Electronic Medical Records

IBM Research (@IBMResearch) may be best known for its scientists who have created technology breakthroughs from moving atoms to analytics systems like IBM Watson. But what might surprise many is the number of deep subject matter experts now working side-by-side with computer scientists to redefine how industries apply technology to improve the services they offer.

The world's healthcare systems, for example, are aging and increasingly complex. Despite amazing advances in medicine and patient care, the system is still plagued by misdiagnosis, misaligned incentives that increase cost, and lack of access to the latest information on patient medical history and treatment options.

Today, a team of medical doctors with more than a century of combined experience is working inside IBM Research with scientists from a variety of disciplines. They are examining ways to improve the healthcare system from all angles - from using data analytics for better-informed diagnoses to understanding why certain diseases flourish in some regions of the world but not others. Perhaps most importantly, they are helping researchers deepen their understanding of the industry, in order to apply technology more effectively.

In the U.S. alone, it is estimated that up to 20 percent of diagnoses are either incorrect or incomplete. In addition, there are an estimated 1.5 million errors in the way medications are prescribed, delivered and taken in the U.S. every year.

These mistakes could be greatly reduced if doctors had access to the latest relevant medical information. But there is so much medical data in the world that it is impossible to keep track of it all.

The MDs are now working with researchers to apply IBM Watson's capabilities to answer natural language queries - which it first demonstrate by defeating the world's best human contestants on the quiz show Jeopardy! - to search through tremendous amounts of both structured and unstructured medical data, from journals to CAT scans, to provide practitioners with vital information.

"Watson will leverage existing evidence in new ways," said Josko Silobrcic, MD, Senior Medical Scientist at IBM Research, "and with natural language processing and machine learning, it may uncover new patterns that may not have been recognized before."

Martin Kohn, MD, Chief Medical Scientist for Care Delivery Systems at IBM Research, is helping design Watson's capabilities in a manner that best supports the way clinicians work. He said physicians tell him, "I'm looking for something to bring help to me when I need it, in a format that is useful."

Tomorrow.

That's not an idea modern society seems to care much about anymore.

So how do health IT professionals, who are busy planning for tomorrow, convey the importance of their activities to individuals and colleagues who are concerned only about what they need today? To bridge that chasm, you first must convey to them why your health IT activities will be important in the future, no small task considering the complexities of HIT architecture and healthcare standards.

And why, exactly, are your health IT activities important? It's simple: patient care.

Sepsis--a medical condition in which the body has a severe inflammatory response to bacteria or other microbes --costs the US approximately $14.6 billion a year, according to the Centers for Disease Control and Prevention. Several initiatives put in place at the University of Kansas Hospital are making a dent in those numbers, resulting in a savings of $18,000 per patient, or $18 million a year.

The university hopes to take this success one step further with the use of CareVeillance, a first-of-its-kind sepsis identification tool that replicates patient information from an electronic medical record (EMR) and monitors patients in real time for the complication.

Bryan Eckert, senior principal at CSC Health Delivery Group, said in an interview with InformationWeek Healthcare that the CareVeillance system includes customer-designed workflows and algorithms to provide this situational awareness, and it automates risk scoring while finding early warning signals often buried in the EMR.

"At a high level, we use industry-standard means of replicating patient information from an EMR: it doesn't matter whose it is," Eckert said. "We're grabbing information at the interface level ... we're grabbing it at two tiers."