Machine Learning, EHR Data Reveal Chronic Disease Associations

A machine learning algorithm can mine EHR data and discover novel associations between common chronic diseases and lesser-known conditions, which could support earlier monitoring or medical intervention, according to a study published in PLOS One. The use of EHRs in large health systems offers the opportunity to conduct population-level analyses that explore disease progression. The team wanted to identify novel comorbidities from routinely collected, anonymized EHRs. Researchers developed a machine learning algorithm, called the Phenome-Disease Association Study (PheDAS), to perform association studies and identify comorbidities across time in EHRs. The team validated the tool using three example conditions: Alzheimer’s disease, autism spectrum disorder, and optic neuritis, which can be the first indication of multiple sclerosis.