Using Big Data to Shape a New LHS

Critical Care 2.0: Integrating Big Data, Clinical Trials, and Implementation Science to Create a Learning ICU System

9:15-11:15 a.m., Sunday

Room C146 (Level 1), KBHCCD

Imagine a health care system so advanced that it continuously gathers and analyzes the massive amount of data generated in the ICU to improve patient care in real time.

“When the system is optimized, patients can be confident that we’re learning from the care we provide every day,” said Vincent Liu, MD, research scientist at the Kaiser Permanente Division of Research in Oakland, California, and a co-chair of the session. “That means safer care and better outcomes for patients.”

Vincent Liu

Vincent Liu

Dr. Liu is speaking of a Learning Health System (LHS) that leverages high-quality evidence, internal data and informatics, and systematic implementation to improve everyday patient outcomes in critical and other care settings.

“We care for hundreds, thousands, millions of patients each day,” Dr. Liu noted, explaining the traditional approach to improving patient care can take a long time to get and implement results.  “The LHS describes a feedback loop that uses data from our practices so we can make our care better with a relatively short lag time.”

An LHS doesn’t happen naturally, though. It requires investment and expertise from health care’s top minds to get the system running. Much of it has to do with extracting big data from existing electronic health record systems (EHRs). “We’re sitting on billions of data elements that are essentially completely unused,” Dr. Liu said.

Matthew Semler, MD, pulmonary critical care physician at Vanderbilt University Medical Center in Nashville, and session co-chair, explained the challenge of extrapolating existing and new EHR data and feeding it into an algorithm to improve hospital operations and determine which treatments are better for which patients. That challenge can be overcome, though, by a combination of medical informatics and IT support in conjunction with hospital leaders, physicians, and medical researchers.

“The challenge the LHS faces is identifying common interventions where there is variation between providers,” Dr. Semler said. “This information, combined with patient outcomes, could provide new evidence that can be applied for better patient care.”

Patient privacy is always a concern. “We often talk about the limits of patient protection, HIPPA, and how we navigate those,” Dr. Semler said, noting that there are encryption tools that can be applied to keep records anonymous.

“This all depends on us making strategic tweaks to the care we deliver every day and analyzing what those results are,” Dr. Liu continued. “The flip side is to never learn from the data derived from the care we deliver patients.”

Want More?
Don’t miss these sessions that also explore how learning may be shaped by data and artificial intelligence, both taking place in Ballroom C One-Two (Level 2), KBHCCD.

8-8:45 a.m.
Data Sharing in the Context of Clinical Trials (K3)

8-8:45 a.m.
What Should Pulmonologists Know About Artificial Intelligence and Machine Learning? (K7)