Moving Beyond Scorecards, Dashboards to Address Quality in Imaging Informatics | Healthcare Informatics Magazine | Health IT | Information Technology Skip to content Skip to navigation

Moving Beyond Scorecards, Dashboards to Address Quality in Imaging Informatics

September 22, 2017
by David Raths
| Reprints
Embrace human-machine collaborative workflow orchestration, says University of Chicago’s Paul Chang, M.D.
Click To View Gallery

There has been quite a bit of speculation, hype and fear mongering about machine learning software eventually replacing radiologists. Paul Chang, M.D., medical director of imaging informatics at the University of Chicago Medical Center (UCMC), argues that imaging informatics needs to leapfrog into what other business sectors are doing and embrace human-machine cybernetic collaborative work flow orchestration to improve both quality and efficiency.

In a Sept. 21 webinar sponsored by the Society of Imaging Informatics in Medicine, Chang said informatics had a central role to play in optimizing quality in radiology. But he started out by saying that although quality is something that everyone in healthcare expects, few leadership teams are willing to fund projects targeting quality improvement.

 “Unfortunately, we have not gone beyond lip service when it comes to quality,” said Chang, who is also a professor of radiology and vice chair of radiology informatics at UCMC. “Every time I go to the C-Suite and say please give me funding, because if you do I will improve the efficiency and improve throughput and reduce cost, that gets prioritized. When I pitch something that will improve quality, they say ‘That is a great idea, write a paper, get a grant,’ but they won’t fund it.”

Chang used an extended metaphor of buying a car to say that quality is like the floor mat of the car. You spend all your time thinking about the size and color and horsepower of the car you want to buy, but you expect the floor mat to be there, thrown in for free. “Quality is the floor mat,” he said. “What do I mean? We are not willing to spend to achieve it on its own. Even though the new models of shared risk demand it, we tend not to get the resources necessary to achieve quality.”

But Chang has decided that if the car is efficiency and productivity, the name of the game now for people in IT is to design systems to improve efficiency and productivity (the car) and quality (the floor mat,) at the same time. “The reason that is frequently possible is because both efficiency and quality loathe the same enemy: variability. What is the greatest source of variability in a hospital enterprise? People.”

Chang noted that other industries have bent over backwards leveraging IT systems to orchestrate workflow systems to minimize the negative potential impact of people. “People are absolutely critical and important,” he said, “but we in healthcare tend to put them in the worst possible positions.”

He said radiology IT efforts can go beyond the dashboards and scorecards that most people seem to be satisfied with when they use IT to approach quality issues. He gave several examples of how UCMC has leveraged IT tools to support radiologists, nurses, and technicians. He described a clinical context lookup tool that scans the EHR for all data relevant to a particular patient and summarizes it.

“We built a system that extracted information from the EMR, pathology reports, lab results, progress notes — to understand what is really going on, to augment the sparse and incomplete information that was in the CPOE,” he explained. “We are now using this with some machine intelligence and natural language processing tools that take this information and create a very rapid abstraction that says something like: ‘We have reviewed the prior reports and history. Here is what you need to know: This patient has lung carcinoma and the white blood cell count is elevated; someone recommended in a prior CT that you look at this lung nodule in the superior segment of the left lower lobe.’

“Instead of me spending time inefficiently searching in the EMR for that information and resulting in variability with respect to quality, we can use machines not to replace me but to augment me, to review this information and present it to me in an efficient manner,” he said. “We can also use just-in-time decision support, to augment not only the clinical context, but also potential variability in my knowledge context. It is not enough to have the imaging information. I need the clinical context. This is an example where decision support tools, human-cybernetic collaboration, machine intelligence can help — not to replace us, but to augment us.”

Chang said, for instance, he is good at selecting a lesion. “But I can let my cybernetic buddy find it in all the other images. That can save 10 minutes.” And if there is an incidental lung nodule that needs to be followed up on in six months, currently that depends on humans to remember to do it. “But as I measure the nodule, the cybernetic buddy can be looking up risk factors and recommendations and automatically populating a database for follow-up.”

“We humans have a great contribution to make to workflow, but one of the things we are not really good at is remembering to do things,” he said. “Like other industries we should be leveraging electronic-based workflow orchestration instead of relying on humans to remember to do the right thing. Evidence-driven workflow is key.”














Get the latest information on Health IT and attend other valuable sessions at this two-day Summit providing healthcare leaders with educational content, insightful debate and dialogue on the future of healthcare and technology.

Learn More