MIE2012
Towards Collaborative Chronic
Care Using a Clinical GuidelineBased Decision Support System
Haifeng Liu, Jing Mei, Guotong Xie
IBM Research – China
2012-08-28
© Copyright IBM Corporation 2010
MIE2012
Outline
§ Objective
§ System architecture and components
§ Integration of decision support with CDM process
§ Result and future work
2
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
Motivation and Objective
§ Guideline-based decision support system has largely
focused on decision support tasks in acute care
§ Crucial factors of improving chronic care include
integrating decision support into a complex care workflow
-  implementing effective patient self-management
-  developing an efficient communication mechanism among care
providers and patients
- 
§ Our work aims to implement innovative chronic care by
developing a collaborative system framework to streamline chronic
care processes across multiple health providers by executing
clinical guidelines;
-  executing clinical guidelines by integrating a business process
management (BPM) engine and a novel decision engine which
evaluates clinical criteria conforming to the HL7 GELLO standard
- 
3
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
System Architecture
GELLO
evaluator
CIG
CIG
adapter
Guideline
processes
CDM engine
General
Practitioner
Commands
Clinical conditions
CDM
portal
Requests
BPM
engine
CDM manager
CDM
services
Queries
Clinical
data
Alerts/Recommendations
GELLO
query
adaptor
vMR data
mediator
Queries
vMR data
CDR
Specialist
Mobile
communication
platform
Institution
manager
Patient
CDM
DB
Enterprise Service Bus
CIS
(Central hospital)
4
CIS (community
center 1)
Haifeng Liu, Jing Mei, Guotong Xie
CIS (community
center x)
MIE2012
Exemplar skeleton of a computerized CDM
process
start
Get
updated
data
Generate
reminder
messages
Generate
alert
messages
Generate
therapy
advices
5
Post reminders
Post alerts
Post
advices
Haifeng Liu, Jing Mei, Guotong Xie
Check
guideline
compliance
end
Post
incompliance
MIE2012
GELLO-based decision support
GELLO query
adaptor
GELLO
evaluator
D1
6
M1
E1, D1
Q1
start
CDM
services
Y
Get data
Hba1c
>10%
?
Haifeng Liu, Jing Mei, Guotong Xie
Insulin therapy
end
N
Oralmedication
MIE2012
Data
Data
standards
Q1 package VMR
GELLO
D1 <LaboratoryObservation classCode="OBS" moodCode="EVN">
vMR
E1 package VMR
GELLO
M1 <advice><advice_type>therapy</advice_type><content>“Applying insulin
Service
message
context Patient
def: HbA1c : CD = factory.CD(‘LOINC’, ‘4548-4’, ‘Hemoglobin A1c, B’)
self.isAssociatedWith ->
select(oclIsTypeOf(LaboratoryObservation)).oclAsType(LaboratoryObservatio
n)) -> select(testCode.equal(HbA1c)) -> sorted(effectiveTime) -> first
<testCode code="4548-4" codeSystem="2.16.840.1.113883.6.1"
codeSystemName="LOINC"
displayName=" Hemoglobin A1c, B"/>
<status code="completed"/>
<effectiveTime value="20104071530"/>
<value xsi:type="PQ" value="11" unit="%"/>
</LaboratoryObservation>
context Patient
def: HbA1c : CD = factory.CD(‘LOINC’, ‘4548-4’, ‘Hemoglobin A1c, B’)
def: HbA1c_threshold : CD = factory.PQ(‘10’, ‘%’)
self.isAssociatedWith -> exists(testCode.equal(HbA1c) and
value.greaterThan(HbA1c_threshold))
7
injections”</content> </advice>
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
Result
§ Computerizing the guideline into an executable care
process composed of
- 
32 sub-processes where 271 clinical steps and 41 clinical decision
points are defined ;
§ Providing diabetic therapy recommendations including
- 
16 drug therapy and 2 referral advices to clinicians through CDM
portal according to the underlying guideline;
§ Sending 7 kinds of follow-up reminders including
- 
HbA1C and blood lipid measurements, eyeground examination,
and so on, 5 kinds of health alerts including hyperglycemia, high
blood pressure, and so on, and care education messages to MCP
which forwards the messages to diabetic patients;
§ Identifying therapy activities that are not complied with the
underlying clinical guideline including mismatched
medications and referrals not conforming to the standards
8
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
A screenshot for viewing guideline-based
clinical recommendations
Guideline-based
recommendation
9
Patient condition
description
Haifeng Liu, Jing Mei, Guotong Xie
Guideline content
reference
Recommendation
type
MIE2012
A screenshot for showing guideline-incompliance status
Statistics about
guidelineincompliance
degrees
Statistics about
guidelineincompliance
reasons
10
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
Future Work
§ We currently deployed the GC3 prototype in a passive
mode where clinical decisions are derived in response to
data captured by vMR data mediator from CIS; We will
enable system work in a proactive model where CIS can
directly interact with the CDM services to feed clinical data
to the CDM engine and receive system recommendations
in real time
§ We will extend the computerization of guideline to provide
decision support over the full chronic disease stages
including prevention, diagnosis, treatment, and prognosis
11
Haifeng Liu, Jing Mei, Guotong Xie
MIE2012
Thanks!
12
Haifeng Liu, Jing Mei, Guotong Xie