Boston-based Massachusetts General Hospital researchers developed machine learning-powered software that can scan EHRs to predict the risk that a patient will be diagnosed with dementia in the future.
The research team, which published the study this week in Alzheimer's & Dementia, used EHR data on 267,855 patients from two large academic medical centers to create the software.
Researchers applied a natural language processing tool to comb through the EHR data and build a list of key clinical terms associated with cognitive symptoms of dementia identified by clinical experts. The software was then able to identify patients who are more likely to develop dementia.
With the software, researchers found that 2.4 percent of patients developed dementia over eight years of follow up care.
Study authors concluded that with further research, the Alzheimer's and dementia-detection tool could be duplicated around the world and offer more data and evidence for treatment trials.