All the data produced by electronic health records offers providers an opportunity to identify key insights into
patient health and outcomes. Some have found this opportunity elusive and have yet to find a way to better analyze the data they record. These providers should consider a more advanced storage strategy as a start on the way to better analysis. That's where enterprise data warehousing comes in. It has become a common practice for healthcare organizations looking for a way to store and analyze the growing volume of patient data.
Most patients that visit a healthcare provider begin to generate data in one system or another the minute they schedule an appointment. From there, information continues to accumulate and is deposited in multiple systems. Lab systems, medical imaging or picture archiving and communication systems, nursing systems, real-time location services, electronic health records, nutrition systems, and other medical and wearable devices are some of the places from which patient data is drawn or where it is stored.
All this information presents an incredible opportunity to assist in population health management, clinical research and care optimization. In order for a health system to tackle the volume and diversity of its data and to implement a fruitful enterprise data warehousing initiative and analytics program, it must take the following steps.
The first step of any enterprise data warehousing initiative is managing the data itself and controlling access to it, either its raw format or via vendor provided tools that act as interfaces to the source. Flat files, cache, Oracle, Structured Query Language or other data engines must be cataloged based on what data is extracted or needed.
The next step is extracting the data. This is when the business intelligence (BI) specialists are likely to leverage the capabilities of the selected BI and analytics platform to connect with and evaluate the available data points. During this stage, data will appear in Health Level 7, X12, plain text and other formats that must be translated for the warehousing engine to process the information.
Once the information is extracted and ready for use, it's time to begin the next critical stage of information processing known as data binding. There are two models that are commonly used.
Early data binding: In this approach, the data is summarized before it gets stored into the reporting system, where it is then presented to end users. Summaries such as reimbursement data, payments over a period of time or the productivity of providers are pre-generated. This model pre-processes the information and enables it to be ready on command, based on specific data layouts and summaries that may be requested. It is widely used by small to midsize independent physicians.
Late data binding: This binding model allows information to be explored in its native format without being translated during the time of the analysis. Unlike traditional approaches such as early data binding, which rely heavily on a common denominator, late data binding has proven to be more flexible and eliminates the need for data approximation or simplification.
Depending on the data binding method used, online analytical processing may be used as a last step. This happens most often when early data binding is used because it gives business intelligence users access to the information to analyze it. During this phase, end users may use a number of methods to process and visualize the information.
Healthcare organizations recognize that the patient health record comes in different formats, shapes and sizes. While some systems talk to each other, the data itself can remain locked in multiple silos and only the appropriate enterprise data warehousing platforms will enable realizing new insights and meaningful use of this wealth of information.
About the author:
Reda Chouffani is vice president of development with Biz Technology Solutions Inc., which provides software design, development and deployment services for the healthcare industry. Let us know what you think about the story; email email@example.com or contact @SearchHealthIT on Twitter.