How to Evaluate a Business Intelligence Tool for Healthcare

One BI Tool to Rule them All

Zemanta Related Posts ThumbnailI used to think I would eventually find the one Business Intelligence (BI) tool for healthcare that would meet all of my needs for data discovery, analysis, visualization, presentation, and reporting.  Now, however, I doubt I will ever find such a “one size fits all” solution. A big obstacle to identifying one single best analytics tool is that analytical needs vary so widely within healthcare—the best tool really depends on the audience that will consume the data, how they will use it, and what the goal is.

Having just one tool to use is not as important as having the tool that accomplishes what you need it to do.  For this reason, I advocate that you consider licensing more than one tool in the toolset.

How do you decide which tools are best for your organization?

BI or Visualization Tool Use Cases

I’ll outline a few example scenarios of different BI/Visualization tool uses that might help in our consideration of which tool is best. You might be involved in one or more of the following scenarios using a BI tool:

  • Scenario 1: You want to use a tool at your desk, on your own time, measuring an outcome and discovering what data is pertinent to the issue. You want to analyze what impact some independent variables will have on that outcome. The results might help answer your business partners’ questions during a small-group discussion later.
  • Scenario 2: You need to teach a business partner or client one-on-one, enabling them to “fish for answers for themselves.”
  • Scenario 3: You want to show findings in your data to a small group (e.g. 3 to 5 clinicians) to help facilitate a weekly project discussion. Your answers (and the tool’s) will beget more questions during the meeting, ideally answered on the spot, during the same meeting. That’s agile!
  • Scenario 4: You’ve been asked to present your findings to a large group to inform and inspire them with the results of your project or analysis.
  • Scenario 5: You need to present your findings to a group of C-level executives, or better yet, you need to create a dashboard for your executive business partner to present findings to his or her executive peers.

EDW Structure and Correct Use is Critical

I would be remiss if I did not highlight that the data and business rules layers are best placed in a separate layer from the visualization. Separating the analysis tool or visualization layer from the data source and from business rules for a subject area encourages the reuse of those data marts and views.  It also encourages agreement of report data across the organization, elimination of redundant reports (or redundant projects!), and much better use of analysts’ time. My colleague, Russ Staheli, mentioned the time it can take hunting for data or transforming it. Analysts love analyzing data—not transforming data!

Some BI tools sell their data transformation capabilities as great strengths. For example, QlikView handles advanced Common Table Expressions (CTEs) and analytic functions in SQL very well. Or JMP has a fairly comprehensive suite for data transformation (e.g. splitting and transposing). However, I caution against performing Extract, Transform, and Load (ETL) functions in the visualization layer since it does not fit into a scalable model of data and object reuse across your organization.

Financial Considerations

Disregard the server license cost comparisons and the multi-user discounts for a moment (though I recognize the deployment cost should be a big factor in the decision). Consider the purchase of just a few single-user licenses for a really good analytics tool, put into the hands of the most capable, advanced analysts—who will use the advanced features of that tool. The successful use of the tool to aid in studying and improving healthcare processes could easily offset the marginally higher cost of the single-user purchase.

For example, consider the following scenario: a hospital system assigns a team to reduce the number of hospital 30-day readmissions for those suffering from chronic heart failure. The hospital system selects and deploys an analytics tool that has some strengths but is deemed less effective for the purposes of a clinical project team assigned to the project. However, it was selected because it costs less. As a result, the clinical team is less able or willing to use the tool in their regularly scheduled meetings to get at exactly the views that will help their analysis.

Now, consider the same clinical team instead applying analysis and process improvement through their preferred visualization tool; the team feels better equipped to view the analysis they need to ensure they succeed in reducing 30-day readmissions. So, the value of the better tool quickly surpasses the higher cost.

I squirm uncomfortably when the decision for a BI tool is sometimes made solely on the basis of cost, or because the Supply Chain department requires dealing with a consolidated vendor list. If you need both an apple and an orange, why should you purchase two apples purely because you were given a 25% discount on them?

What criteria do you use when making a decision to purchase a BI or visualization tool?

Loading next article...