6th Annual Public Health Information Network Conference: How to Visualize and Do Analytics in a Federated Data Model

Sunday, August 24, 2008

South/West Halls

Russell Gann, M, Ed , National Center for Public Health Informatics, Division of Emergency Prepared and Response, Center for Disease Control and Prevention, Atlanta, GA

Barry Rhodes, Ph, D , Division of Emergency Preparedness and Response, Centers for Disease Control and Prevention, Atlanta, GA

Gail Scogin, MS , Division of Emergency Preparedness and Response, Centers for Disease Control and Prevention, Atlanta, GA

How to Visualize and Do Analytics in a Federated Data Model

Within the BioSense project the CDC is actively pursuing a federated data sourcing model in an effort to move away from a costly centralized data warehouse solution and maintain local autonomy and control of data. This presentation provides insight to the structure and components of such a federated system. Federated data systems are relatively new in the IT industry. Standards are just now being implemented and stable federated data warehouse solutions are still being analyzed in the market place.

The federated data solution would be based on certain assumptions attained through a central architectural governance and common federated interface. The systems involved would include data housed at the CDC and the federated sites. The infrastructure would include open source portal technology and tool kits for reporting, plots/charting, connectivity, container, and build environments.

Data analyses include standard definitions for outcome variables including syndromes and notifiable diseases; flexible user-defined case definitions and wild-card searches; time series analysis on disease/syndrome count data that include at least 1 year of previous data to evaluate seasonal trends; map displays of counts, rates, and measures of statistical significance; spatial-temporal analysis; analysis of data at varying levels of aggregation; and stratification of the previously listed analyses by age group and other demographic variables.

Next steps for BioSense would include a remodel of the application with portal-basing, open source products, and SOA-compliant architecture; creation of a pilot model; creation of a design structure; and implementation of pilot.