A study by the Regenstrief Institute and Indiana University has demonstrated the potential of using natural language processing technology to extract social risk factor information from clinical notes.
The NLP system developed by the research team can “read” and identify keywords or phrases indicating housing or financial needs (for example, a lack of a permanent address) and deliver highly accurate performance, the institutions reported.
By using NLP to search for social determinants of health, which often lack the standardized terminology found in a patient’s electronic health record, healthcare providers can more easily find and extract this data from clinical notes.
The study, published in the International Journal of Medical Informatics, analyzed more than six million clinical notes from Florida patients.
The system’s generalizability and portability were evaluated, showing ease and accuracy in adapting to new environments and data needs.
In addition, advancement in NLP technology could result in more cost-effective data extraction, thus allowing for a population health perspective and proactive interventions addressing housing and financial needs.
WHY IT MATTERS
Dr. Chris Harle, Regenstrief and IU Fairbanks School faculty member and senior author of the study, explained that, with more accurate or complete social information about their patients, healthcare providers may be able to tailor patients’ medical care with social needs in mind or refer patients to other services that help address them.
“This approach can be re-used for extracting other types of social risk information from clinical text, such as transportation needs,” he said. “Also, NLP approaches should continue to be ported and evaluated in diverse healthcare systems to understand best practices in dissemination and implementation.”
He pointed out text data in healthcare varies across organizations and geographies and over time.
“As such, methods to automatically extract information from text must be evaluated in diverse settings and, if implemented in practice, monitored over time to ensure ongoing quality,” Harle said.
THE LARGER TREND
Previously, Regenstrief Institute researchers developed three NLP algorithms to extract housing, financial and employment data from electronic health records.
The goal was to measure social determinants well enough for researchers to develop risk models and for clinicians and health systems to be able to use various factors.
The organization also developed an app called Uppstroms, which helps predict patients in need of a referral to a social service, such as a nutritionist.
The researchers said these studies show how AI models and NLP can leverage clinical data to improve care with “considerable performance accuracy.”
Dr. Harvey Castro, a physician and healthcare consultant, said he agrees integrating NLP for extracting social risk factors has tremendous potential across the healthcare spectrum.
“It can lead to faster, more personalized care and proactive interventions,” he explained.
He added the applications are vast, from the ER to primary care, mental health, chronic disease management, pediatric care, geriatric care and public health.
“NLP allows clinicians to treat the whole person by understanding their social risk factors,” Castro said. “It can lead to more personalized care, such as directly connecting a homeless patient with social services from the ER.”
Enhancing NLP with more complex algorithms can increase understanding of patient-specific nuances while they predict possible substance abuse issues or analyzing speech patterns might aid addiction intervention, he added.
“There is significant potential and wide applicability in using NLP to identify and address social risk factors, aligning with achieving health equity,” Castro explains.
For NLP applications to reach their full potential, however, providers must be on board and cater to caregivers, one expert said.
Nathan Eddy is a healthcare and technology freelancer based in Berlin.
Email the writer:Â nathaneddy@gmail.com
Twitter:Â @dropdeaded209
