
I was recently asked about the influence of business intelligence (BI) in schools of medicine (SOMs), what the difference is between a data warehouse and a data mart, and which one of the two is better suited for an SOM.
SOMs have two choices to make in determining the overarching scope of the BI effort: a full-scale data warehouse or an academic mission- or department-specific data mart (or series of marts). A data warehouse integrates information from multiple information systems. Data marts, however, are specific to one subject area or system – for example, finance, research, or grants management – and are often smaller in size and scope than a warehouse.
A mart strategy is very consistent with an SOM environment that tends to have data, systems, and budgets specific to missions, departments, or institutes. As data marts are developed and mature, data can be made consistent between marts so that information may be integrated across the SOM. For example, the definition of a “student” or a “grant” would need to be made consistent so that this information can be shared across marts. In a warehouse, these definitions have to be standardized as part of the initial ETL (Extract, Transform, Load) process to ensure consistency. By applying a consistent data architecture, an SOM can begin BI efforts on a smaller scale with marts and move toward a larger data schema in the long term. While data marts may offer some value early, if they are built without an eye to future integration with other data marts, it can mean constantly reorganizing data mart solutions and, ultimately, higher costs.
On the other hand, a well-designed and comprehensive data warehouse provides a solution that is more responsive to the needs of the enterprise as a whole. It can be capitalized in a single effort and have a shorter total implementation timeline. However, keep in mind that a warehouse requires significantly more capital and management up front. As a result, most data warehouses are developed in phases.
Read more about how BI is affecting SOMs and its importance in the success of your SOM going forward.
Published September 13, 2012