March 25, 2021 – According to a study, an EHR extraction system could be key to translating unstructured text about patients’ travel history into actionable health data released in JMIR publications.
Without an automatic extraction system, doctors would have to manually review the travel graphs, use a specific EHR system that dictates documentation of the itinerary, or ignore the itinerary entirely.
The spread of COVID-19 was urgently needed integrate Travel history information in the EHR. Implementing the itinerary into the EHR can help relate infectious symptoms for doctors.
When implemented as vital signs along with temperature, heart rate, respiratory rate and blood pressure, the travel history can add detailed patient data, initiate further tests and trigger protective measures for those who come into contact with the patient.
EHRs can also be incorporated into travel history to customize instant diagnosis for returning travelers, similar to cardiovascular disease Risk calculator can show the patient a personalized list of possible lifestyle changes.
Although the Department of Veterans Affairs (VA) is currently integrating travel history into patients ‘EHRs, the research team evaluated the feasibility of annotating and automatically extracting mentions of travel history from physicians’ notes that are available as unstructured text in various healthcare facilities for reference react Public health emergencies.
Researchers created a standard for EHR recognition of patient journeys through manual abstraction of patient charts and developed an automated text extraction pipeline.
Of over 4,500 annotated EHRs, 58 percent contained travel history mentions, 34.4 contained no travel history, and the remainder were indefinite. The research team said the accuracy of the automated word processing and the burden on doctors were acceptable enough to allow for rapid screening in the future.
The travel history varied from semi-structured questionnaires such as “Have you visited a region known for the transmission of Zika?” to “Has the patient recently returned from Brazil, Mexico, or Miami?” to “Went to Europe”.
Several disagreements among researchers arose from the different attributions of past confirmed trips versus future or hypothetical trips.
For example, one researcher identified “traveling to visit sister in Hungary in May” as a future trip, while another identified this example as a past confirmed trip.
In addition, the study’s authors expected military deployment sites, but the patient was not always deployed. Some EHRs would display “Service Era: Vietnam” but that does not mean the patient has traveled to Vietnam.
“The site agreement was calculated for all annotated site text areas and required an exact match between text offset and negation status,” the researchers explained. “Each difference in status was treated as a disagreement, and each difference in the range of text was treated as a separate annotation element. The record agreement combined each annotated location status so that each snippet was assigned a class that was either unmentioned, negated, positive, or mixed. “
The research team identified 561 different locations across 8,127 location areas.
“Our results show that training an accurate model to extract travel mentions is possible in an automated system,” the study authors wrote. “Both the flagged sets and the modeling approaches were chosen to minimize the development time and computational resources required to continue monitoring day-to-day operations. The baseline comparison presented here is a simplified assessment, but shows that general-purpose geoparsing solutions alone result in lower accuracy. “
Since the research team developed the technology three years before COVID-19, its use was limited during the spread of the coronavirus, as travel was only a relevant risk factor in the early stages of transmission, the study authors wrote. When researchers developed the tool, its capabilities were mostly focused on people bringing infectious diseases to the United States.
“The Centers for Disease Control and Prevention (CDC) guidelines for examined subjects on February 12, 2020 contained explicit mentions for traveling to Wuhan or Hubei Province,” the study’s authors said. “By March 4, the CDC removed these criteria and instead encouraged clinicians to use best judgment when testing virus. In some surveillance efforts, travel history was considered less important in risk assessment once acquisition increased in the community. “
Researchers could use the method in the future to prevent and contain the further spread of COVID-19 and the spread of other infectious diseases.