The Golden Age of Biomedical Informatics? Or the Best of Times, the Worst of Times?
A keynote presentation at the Oct. 23 Mid-Atlantic Healthcare Informatics Symposium in Philadelphia highlighted some of the reasons it is exciting to be an informatics specialist these days. Jason Moore, Ph.D., director of the Institute for Biomedical Informatics at the University of Pennsylvania, said we are entering a “golden age of biomedical informatics. “This is our time,” he said. “We have more data than we know what to do with, access to high-performance computing resources, and talented trainees who will become the next generation of informaticians.”
Moore went on to say it also is a great time to be building informatics communities. There is more government recognition than there used to be. NIH has hired a chief data scientist and started the Big Data to Knowledge effort. “Previously, NIH had not paid as much to informatics as it should have, but now they are,” he added. Also there is increasing industry recognition, and university investment. At a time when many medical schools are struggling financially, the fact that universities are carving out funding for new informatics programs signals the recognition of its increasing importance, he said.
Moore focused the rest of his talk on progress taking place in artificial intelligence, visual analytics, and what he called “no-boundary” thinking. He said IBM’s Watson is a pretty important advance, and a “signal to our community that artificial intelligence is ready for prime time. I have issues with it, but it signifies this golden era of natural language processing, machine learning, and knowledge representation.” He said visual analytics originally was overhyped and under-delivered on its promise, but he expects that to change soon. He mentioned that next year Penn’s Institute for Biomedical Informatics plans to open an “Idea Factory” visual analytics space with 3-D visualization and data analysis capabilities.
Explaining his “no-boundary thinking” concept, Moore said that rather than being advisors to research projects, informaticians may start to play a more proactive role in asking good questions of the data, because they are agnostic to the specific medical discipline. “We can look across knowledge sources and ask more impactful scientific questions. Now that the data is accessible, we are in a much better position to use others’ data and institutional data.”
If Moore sees a golden age opening up, another presentation had an intriguing title “A Tale of Two Cities: A Clinical Decision Support Update from Two Leading Children's Hospitals.” Both speakers, Eric Kirkendall, M.D., associate chief medical information officer at Cincinnati Children's Hospital Medical Center, and Eric Shelov, M.D., associate chief medical information officer at the Children's Hospital of Philadelphia, referred to the first line of the famous Dickens novel: “It was the best of times, it was the worst of times,” but they weren’t sure how it applied to their talk. But their work is definitely more in the trenches, trying to decrease alert burden on clinicians, increasing clinical decision support alert salience, and improving patient outcomes.
Kirkendall described the work in Cincinnati fine-tuning alerts about antibiotic drug overdosing. By cutting down on the volume of alerts and making them more clinically appropriate, they are making progress on cutting down on patient safety issues. One challenge is how to scale that work to the hundreds or thousands of rules in the system.
Kirkenall also described work on a situational awareness tool called GARDIANS (Global Automated Risk Detection Interface and Network System) for Operational Excellence. It serves as a front-line tool that comprehensively presents information to enable front-line leaders to identify and mitigate risk for flow failures, patient and family experience failures, and safety events. “It is a real-time web-based application that aggregates situational awareness data around safety, patient flow, and patient and family experience and is used by clinical leaders on secure mobile devices.” He said the hospital built a custom technology framework to leverage any vendor API. “It gets key data points to the right audience at the right time, in the right format.” For instance, for patient flow, an executive can quickly look unit by unit and scroll the whole hospital for hot spots, which patients are on high-risk therapies, the number of beds per unit and a predictive number for the end of the day. They can drill down to a patient-level view.
Shelov talked about an emerging partnership between the informatics team and the hospital’s quality improvement program. He said a common problem with CDS is that someone would get enthusiastic and build a solution without clearly understanding the problem or the potential consequences of the solution. Those interventions often proved inefficient, led to poor satisfaction, and were sometimes dangerous, he said.
Working with the Office of Quality Improvement, informatics experts and IS staff are now following a process of asking questions about the specific problem and looking at consequences more clearly. It also is a more rapid cycle “agile” development process. “It has really improved the development of CDS and driven toward better processes,” Shelov said. “Both teams had to step outside their comfort zones. The quality improvement team got a crash course in unintended consequences. For the informatics team, it was a learning experience on how we can do new tool exploration. Rarely do we have opportunities to do something proactive, because we are so busy keeping the lights on. But this allows us to get agile and iteratively test and change as we went. That was unfamiliar to us in applied informatics. So instead of starting with the intervention, we start with the problem we are trying to solve and the CDS will follow,” he said. It is reducing inefficiencies and dissatisfaction, and helped streamline the development process.
Overall, it was a great symposium, once again. I hope to write about a few other sessions next week.