Research Databases The Research Patient Data Registry and Strategic Directions

1
Research DatabasesThe Research Patient Data
Registry and Strategic Directions
Henry Chueh, MD, MS Director, Laboratory of
Computer Science Division of Clinical and
Research Informatics Department of
Medicine Massachusetts General Hospital
2
Agenda

• Problem statement
• Research Patient Data Registry (RPDR)
• Brief RPDR demo
• Challenges overcome, anew
• Strategic directions

3
A Problem?

• Assets for Knowledge Discovery
• World-class investigators
• Large clinical enterprise
• Numerous research data collections
• Pioneering but decentralized research activities
• Active, high quality clinical care
• Expanding clinical information systems

4
The Problem Knowing our population

• Simple or complex attributes
• Lack of systematic, continuous phenotyping for
  research

5
Research Patient Data Registry

• Innovation from the Clinical Research Program and
  Lab of Computer Science
• 1997 Initial prototype by CRP/LCS
• 1998 Partners research sponsorship
• 2000 Initial RPDR pilot
• 2001 General availability

6
What is the RPDR?
- Terabyte warehouse of PHS patient data.- 2
million MGH BWH patients.- 500 million
diagnoses, medications, procedures,
laboratories, with patient demographic
encounters- Authorized use by faculty status,
1000 users.- Researcher can construct complex
queries.
Web-based Query Tool
Research data is digested into facts that can be
queried with patient data
Raw data is digested into 2 facts.
Person ID
Concept ID
Value
Other specific data
Z45920
AL.141
1.0
Z45920
AL.153
1.0
raw data goes into the RPDR as an encrypted object
7
Characteristics of the RPDR

• Unique tool that clinical staff can use directly
  for research needs
• Proven track record of usage with clinical
  investigators, and can also be used for clinical
  operations
• Many person-years of embedded knowledge about the
  clinical domain

8
RPDR impact as of 2004

• 1000 registered users, 321 new in 04
• 100 teams preparing grants, 150 teams preparing
  patient cohorts for IRB approved research, 100
  teams doing clinical studies, 50 teams reviewing
  hospital operations
• Over 1 million retrieved patient charts to date
• 40 million in research grants

9
Obtaining grants with the RPDR
10
Finding Cohorts with the RPDR
11
Other uses for the RPDR
12
Demo
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
Challenges overcome by the RPDR

• Data normalization and metadata
• IRB issues
• HIPAA privacy conditions
• Politics of consolidated data
• Practical issues of access
• Investigator fluency with technology

24
RPDR
25
RPDR Access Policy
1. Access to Query Tool
PHS faculty
2. Perform queries returning aggregate data only
IRB exempt
3. Obtain non-identifying data for cohort
IRB exempt
4. Identify cohort with or without additional
data, from own institution only
PI requires full IRB approval
Requires co-investigator from all institutions,
PI requires full IRB approval
5. Identify cohort /- additional data, from ALL
institutions
Adhere to current IRB rules
6. Contact patients
26
New Challenges

• Expanding set of datatypes (genomics)
• Increasing pressure for outcomes data
• Associational research
• Mandatory reporting

27
Common theme need more detail

• Implies going deep as well as wide
• Means getting the right information
• Patient-based
• Disease-based
• Knowledge-based
• Disease-specific workflow and data collection

28
Optimize RPDR for knowledge discovery

• Integrate deep domain registries with wide
  repositories like the RPDR
• Add complex datatypes (genotypic data, images,
  etc.)
• Integrate biological sample availability
• Create a modular software architecture to
  integrate the RPDR with other biocomputational
  systems and resources

29
Associational studies

• Long term outcomes after head trauma and APOE
  expression

Gene expression in APOE e4 Allele
person
concept
date
raw value
Z5937X
3/4
Outcomes calculated every week
Surgery
microarray (encrypted)
Alzheimer's
ER visit
Z5937X
3/4
Seizures
Trauma
Z5937X
3/4
ER visits
Gene-Chips
Z5937X
3/4
Clinic visits
Trauma
Seizure
Z5937X
4/6
Surgery
Gene-Chips
Z5956X
5/2
Multiple sclerosis
microarray (encrypted)
Seizure
Z5956X
5/2
Alzheimers
Z5956X
5/2
Diabetes
Z5956X
5/2
CT Scan
Z5956X
3/9
Hemorrhage
Z5956X
3/9
Trauma
Z5956X
3/9
Thalamus
Z5956X
3/9
30
Clinical Research Chart (I2B2)

• Collection of software services into a clinical
  research framework

Surveyor
Surveyor
Study mgmt
Study mgmt
statistics
statistics
Publisher
Publisher
application
application
External access
External access
allele
allele
server
server
mapping
mapping
PostOffice
PostOffice
linux
linux
Messaging
Messaging
haplotype
compute
haplotype
compute
Reg
Reg
cluster
cluster
visualization
visualization
Registration
Registration
graphics
graphics
BLOB
BLOB
server
server
storage
storage
indexed
indexed
file server
file server
31
Beyond the RPDR

• Enhanced clinical attributes, domain pertinent
• Define consistent outcomes oriented clinical
  research datasets
• Data of research quality from the clinical
  enterprise

32
Patient-based episode attributes
33
Domain-oriented datasets (NSQIP)
34
Strategic Directions

• Clinical disease registries with extended
  datasets
• Virtuous cycle of clinical care driving research
  driving clinical care
• Sponsorship of an interdisciplinary MGH approach
  for integrated outcomes registries and related
  research data warehouses

35
End