BOSTON—Healthcare is at the cusp of being transformed by Big Data, said Thomas Hill, PhD, executive director for analytics, Dell Statistica Software, Information Management Group, speaking at the Big Data and Healthcare Analytics Forum.
Big Data will impact how the industry does research and optimizes processes, as well as how it evaluates drugs, patient outcomes, and alignment of key performance indicators. These all will be “fundamentally transformed through these technologies and it’s happening right now,” said Hill.
Healthcare also is at the beginning of individual use cases about readmissions, managing risk and how to incorporate text mining—the “bread crumbs that describe our lifestyle.” Marketing can trace people down to the house level, he said, with up to 1,500 points of data. That’s beginning to happen in healthcare because “when you walk out of the hospital, that’s when the healing starts. Lifestyle choices have a great impact.”
Much of these data don’t live in the EHR or billing data but in the unstructured notes. Systems weren’t set up for predictive analytics, Hill said, but to support the core business. Finding ways to mine the notes will help predictions get better. “You will find things you never imagined before. There’s a novelty to the things you find in unstructured data.”
The bread crumbs we leave on the internet are now very easy to collect, he said. The blogs a person reads, the groups a person belongs to all describe a lot of lifestyle choices which impact health and risks.
However, reaping the rewards of those data is difficult without the right people. “Data scientists are hard to find, expensive and difficult to keep,” said Hill. However, analytics will become more and more automated, he predicted. And because we cannot rely on data scientists, Hill expects to see the rise of citizen data scientists. Automation will allow for templates that can be understood and managed by the end user. A doctor or nurse, for example, could gain insights on their own based on predefined workflows.
Fraud and security are ongoing challenges, Hill acknowledged. He also noted that more and more people are studying algorithmic decision making. Those algorithms can be used to select people to whom to give a coupon. Even though it’s done without any human intervention it’s not fair to other people. Patients also are classified in some way when they enter the hospital.
“Most people are the same but also everybody is unique," Hill said. "Why should a treatment not be recommended to me when others believe it might be indicated?” He expects to see a regulatory framework that will govern how algorithmic decision making and modeling needs to be done when it impacts people.