Too often organizations rush into big data projects without keeping an eye on the big picture, Booz Allen Hamilton's Steven Escaravage and Joachim Roski write in a Health Affairs Blog post.
They propose some best practices that, along with disciplined project management, can eliminate pitfalls they've encountered in such projects with government agencies including the National Institutes of Health (NIH), Centers for Disease Control (CDC) and the Department of Veterans Affairs (VA).
Among their advice:
- Acquire the right data for the project: Too often organizations go with the data they have on hand when more useful data might be difficult to obtain. Look for high-impact data sources beyond that you've used to look at the problem before.
- Ensure that initial pilots have wide applicability: They tell of a government agency's pilots focused on specific, computationally complex and storage-intensive challenges. While the projects overcame the technical challenges, the results didn't provide fodder for transformational change in the organization as the bosses had hoped. Instead they recommend projects with wider applicability that can be more easily understood.
- Don't start with a solution: Give subject-matter experts direct access to the data to explore and find unexpected patterns. Using business-intelligence reports and other conventional approaches will give you more of what you've had in the past.
Cincinnati Children's Hospital's Yiscah Bracha has argued that analytics pros should be part of every healthcare organization's data warehousing project because they know the data they need and how to best use it.
Researchers recently laid out the plan for the NIH Big Data to Knowledge (BD2K) initiative, pointing to a number of challenges including developing the required technology, developing an effective search mechanism and incorporating policies and practices that protect patients' privacy.
To learn more:
- read the post