Presentation is loading. Please wait.

Presentation is loading. Please wait.

Data Democratization Using i2b2, Excel, REDCap in End-User Extracts

Similar presentations


Presentation on theme: "Data Democratization Using i2b2, Excel, REDCap in End-User Extracts"— Presentation transcript:

1 Data Democratization Using i2b2, Excel, REDCap in End-User Extracts

2 Data “Democratization”
Facilitate access to clinical data in a end-user drag and drop, guided, fully self-service experience

3 UR Medicine Overview Academic Health Center located in Rochester, New York Four hospitals (including cancer center and children’s hospital) with 1,364 beds UR Medicine clinical enterprise oversees*: 71,371 inpatient discharges 234,742 ED visits (includes Urgent Care) 40,498 ambulatory surgeries 1,691,135 outpatient visits 200+ clinics in a 115-mile radius 1,537 specialty and primary care attending physicians  744 residents 18,500+ clinicians using eRecord *Figures are based on FY2015 reports

4 Users Q1:540 Q2:550 Q3:612 Queries Q1: 1995 Q2: 1025 Q3: 1364
General 2016 i2b2 Statistics Users Q1:540 Q2:550 Q3:612 Queries Q1: 1995 Q2: 1025 Q3: 1364 Distinct Concepts Q1: 1696 Q2: 729 Q3: 808

5 Preparatory to Research
Pathways Simple attestation, immediate access to perform deidentified excel exports Preparatory to Research On-Demand ETLs Identification of datamart via Excel upload or previous query IRB Approved Projects

6 Data Access Validate Apply Data Governance Description of project
Documentation IRB Approval Validate Honest Broker Review Minimal Necessary ETL Manager Data Access ETL on demand Custom i2b2 mart

7 Provisioning Process Production workflows:
IRB Preparatory Preparatory is generally deidentified, limited excel output, no REDCap

8 IRB Project Production workflows: IRB Workflow: IRB Preparatory
Confirmation Upload abilities “Re-Identification”

9 Project Creation Process
Investigators can choose desired data elements to be populated into a data mart, vetted by broker Datamart is late-bound, loaded into local observations in each end-user mart

10 Excel Upload Or I2B2 Patient Set

11 Storage Execution Results Back End Process
Uploaded Excel, Redcap merge data , patient enrollment lists are kept separately from i2b2 Storage We have our “JavaServices” component create, delete, ETL datamarts, compartmentalized into each mart. Execution We have over 500 individual i2b2 data marts, sharing common concepts, metadata. Results

12 Summary of Integrations

13 End User Exporting

14

15 A Message From Our Regulators
Access to data is granted via excel direct download or via tight integration with REDCap

16 Self-Service Aspect 1 record, 1 encounter pull Excel - Fu
Class Crash Course Guidelines on ontology use Framework for future lessons and support Excel 1 record, 1 encounter pull Excel - Fu Questions raised re: pivoting, better output REDCap Complicated setup allows for pivoting on specific events Biostats, DDP  Granular control & Validation

17 Full Screen REDCap

18 Full Screen Redcap Popup Box

19 Redcap I2b2 CDW Enrolled Patients
Pulls enrolled patients, matched before ETL into datamart For Each REDCap Event Look at the mappings (red) and match requested concepts to observations (blue) for time periods where events (green) occur. Look at aggregate function Output data per patient/event/form Concepts Projects has: REDCap Events Observations Matches/ joins REDCap Forms Enrolled Patients Matches/ joins REDCap Fields Visits REDCap Mappings

20 Too Much Data

21 Use Cases REDCap event schedule is set up monthly Questionnaire is built identically in trial data management system End user sends questionnaires to be filled out by patients once a month Excel Export, most recent data element is the pivot option (default) Export a common questionnaire filled out online by patients as a measure

22 Ontologies For MyChart Data
EMR Special Data ETLs Engagement & Usage Messaging Workflows Smart Codified Notes Questionnaires Biweekly ETL Epic Userweb

23 Example SmartData Note

24 Form has a Y/N Additional Use Cases Build as Y/N
Memo box with items matching, all dates “True” constants (Gender, Age) Form has a Y/N

25 Gotchas Reliant on external timepoints
REDCap design tends to be lacking Edit Projects before integration Timepoints, not an afterthought Scheduling the surveys Validations. OFF. Aside from dates, validations are a pain ALPHA for medication, source has “~”

26 More Gotchas Moving Goalposts
Questionnaires design and notes should occur before the study The most recent questionnaire version is maintained on the ontology Same Downsides as i2b2 Not Particularly Fast Investigators lost on how things are documented, - smoking history - scans - notes Careful Naming Investigators assume things “Deidentified NOT MRN”

27 Projects Created as of 3/16

28 REDCap “Chart Reviews”
Usage Statistics 2015 : 765 2016 : 916 Excel Generations 6660 Concepts mapped 2015 : 231, 2016 : 88 REDCap Syncs 2015 : 468 2016 : 320 REDCap “Chart Reviews”

29 How do I get this open-source awesomeness?
i2b2 plugins available here, i2b Epic Clarity ETL via Epic UserWeb Wordpress modifications, java service, plugins Do not support Postgres or SQL Server yet. Working on a faster merge/pivot using excel process.

30 Questions? Philip Ng Research Integration Group Lead HL7 Interface Development Group Lead University of Rochester Information Systems Division 30 Corporate Woods, Suite 350 Rochester, NY Office Phone: 1(585)


Download ppt "Data Democratization Using i2b2, Excel, REDCap in End-User Extracts"

Similar presentations


Ads by Google