Unstructured EHR data may help clinicians identify seniors in need of additional coordinated geriatric care, according to a study published in the Journal of the American Geriatrics Society.
A team of researchers from Baltimore-based Johns Hopkins University and Newton, Mass.-based Atrius Health used a natural language processing algorithm to analyze free-text clinical notes from an estimated 18,000 senior patients at a large multispecialty group practice. The researchers' goal was to see if the NLP algorithm could help them detect various geriatric health conditions.
For the analysis, the team compared the number of geriatric syndrome cases it identified from unstructured and structured EHR data with the number of geriatric syndrome cases that would have been detected using only structured claims data. The researchers found that the NLP algorithm's analysis of unstructured EHR data helped them identify many more geriatric syndrome cases.
As an example, the researchers wrote that incorporating unstructured EHR notes allowed them to identify considerably higher rates of dementia among the patients in the study. The NLP algorithm detected 6.7 percent of patients exhibited symptoms of dementia, 1.5 times higher than previous estimates that had only used structured claims data.
The study authors concluded claims and structured EHR data provide an incomplete picture of geriatric syndromes.
"Geriatric syndromes are likely to be missed if unstructured data are not analyzed," the study authors concluded. "Pragmatic NLP algorithms can assist with identifying individuals at high risk of experiencing geriatric syndromes and improving coordination of care for older adults."