Machine Learning Algorithm Predicts Dementia Risk Based On EHR Data

A machine learning algorithm used with consumer electronic health records (EHRs) was able to predict the risk of an individual developing dementia eight years before any diagnosis. The initial analysis included data from 267,855 hospitalized adults with no previous dementia diagnosis; they were followed for eight years. Each individual was assessed for dementia risk based on data in the EHR. During the follow-up period, 2.4% (6,516 individuals) were diagnosed with dementia. Those in the initial highest risk quartile were 7.4 times more likely than those in the lowest risk quartile to receive a dementia diagnosis . . .