A panel highlights potential pitfalls for healthcare analytics

In a panel discussion at MedCity CONVERGE this week, participants shared what they’ve learned from working with big data, especially when their big data ambitions hit bumps on the road and things go pear shaped.

Dr. Edward Ewen brought an interesting perspective to the panel discussion at the MedCity CONVERGE conference in Philadelphia this week. A physician, he now works as director of clinical data and analytics at Christiana Care Health System. Back in the day, he was one of the architects behind the assembly of Delaware’s health information exchange, the first in the country. It was no mean feat. Now, Christiana Care is “beginning our ACO journey.” He said, “I think a big part of the transition is shifting activities away from physicians and to the care team….In my experience, physicians love innovation and hate change. If you can craft tools to allow me to do what I want to do, you will see rapid adoption… We are not resistant to new tools.”

Hal Andrews, president of healthcare at Digital Reasoning called attention to one example of a failed analytics effort that was intended to identify sepsis cases. Unfortunately, the data the customer provided for the model had already been coded as sepsis, which pretty much defeated the purpose of the exercise.

“It’s one thing to have advanced analytics but getting it into a workflow in a timely manner is something else,” Andrews observed. “A critical part of the journey is the worklflow — delivering it to the right person who needs it when they need it.”

Nicholas Stepro of Arcadia observed that the need to avoid interfering with workflow can have it’s downsides, too. “It’s important to listen to end users, but you cannot be a slave to them.” He used the example of electronic health records. Because health IT vendors did not want to disrupt existing workflows, they did not take risks and create something that could have been easier to use.

In the question and answer session, an audience member wondered how  Ewen would handle a patient’s Fitbit data. “Where would you draw the line on patient-generated data? Where does that fit on the ethical line for you, as a physician? Ewen answered this way:

“I really feel like you need to have patient consent to do that and have transparency so they know how you are using the data. Having a default opt-in or opt-out will undermine trust or slow adoption.”

CORRECTION: An earlier version of the story had the wrong title and company association for Hal Andrews. It also misspelled Nicholas Stepro’s name.

Photo: Meghan Uno/Breaking Media 

Read this article:

A panel highlights potential pitfalls for healthcare analytics