Data ware housing

Content:

As I understood from the lecture on health IT systems, data warehousing is the key issue which is acting as a hindrance in implementation of successful EMRs. Two types of data repositories were discussed. One was the clinical data repository or CDR which is patient specific information consolidated in a single database from various sources like clinical notes, laboratory, radiology etc, and the other was enterprise data ware house (EDW) which includes data from financial and business sources in addition to clinical data.

The key issues with CDR include data normalization, use of standards across systems used to input data, storage capacity, interface between the CDR and various sources which are linked to the CDR which input the data into CDR. Storage issue further give rise to the question of using relational database structure or a object repository (I think this is the EAV model of database modeling).

Enterprise data warehouse is used in various other industries including finance, banking. Its use in medicine is particularly challenging because of the lack of standards across various platforms. Therefore standardization of the semantics is one of the key issues. I think an effective EDW can actually solve the issue of electronic health records. That’s easier said than done due to obvious lack of standards and due to role of multiple stakeholders in making an EDW. The merit of having one is undisputable, as this data can be used for research and public health reports. Infact EDW is used extensively in clinical research which has data inputs from many sources which do not necessarily follow same standards. SAS software package has its own EDW system for clinical trials. But as pointed out in the lecture, clinical data is complex and this solution may be complicated than it appears.

Posted by
Sheetal Shetty