Upload
clifford-russell
View
247
Download
2
Embed Size (px)
Citation preview
INDEPTH Network 1
iSHARE – Let’s Share to Learn More
INDEPTHINDEPTH Data Sharing Initiatives
By
Team iSHARE
INDEPTH Network 2
iSHARE – Let’s Share to Learn More
Presentation Agenda• Data Sharing Initiatives
• Data Sharing with INDEPTH – History, Purpose, Initiatives
• Concept of the Data Repository
• Data Extraction Methodology
• The ETL process
• The Application and the Process
• Dynamic Reports
• The Framework
• Current Limitations and Challenges
• Future plans
• QA
INDEPTH Network 3
iSHARE – Let’s Share to Learn More
Data Sharing Initiatives• INDEPTH Data System (IDS)
– Efforts so far led by Prof. Abraham Kobus Herbst– If funded would lead to Standard Data Management System
(OpenDSS) + A web-based repository– This would greatly enhance cross-site data analysis
• Data Documentation Initiative (DDI)Documenting data within INDEPTH sites using standard machine readable formats
• Data Sharing on the web within INDEPTH Sites (iSHARE)
INDEPTH Network 4
iSHARE – Let’s Share to Learn More
Data Sharing History• Growing call within the funding community & the
scientific community for data to be shared
• Some individual INDEPTH sites; (Agincourt, Africa Centre) had already started taking steps in the direction of sharing data documentation and/or actual data
• In 2007, three INDEPTH HDSS sites in Asia (Vadu -India, Kanchanaburi -Thailand & Wosera - Papua New Guinea) came together to share their data on a web-based repository, with funding from the INDEPTH Secretariat, and technical support from I2IT
INDEPTH Network 5
iSHARE – Let’s Share to Learn More
Why Data Sharing?
• To encourage INDEPTH sites to share their data with the broader scientific community
• To help bring about transparency in scientific inquiry and also allow for verification and refinement of findings, more economically and effectively
• To encourage collaboration with other institutions and communities
INDEPTH Network 6
iSHARE – Let’s Share to Learn More
iSHARE Initiative• iSHARE – INDEPTH Sharing and Accessing Repository
• Funding from the Hewlett Foundation for expansion - to include three African sites
• In response to call from Secretariat; Agincourt and Dikgale from South Africa and Magu from Tanzania joined this initiative, totalling to six HDSS sites on the platform
• All participating sites submitted draft data to be used for development of the repository
• New website (http://www.indepth-ishare.org) beta launched in October 2009 and final to be launched in February 2010
INDEPTH Network 7
iSHARE – Let’s Share to Learn More
Concept of Data Repository
• Standardized and Harmonized dataset
• Collect data from participating HDSS sites (Push / Pull Extraction)
• Clean and transform datasets to standard format
• Upload data to centralized database
• Data Repository created!
• Repeat cycle for addition of more datasets
INDEPTH Network 8
iSHARE – Let’s Share to Learn More
Standardized DatasetStandardized Dataset• Five table
– Base table: one record for each individual under observation
– PregnancyOutcome: one record for each pregnancy experienced by a women under observation
– Deaths: one record for each death that occurs under observation
– In migrations: one record for each in migration into a location under observation
– Out migrations: one record for each out migration from a location under observation
INDEPTH Network
INDEPTH Network 9
iSHARE – Let’s Share to Learn More
Potential Uses of the DatasetPotential Uses of the Dataset• Basic Demographic rate and statistic calculations. Can
character the populations from each site
– Person years calculations
• Assessing vital registry systems with in the sites
– Birth registration
– Death registration
• Other analysis of
– Education
– Occupation
– Reason for migration
INDEPTH Network
INDEPTH Network 10
iSHARE – Let’s Share to Learn More
Dataset StructureDataset Structure
INDEPTH Network
•Individual level•PID uniquely identifies the individual
•Event table link to Individuals
•EID uniquely identifies an event
•Event liked to household(locations) where they occur identified by HID•Social groups simplified to individual living at the same location (HID)•Pregnancies linked to mother. Live born children linked to mother in Individuals (base) table
INDEPTH Network 11
iSHARE – Let’s Share to Learn More
The ETL ProcessStart
Data Extraction
Store the data in dummy tables in Excel/Mysql format
Remove errors in the data
Enforce data standards (Ex: ICD-codes)
Validation and
Integrity test
StopLoad anonymized data into iSHARE database using FTP protocol
Insert data into Error table
Test Passed
Test Failed
More data in
future ?
Yes
No
INDEPTH Network 12
iSHARE – Let’s Share to Learn More
Data Extraction Methodology• Sites send data as per standardized dataset
requirements (Push Method)
– Sites send data in csv, xls, mdb, frm, scripts, etc formats over FTP or eMail
• Sites upload data at specified location; application access that to populate repository (Pull Method)
INDEPTH Network 13
iSHARE – Let’s Share to Learn More
The Application
Wosera HDSS
Kanchanaburi HDSS
Vadu HDSS
Agincourt HDSS
Digkale HDSS
Magu HDSS
iSHAREDB
ETL Operations
ETL Operations
iShare Web
Server
Client
Client
Client
INDEPTH Network 14
iSHARE – Let’s Share to Learn More
The ProcessStart
User Registration
Login & Password Generated
Send Download Request
Accept/Reject Download Request by Committee
Member
Is Download Request
Accepted ?
Accepted
RejectedStop
Download Data
INDEPTH Network 15
iSHARE – Let’s Share to Learn More
The FrameworkUser Interface Layer
Application Layer
Registration Login Download Request
Approval Feedback Reporting
Database Layer
Site 1
Site 2
Site 3
Site n
INDEPTH Network 16
iSHARE – Let’s Share to Learn More
Dynamic Reports• Reports generated on-the-fly providing real-time
data for faster analysis.
• It dynamically loads data from the database
• iSHARE dynamic reports provides:
– Customizable reports as per user needs
– Sophisticated actionable information without exposing internal complex data structures
• Example: Migration Reports – By Year – By Site – Drill Down to gender and generate bar. Line and pie charts for better visual simplicity
INDEPTH Network 17
iSHARE – Let’s Share to Learn More
Current Limitations
• Error findings on datasets is manual process but cleaning is automated
• Pull method of data extraction not yet implemented – a framework for this has to be developed
INDEPTH Network 18
iSHARE – Let’s Share to Learn More
Challenges• Re-coding existing data into agreed categories /
standards come at significant costs and requires funding
• Conflicting conditionality imposed by different parent institutions and funding agencies
• Cost of maintaining the repository as versions and contributing sites increase
• Defining policies for research data in repositories
• Abuse of data downloaded
INDEPTH Network 19
iSHARE – Let’s Share to Learn More
Thank You
http://www.indepth-ishare.org