EMERSE Search Engine Accesses ‘Untapped Data’ In EHR Clinician Notes

More than 12 billion words. That’s how much text is locked in the “free text” portion of electronic medical records at UMHS. And the standard way to access those words is to look through the medical records individually.

That is, unless you use EMERSE.

The University of Michigan Health System’s Electronic Medical Records Search Engine (EMERSE) is a secure, user-friendly tool that allows users to search the “free text” portion of electronic medical records. 

Initially released in 2005 (and with development ongoing), EMERSE provides easy access to data that was previously untapped. David Hanauer, MD, MS was interested in using search engines for clinical data and wanted a way for researchers to access the free text notes. “This was just untouched,” he says. “Here was the majority of the valuable information that was basically just this black hole that nobody could use.” He originally developed it as a tool for researchers in the Cancer Center, but opened it up for use by other researchers and U-M units. 

Tapping Difficult-to-Use Data
UMHS has had electronic health records (EHR) since 1998. Each record includes a section that clinicians use to make notes about a patient’s care, but Hanauer says it’s difficult use that information after the visit. “It’s basically just free-form text, and it’s very hard to make use of those kinds of data,” he says. “We have now amassed a collection of one hundred million of these clinical documents – so the question, of course, is ‘what do you do with all of this free-text information?’” 

EMERSE allows users to search the free text portion of EHRs from both the original “homegrown” UMHS system and the replacement vendor EHR system, MiChart, which was introduced in 2012. “It’s basically a search engine for the medical record that allows people to find information very quickly in this large collection of documents,” Hanauer says.

EMERSE has features that a typical search engine doesn’t have. For instance, users can save lists of patients that they can work through for a study. A collection of more than 100,000 synonyms and related key words lets users look for anything related to their initial search. “If you search for, say, ibuprofen, it will also look for things like Motrin, Advil, etc. because these are all variations on the same thing,” Hanauer says. Search results can be displayed and sorted in a variety of ways.

Usability is a big focus of the EMERSE project, and there is no instruction manual. “There are training videos, and some built-in ‘help’ functions, but the goal is for it to be really intuitive,” Hanauer says. 

Hanauer serves as Associate Chief Medical Informatics Officer for UMHS, Assistant Director of the Comprehensive Cancer Center Informatics Core, Director of Clinical Research Informatics for the Michigan Institute for Clinical & Health Research (MICHR), and Associate Professor in the Department of Pediatrics.

‘A Big Time-Saver’
EMERSE is primarily used by three groups of people: clinicians, operational unit staff, and research teams.

Clinicians can use the tool to access past notes about patients. For instance, a slight heart murmur is detected during an office visit. The patient remembers something about a heart murmur in the past, but doesn’t remember when or the details of the condition. The clinician could type in the word “murmur” and search the patient’s entire EHR to find where murmur had been mentioned before – EMERSE will highlight the word throughout the patient’s EHR, instead of making the clinician pore over all of the patient’s past records.

Several operations units also make use of EMERSE. For instance, Infection Control & Epidemiology uses it for infection surveillance and other kinds of problems. Risk Management employs EMERSE for safety evaluations, and Health Information uses EMERSE for a variety of purposes including the support of billing and quality improvement projects.

EMERSE was originally developed to assist researchers, and is still heavily used in that capacity. It can be used for general data abstraction, cohort identification, identifying adverse events in a clinical trial, and more. “It’s a big time-saver. Without this, they might have to read through 500 notes. Here, they might have to read through three notes,” Hanauer says. “It also lets them be a little bit more thorough.”

More than 900 studies have made use of the EMERSE search engine, and about 120 publications have resulted from the studies.

Offering EMERSE to U-M and Beyond
Hanauer would like to see the tool used at other institutions. “We’re encouraging others to adopt it instead of creating their own because we’ve got a pretty mature tool in place already,” he says, comparing it to REDCap, a research data capture tool developed by Vanderbilt University and now in use at many research institutions, including those of the Clinical & Translational Science Awards (CTSA) consortium. No one is building their own REDCap, he says, because the tool already exists and is available for their use; the same should be true for EMERSE.

Hanauer says that EMERSE is currently free for academic use. U-M provides the full source code and a detailed technical manual to institutions that use EMERSE and will provide support to help get the search engine up and running. 

“This requires a highly technical IT team to install it centrally and then make it available to all users,” Hanauer says. “It’s not something the individual user can install … but once it’s installed, it’s a self-service tool.” It will be worth it, he says, once people realize the capabilities of EMERSE.

The CTSA consortium has identified EMERSE as one of several tools that they consider “exemplars” of the consortium’s work, and is encouraging network-wide adoption of the tool. The U-M CTSA award is housed in the Michigan Institute for Clinical & Health Research (MICHR), which has provided salary support to allow Hanauer time to develop and maintain EMERSE. Hanauer is director of MICHR’s informatics program.

To learn more about EMERSE, visit the EMERSE website or email Dr. Hanauer.