Clinician Decision Support Dashboard: Extracting value
from Electronic Medical Records using Text Mining
Iccha Sethi, Zhaohui Sun, Harold R. Garner
Abstract Introduction Proposed Interface
The conversion of medical records to electronic form An Electronic Medical Record is an electronic record of health-related
is proceeding quickly; for example, Carilion Clinics is information on an individual that is created, gathered, managed, and
one of the leading patient care organizations with over consulted by licensed clinicians and staff from a single organization who are
70% (and heading towards 100% in the next few involved in the individual's health and care. The Clinical Decision Support
years) of its patients have Electronic Medical Records Dashboard aims to extend the purpose of EMRs beyond information storage
(EMRs). Although EMRs have been a benefit for and use the data to help doctors find similar de-identified electronic medical
efficiency and administrative needs such as billing and records and other related medical information from medical literature and
logistics, doctors want to see enhanced benefits for the journals, thereby aiding them to provide better health care.
patient and their delivery of care, and patient care
facilities want to see more cost saving enhancements Design
without sacrificing quality or delivery of care.
Therefore, what is needed is a system that will analyze Clinician
a patient’s evolving EMR in context with all available Input
biomedical knowledge and the accumulated
experience recorded in the EMRs of other patients.
The aim of the Clinician Decision Support (CDS)
Dashboard is to provide interactive, automated,
actionable, EMR text-mining tools that helps improve
both the patient and clinical care staff experience. Related Similar Case Clinical
Literature EMRs Reports Trials
The CDS Dashboard, in a secure network, will help
physicians find de-identified EMRs similar to their
patient's medical record thereby aiding them in
diagnosis, treatment, prognosis and outcomes.
Hypothesis Predictions Best matches based on Conclusion
Because some cases involve complex disorders, it will The clinical decision support dashboard described is a system which
also allow physicians to search medical literature, integrates text mining methodology with various sciences like medicine,
recent research findings, clinical trials and medical marketing, human computer interaction and medical informatics thereby
Diagnosis Treatments Outcomes Warnings
cases, enabling clinicians to also become researchers, providing a comprehensive approach to medical healthcare.
while simultaneously making certain their patients get
the latest in care. Another feature of the Dashboard is
that it will also provide suggestions for drug
Hidden Dx References
alternatives, counter indications and drug-drug Errami, M., J. M. Hicks, et al. (2008). "Deja vu--a study of duplicate
interactions. Text similarity searching technology citations in Medline." Bioinformatics 24(2): 243-249.
(etblast.org) developed by our group will be used as Challenges Errami, M., J. D. Wren, et al. (2007). "eTBLAST: a web server to
the similarity search engine for the Dashboard. identify expert reviewers, appropriate journals and similar publications."
There are numerous challenges in processing EMRs, a few of which are
Another critical part is the development of local Nucleic Acids Res 35(Web Server issue): W12-15.
databases to which the patient’s EMR is compared, Lewis, J., S. Ossowski, et al. (2006). "Text similarity: an alternative way
There are a variety of ways each physician makes entries into the
and these include MedLine, the Orange Book, the to search MEDLINE." Bioinformatics 22(18): 2298-2304.
medical record; the use their own set of acronyms and shortcuts.
Physician Desk Reference and others which then link Report, I. (1999). "To Err is Human: Building a Safer Health System."
The presence of billing and CPT codes.
the latest in medical knowledge to a given case. National Academies Press.
Combinations of unstructured and structured laboratory data.
Zhou, L., C. S. Soran, et al. (2009). "The Relationship between
Presence of lexical variants and synonyms.
The design, features and functionality of the CDS Electronic Health Record Use and Quality of Care over Time." Journal
Dashboard are described. They require a of the American Medical Informatics Association 16(4): 457-464.
multidisciplinary approach to the construction of the
system, including computer science, medicine,
biomedical research, marketing and human-machine