Our local health insurance provider client needed to extract data from multiple sources, process it using a diagnostic grouper, then capture the output in a new data mart to be used for predictive analytics.
The broad goals were:
• create a data mart that was user-friendly for predictive analysts
• keep manual processing effort to a minimum
• ensure high data quality
• include 10 years of history for go-live
The main challenges were:
• upstream data quality issues
• multiple heterogeneous data sources that needed to be processed uniformly
Aegeus designed and developed a comprehensive solution centered around the Johns Hopkins ACG System using Microsoft SQL Server, SSIS and C#.NET. The solution provided an automated way to do the following:
1) Extract Transform and Load data from multiple heterogenous sources into the data mart, aggregating, de-duplicating, cleaning and conforming data to improve data quality while providing a uniform representation.
2) Invoke the Johns Hopkins ACG and process the newly formatted data through it using custom configuration options.
3) Load the diagnostic grouper’s output into the datamart, providing easy-to-use monthly snapshots and cumulative roll-ups.
The solution Aegeus delivered provided a stable, efficient platform for predictive analytics, as well as an extensible foundation for future development efforts.
Leave a Reply