By John Otrompke 

April 27, 2015 | CHICAGO—Children’s Healthcare of Atlanta had a problem. Research on neonates had shown that the higher the decibel level in the ICU, the longer the length of stay. Now researchers wanted to explore exactly how external stimuli affected the outcomes of babies in the ICU, but were not sure how to go about it. 

“Data is not only used to test hypotheses,” said Michael Thompson, vice president of business intelligence and data architecture at CHOA. “Sometimes researchers say, ‘I need the data, so I can find my hypothesis.’ Of course that means having plenty of data on hand. Commonly, the data from an ICU bedside monitor is only kept three days, and then it gets thrown away. Thompson needed more than that. “So I became a data hoarder,” explained Thompson, who spoke Wednesday at a roundtable discussion, Better Care with Big Data, at the 2015 HIMSS conference.

“Someone from a university came in and said, ‘We’ll measure the decibel level every five seconds.’ I said, ‘That’s great, but what about the seconds in between? We’ll dumb it down to a Hadoop buster.’ Management asked, ‘Do we have that?’ ‘No, but we can build it.’”

The researchers ultimately learned that it was not the gradual increase in volume during the change in staff shifts, but the abrupt noises that occurred when an individual regularly opened a garbage can, for instance, that accounted for the change in outcomes. “Now we have millisecond data on the volume level, and it took five PCs and $600. I hope you all become data hoarders, too,” Thompson added.

While plugging huge amounts of data into open-source software like Apache Hadoop is inexpensive, the savings implied are huge, according to Ryan Goldman, director of product marketing at Cloudera, who also spoke at the roundtable. “There is a huge problem with access, and a problem with quality in health care in the U.S. But according to McKinsey, out of a potential savings of $750 billion in healthcare overall, there is an opportunity to save $450 billion from scaling big data analytics,” Goldman explained. “Imagine what it would mean if you could have 100% of the data on light quality and air quality in that ICU at marginal additional cost.”

One method hospitals are using to deal with the growing amounts of data they accumulate is to migrate to the cloud.

“With functions like patient education, HR, payroll, and for every application, we’re asking ourselves, ‘Does it make sense to have it in the cloud?’” said Melanie Rivero, Information System & Services Manager of Enterprise Management Systems, Legacy Health System in Portland, Oregon. Of 100 critical applications, one third are already cloud-based. The institution, which is affiliated with a burn center and a research institution, achieved the highest level of EHR adoption 2013, she said.

“Last year at HIMSS, everybody was asking, “Why the cloud?’ This year, attendees are asking, ‘Why not the cloud?’” explained Steve Fanning, vice president for industry strategy at Infor, who spoke with Rivero on Tuesday at an education session called Healthcare in the Cloud: Solutions for Transformation.

“A cloud provider can increase compliance. For example, with cloud-based functions, there is no physical tape [backup]. That’s just another opportunity for liability, because it’s not encrypted. Plus, you have to carry the tape,” explained Fanning.

Legacy Health has identified several keys to success to making a transition to the cloud, a year and a half after transitioning to Amazon Data Center. However, changing to cloud-based IT services is not without risks, so ‘caveat emptor,’ suggested Rivero.

“We don’t even know where the data center is located. I have asked, and they say they’re not able to give that information out. One instance occurred where half of Amazon’s west coast region went down. It… lasted for three hours,” she said.

More from HIMSS:  

HIMSS Keynote Speakers Highlight the Speed of Innovation