In order to implement population health management and accountable care, healthcare provider organizations (HCOs) will have to deploy dozens of analytic applications, including clinical, operational and financial applications. With demand for applications outstripping the ability of IT to develop these applications in-house, many of the required applications will be purchased — frequently by end-users with little or no input from IT.
Consequently, most HCOs will end up with a mishmash of analytical applications with different software components. Short-term gains for end-users could cause long-term pain for IT if too many of these standalone applications get installed, all requiring different overlapping infrastructure technology, data, and support. For example, one application may use Microsoft’s data integration, database, or end-user Business Intelligence (BI) software, another could use IBM’s, even though IT has chosen a third vendor as the preferred standard for data integration, end-user tools, and database.
Our report, Building an Infrastructure to Drive Healthcare Analytics, examines this challenge and documents how to create a flexible infrastructure framework to support analytic applications. And how to weave all the technology together into a consistent and manageable platform
The bottom line is building an analytics infrastructure is a design and construction process pulling together and assembling a handful of mature and capable software components — typically from different vendors. The overall architecture and technical requirements of each of the components, as we outline in the report, are well known and tested. It is not rocket science
What makes analytics hard is that the outcomes, particularly organizational impacts, are not known with certainty at the beginning of the project, and the economics are merely uncertain projections until the project moves towards production. The biggest challenge is managing and integrating each analytic project into the platform using the IT best practices we outline.
One other challenge that we discuss in-depth within this report is the need to look at IT analytics projects as different from virtually any other IT project. Analytics projects have an exceptionally high rate of failure. Therefore, rather than using a common project management approach with set deadlines and milestones for analytics projects, we encourage users to use a portfolio management approach that will be flexible and anticipate potential failure, allowing for quick resource alignment once fate of a given analytics initiative is known.