/dl. The patient was treated aggressively
with hydration and diuretics, but expired shortly
after admission. A diagnostic report was
received.
Prologue as Epilogue
The pathology report from the wound
revision the previous September included
the following phrase:
“Metastatic adenocarcinoma of
uncertain origin is noted at the
tissue margins”
What Happened?
• The primary use of the data was to
produce a report
• An information system could have
managed that report
• High-quality data representation could
have supported the reuse of the data by
a decision support system
Pathology Information Today
• Narrative reports
• Structured reports
• Coded data (ICD9-CM vs. SNOMED vs.
Local Codes)
Potential for Reuse of
Pathology Information
•
•
•
•
•
•
Direct use by care providers
Billing
Quality assurance/Case management
Clinical research (including Outcomes)
Automated decision support
Integration with other information systems
Example:
Retrospective Clinical Research
Pathology System
Financial System
Data Monitor
Laboratory System
Pharmacy System
Repository
Research
Database
Example:
Prospective Clinical Research
Pathology System
Financial System
Data Monitor
Laboratory System
Pharmacy System
Repository
Research
Database
Decision
Support KB
Example:
Prospective Clinical Research
• Data monitor checks eligibility criteria
• Alert sent to subject recruiter
• Example: Biphosphonate Study
Example:
Automated Decision Support
• Data monitor checks for triggering conditions
• Medical Logic Modules decide if warning
conditions are present
• Message sent to appropriate channel
• Example: Tuberculosis culture result
Caveats: TB Example
• Monitors for delayed culture results
• Sends message if result not equal to the
code “No growth”
• One day, dozens of alerts about positive
results but no organism was reported
• What happened?
How the Lab Fooled the Alert
• Alert looked for results “No Growth”
• Lab started reporting “No Growth to Date”
• “No Growth to Date”  “No Growth”
• Solution: Use the controlled terminology
to map all No-Growth-like lab terms into a
single class, and have the alert logic refer
to the class.
How we Outsmarted the Lab
“No Growth”
Results
No Growth
No Growth
to Date
No Growth
after 48 Hours
No Growth
after 24 Hours
No Growth
after 72 Hours
No Growth
after ...
Example:
Linking to On-line Resources
• Clinician reviewing reports will have
information needs
• On-line information sources can satisfy
that need
• Data from report can be used to
automate the query
Linking Text Reports to
On-line Information Sources
• Natural Language Processing
• Data representation to support reuse
• Codification of information needs
How does CPMC
Support Data Reuse?
• Natural Language Processing
• High-quality controlled terminology ( “codes”)
High-Quality Terminology
• Concept oriented - with concept permanence
• Multiple synonyms
• Multiple hierarchies
• Semantic network of interconcept relations
• Integration of low-level (clinical) terminology
with high-level (aggregation) terminology
Medical Entities Dictionary
Laboratory
Test
Serum
Gentamicin
Test
Antibiotic
Random Gentamicin Level
Main-MeSH:
Supplementary-MeSH:
"Gentamicin/bl"
Measures: Gentamicin
Drug
Chemical
Aminoglycoside
Gentamicin
Injectable Gentamicin
Trade-Name: Garamycin"
Has-Ingredient: Gentamicin
Pathology Informatics
“Grand Challenges”
• Representing the concepts in reports
• Represent the relationships among concepts
in reports
• Representing the concepts for aggregation
and retrieval
• Integration of the concepts
What Will Be Possible
• Data retrieval
• Data mining
• Improvements in report quality
• Improved reuse of data…
…including patient care
My Challenge to You
Ask not what informatics can do for you.
Ask what you can do for informatics.
Questions?