VA improves treatment with natural language processing tool
The new tool extracts relevant data from the case notes narratives in electronic heath records.
Natural language processing could dramatically reduce the time it takes for doctors to review electronic health records, thereby improving patient care.
A recent study by Dr. Jennifer Garvin with the Department of Veterans Affairs in Salt Lake City, Utah, tested the prototype Congestive Heart Failure Information Extraction Framework. CHIEF is aimed at improving the care of patients by reducing the time it takes for doctors to read through patient care notes to determine whether patients received an appropriate assessment of the their condition (in this case, the functioning of the heart's left ventricle) and are receiving the appropriate medications.
The study found that using natural language processing to review clinical information in text notes “makes fast and scalable quality improvement approaches possible, eventually improving management and outpatient treatment of patients suffering from [chronic heart failure],” according to the abstract in the Journal of the American Medical Informatics Association.
CHIEF is based on the Apache Unstructured Information Management Architecture framework, which supports analysis of unstructured information using search. It uses rules, dictionaries and machine learning methods to find mentions in text notes of electronic health records related to heart function, medications and any documented reasons why a patient would not be receiving these medications.
The study included 1,083 heart failure patients discharged from Health Administration medical centers in 2008 and 2009. CHIEF scanned and processed the patients’ electronic health records stored in the VHA data warehouse and created a table of summary information for easier analysis of the records.
CHIEF had an accuracy rate of at least 95 percent in finding mentions of the two drug therapies and of past measurements of left ventricle functioning.
“Garvin's study shows that an extraction system has the potential to virtually eliminate the need for primary care providers to conduct tedious searches in free text for critical details,” according to an article on the study, written by the VA’s Office of Research and Development. The study demonstrated that clinicians could “accurately process veterans' records with minimal to no manual chart review to learn if they received recommended care.”
According to Garvin, "the overall goal was to reduce the burden on primary care providers and the health care system, to undertake quality measurements, and to use the data in applications such as clinical decision support.”
Garvin focused her study on cardiovascular disease because it is the number-one killer of Americans and the leading cause of hospitalization in the VA. She said her group may expand the software’s use to better understand other aspects of heart failure, such as the disease’s progression, as well as extracting information on other illnesses, perhaps even throughout the VA health care system.
NEXT STORY: Defense applications ripe for quantum computing