It goes without saying that the topic of big data and predictive analytics in healthcare is, well, pretty big.

And judging from press releases and user-group meetings at big EHR companies, it should be any minute now that we have IBM's Watson installed and crunching predictive algorithms to prevent readmissions and diabetic mismanagement.

Don't hold your breadth.

Here's the truth: effective healthcare analytics cannot be bought. It must be grown and cultivated with culture, discipline and conservative, consistent funding.

Harnessing data requires more than expensive, fully-featured software. There needs to be a certain clinical-corporate culture, and distinct hiring practices; you need common ground and expectations set so that everyone, everyone, understands:

  1. Where to find data. (Metadata Management)
  2. Where the data came from. (Data Lineage)
  3. What the data means. (Business Glossary)

Litmus Test

Step on uppp! Don't be shy! The test below is simple and guaranteed to be as accurate as an online IQ test (or more!)


With your left hand, indicate with a finger each of the following that apply:

  1. My organization recognizes that effective analytics is needed to stay competitive with new operating models like All-Payer Reimbursement and Accountable Care in Vermont.
  2. My organization understands the importance of Data Governance and Stewards.
  3. My organization has a dedicated clinical analytics department.
  4. My analytics dept's leadership has experience with software product development, not just IT project management.
  5. The analytics department has one or more Data Scientists or Software Developers.

Now, with your right hand:

  1. My organization's analytics dept is comfortable with Data-Warehousing, able to quickly integrate data from different source systems (clinical, billing, patient satisfaction, gov't-released data sets.)
  2. My organization has invested in a Business Intelligence software stack like Qlikview, Tableau, Business Objects, or others.
  3. My organization has invested in a Metadata Management tool like Collibra.
  4. My organization's software developers can develop custom "side-car" applications that extend vendors' functionality using Application Programming Interfaces (APIs) and other integration tools.
  5. My organization's analytics dept has in place best-practice Development Operations (DevOps) tools and workflows to continuously build, test, release, and maintain quality software (e.g. JiraProject Management, GithubVersion Control, JenkinsAutomated Build/Test Server, ConfluenceDocumentation)

Count your fingers.

0 Fingers - Bad. Really bad. Shame on your organization if it has 300+ beds. It's not preparing for the changes to reimbursement models only 4-7 years down the road.

1-4 Fingers - Okay. Your organization is on the right path, but you've still got work to do and hearts & minds to win. If you are a large organization/medical system, you are falling behind.

5-8 Fingers - Solid. It sounds like your institution has already begun investing in analytics. Expect growing pains as you get buy-in from clinical leadership and learn how to manage these new technologies, together.

9-10 Fingers - Alright alright alriiight. That's how we do at AAMC.