A Model For Fostering Health Data Sharing

Sometimes, I’m amazed by what Facebook’s advertising algorithm can do. While most folks get pitches for hot consumer devices, shoes or casual wear, I get pitched on some cool geek stuff.

Most recently, I got an interesting pitch from data.world, a social networking site that helps members share and discover open datasets. The site is free to join, and if there’s a paid “premium” setting I haven’t found it. From what I’ve seen, this is a pretty nifty model which could easily be adapted for use by health IT organizations.

The site, which looks and feels something like Facebook, features data from a wide range of industries, tilted heavily toward government databases. For example, when I checked in, a front-page column listing the most commonly used tags includes “GIS,” “Homeland Security,” “police,” “SBA” and “DC” (which lead the pack with 688 mentions).

And there’s plenty of healthcare industry data to grab if you’d like. If you search for the term “healthcare” some useful datasets pop up, including a list of last year’s hospital HCAHPS ratings, California-specific data from 2005 to 2014 on the number and rates of preventable hospitalizations for selected medical conditions and New York state data on payments it made under its Medicaid Electronic Health Record Incentive Program. (You’ll have to become a site member to access these records.)

What makes the site truly interesting is the data sharing mechanism it offers. As a member, you have a chance to both upload open datasets, download datasets, post a project or join someone else’s project already in progress. Want help crunching the data on preventable hospitalizations in California? Let other site members know. There’s at least a chance you’ll find great project partners.

Of course, I’m not here to shill for this particular venture. My point in writing about its features is to draw your attention to what it does.

I think it’s more than time for healthcare organizations to collaborate on shared data projects together, and this is perhaps one mechanism for doing so. True, most of the data health systems work with is proprietary, but perhaps it’s possible to work past this issue.

Some healthcare organizations have already decided that sharing otherwise proprietary data is worth the risk. For example, late last year I wrote about a project undertaken by Sioux Falls, SD-based Sanford Health, in which the health system shared clinical data with a handful of academic researchers.  Benson Hsu, MD, vice president of enterprise data and analytics for the system, told Healthcare IT News this “crowdsourced” approached helped Sanford predict risk more effectively and improved its chronic disease management efforts.

Admittedly, Sanford’s approach won’t work for everyone. Today, healthcare organizations aren’t in the habit of cooperating on clinical data analytics projects, and anyone who suggests the idea is likely to get some serious pushback. Yes, in theory we all want interoperability, but this is different. Sharing entire clinical data repositories is a big deal. Still, how are we going to tackle big problems like population health management if we aren’t open to data analytics collaboration?

Sometimes new initiatives happen because people learn to understand each other’s needs, and decide that the prospect of mutual gain is worth the risk. I think a community devoted to data analytics could do much to foster such relationships.

About the author

Anne Zieger

Anne Zieger is a healthcare journalist who has written about the industry for 30 years. Her work has appeared in all of the leading healthcare industry publications, and she's served as editor in chief of several healthcare B2B sites.

1 Comment

Click here to post a comment
   

Categories