Transcript Active Semantic Electronic Medical Records

Active Semantic
Electronic Medical Records
an Application of Active Semantic Documents in Health Care
Amit Sheth , S. Agrawal, J. Lathem, N. Oldham,
H. Wingate, P. Yadav, K.Gallagher
Athens Heart Center & LSDIS Lab, University of Georgia
http://lsdis.cs.uga.edu
Semantic Web application in use
In daily use at Athens Heart Center
– 28 person staff
• Interventional Cardiologists
• Electrophysiology Cardiologists
–
–
–
–
2
Deployed since January 2006
40-60 patients seen daily
3000+ active patients
Serves a population of 250,000 people
Information Overload
• New drugs added to market
– Adds interactions with current drugs
– Changes possible procedures to treat an illness
• Insurance Coverage's Change
– Insurance may pay for drug X but not drug Y
even though drug X and Y are equivalent
– Patient may need a certain diagnosis before some
expensive test are run
• Physicians need a system to keep track of
ever changing landscape
3
System though out the practice
4
System though out the practice
5
System though out the practice
6
System though out the practice
7
Active Semantic Document (ASD)
A document (typically in XML) with the following
features:
• Semantic annotations
– Linking entities found in a document to ontology
– Linking terms to a specialized lexicon
• Actionable information
– Rules over semantic annotations
– Violated rules can modify the appearance of the
document (Show an alert)
8
Active Semantic Patient Record
• An application of ASD
• Three Ontologies
– Practice
Information about practice such as patient/physician data
– Drug
Information about drugs, interaction, formularies, etc.
– ICD/CPT
Describes the relationships between CPT and ICD codes
• Medical Records in XML created from
database
9
Practice Ontology Hierarchy
(showing is-a relationships)
facility
insurance_
carrier
owl:thing
ancillary
insurance
ambularory
_episode
insurance_
plan
encounter
person
event
patient
10
practitioner
insurance_
policy
Drug Ontology Hierarchy
(showing is-a relationships)
non_drug_
reactant
interaction_
property
formulary_
property
formulary
indication
monograph
_ix_class
prescription
_drug_
property
cpnum_
group
property
indication_
property
brandname_
individual
brandname_
undeclared
prescription
_drug_
brand_name
brandname_
composite
generic_
composite
prescription
_drug
prescription
_drug_
generic
generic_
individual
11
owl:thing
interaction
interaction_
with_prescri
ption_drug
interaction_
with_non_
drug_reactant
interaction_
with_mono
graph_ix_cl
ass
Drug Ontology showing neighborhood of
PrescriptionDrug concept
12
Part of Procedure/Diagnosis/ICD9/CPT Ontology
specificity
diagnosis
13
maps_to_diagnosis
procedure
maps_to_procedure
Local Medical Review Policy (LMRP)
support
ICD9CM
Diagnosis Name
Example – a partial list
of ICD9CM codes that
support medical
necessity for an EKG
(CPT 93000)
Data extracted from the
Centers for Medicare
and Medicaid Services
15
244.9
Hypothyrodism
250.00
Diabetes mellitus Type II
250.01
Diabetes Mellitus Type I
272.2
Mixed Hyperlipidemia
414.01
CAD – Native
780.2780.4
Syncope and Collapse
Dizziness and Giddiness
780.79
Other Malaise and
Fatigue
785.0785.3
Tachycardia Unspecified Other Abnormal Heart
Sounds
786.50786.51
Unspecified Chest Pain –
Precordial
786.59
Other Chest Pain
Technology - now
• Semantic Web: OWL, RDF/RDQL, Jena
– OWL (constraints useful for data consistency), RDF
– Rules are expressed as RDQL
– REST Based Web Services: from server side
• Web 2.0: client makes AJAX calls to ontology,
also auto complete
Problem:
• Jena main memory- large memory footprint,
future scalability challenge
• Using Jena’s persistent model (MySQL) noticeably
slower
16
Design and Implementation Issues
•
•
•
•
•
Schema design
Population (knowledge sources)
Freshness
Scalability though client side processing
Rules: “Starting at instance A is it possible to get
to instance B going through these certain
relationships, if so what are the properties of the
relationship” (e.g., “Does nitrates or a super class
of nitrates interact with Viagra or one of its super
classes, if so what is the interaction level” )
17
Architecture & Technology
18
Demo
On-line demo of Active Semantic Electronic Medical Record
deployed and in use at Athens Heart Center
19
Evaluation and ROI
• Given that this work was done in a live,
operational environment, it is nearly impossible
to evaluate this system in a “clean room” fashion,
with completely controlled environment – no
doctors’ office has resources or inclination to
subject to such an intrusive, controlled and
multistage trial. Evaluation of an operational
system also presents many complexities, such as
perturbations due to change in medical personnel
and associated training.
20
Athens Heart Center Practice Growth
1400
1300
Appointments
1200
1100
1000
2003
900
2004
800
2005
700
2006
600
500
Month
21
se
p
oc
t
no
v
de
c
ju
l
au
g
ju
n
fe
b
m
ar
ap
r
m
ay
ja
n
400
Chart Completion before the
preliminary deployment of the ASMER
600
400
Same Day
300
Back Log
200
100
04
M
ar
04
M
ay
04
Ju
l0
Se 4
pt
04
N
ov
04
Ja
n
05
M
ar
05
M
ay
05
Ju
l0
5
0
Ja
n
Charts
500
Month/Year
22
Charts
Chart Completion after the preliminary
deployment of the ASMER
700
600
500
400
300
200
100
0
Same Day
Back Log
Sept
05
Nov 05
Jan 06
Month/Year
23
Mar 06
Benefits of current system
• Error prevention (drug interactions,
allergy)
– Patient care
– insurance
• Decision Support (formulary, billing)
– Patient satisfaction
– Reimbursement
• Efficiency/time
– Real-time chart completion
– “semantic” and automated linking with billing
24
Benefits of current system
• Biggest benefit is that decisions are now in
the hands of physicians not insurance
companies or coders.
25
Technology - Future
• BRAHMS (with SPARQL support and path
computation*) for high performance main
memory based computation
• SWRL for better rule representation
• Support for example user specified rules, possibly
for integration with clinical pathways:
– If patients blood pressure is > than 150/70 prescribe
this medicine automatically.
– If patients weight is > 350 disallow a nuclear scan in the
office because our scanning bed cannot handle such
weight.
– If patient has diagnoses X alert, the user to suggest a
doctor to refer patient to Y.
26
* Semantic Discovery http://lsdis.cs.uga.edu/projects/semdis/
Value propositions & Next steps
• Increasing the value of content, and
content in context – highly customized
using one of the ontologies (not just
CTP/ICD9, but also specialty specific), at
the point of use; no separate search, no
wading through delivered content
• Actionable rules
• Possible trial involving alert services: “When
a physician scrolls down on the list of drugs and clicks on the drug
that he wants to prescribe, any study / clinical trial / news item
about the drug and other related drugs in the same category will
be displayed. “
27
Comments on Evaluation
Questions?
More? See Active Semantic Document Project
(http://lsdis.cs.uga.edu/projects/asdoc/)
at the LSDIS lab
Or resources (example ontologies, Web
services, tools, applications):
Google: LSDIS resources, or
http://lsdis.cs.uga.edu/library/resources/
28