With the digitization of hospitals and the growing body of EHR data, there is increased demand to generate knowledge from the available data and create innovative solutions to challenges in healthcare.

Unfortunately, major obstacles exist to achieve these goals, hindering the speed of healthcare innovation. Significant barriers include the investment required to transform clinical data repositories into a suitable format for analysis and the dearth of available expertise in analyzing complex big datasets.

Other obstacles include the need to respect and protect patient rights and privacy, the time required to cleanse and curate data in a reproducible way and keeping up with the rapidly changing field of data science and artificial intelligence.

"Being able to analyze data across a diverse set of hospitals is vitally important to creating knowledge and tools that will benefit patients," said Dr. Omar Badawi, director of medical device safety at the Medical Device Innovation Consortium. "There are several advantages to crowd-sourcing knowledge-generation from EHRs."

First, because EHR data was not collected for research, it is inherently messy and difficult to curate: No one person, or even one team, has the capacity or capability to understand and address all the nuances within databases representing such a diverse set of clinical scenarios, variations in clinical practice, documentation patterns and software usage.

"Crowd-sourcing enables others to refine the way we look at the data and improve all of our understanding by sharing their methods with the broader community," he said.

Badawi, who will address the topic at HIMSS21 next month, said it is understandable that healthcare systems want to view their data as an asset, but this perspective is stifling towards innovation and ultimately not in their or their patients' best interests.

"They should look at creative ways to safely collaborate with others, as there are many recent innovations allowing this to happen more easily than before," he said. "For example, I am currently leading an effort ... to create a federated learning network to support active surveillance for medical device safety."

Badawi said this type of system supports data to be analyzed locally and only shares derived statistics, ensuring patient privacy.

"By participating in these types of networks, health systems can learn from their peers without sacrificing their privacy," he said.

He said the biggest challenge to generating clinically useful knowledge from clinical information systems and EHRs is that things are not always as they seem.

"Everybody working with this data needs to recognize the likelihood of unknown unknowns," he said. "Basically, you have to question everything and assume nothing when interpreting data."

Dr. Omar Badawi will speak on using data crowd-sourcing at HIMSS21 in his session, "An Ecosystem for Innovating Through ICU Data Crowd-Sourcing." It's scheduled for Wednesday, August 11 from 2:30 to 3:30 p.m. in Wynn Lafite 2.

Nathan Eddy is a healthcare and technology freelancer based in Berlin.
Email the writer: nathaneddy@gmail.com
Twitter: @dropdeaded209