Massachusetts General Hospital researchers have found a way to maximize patient information in EHRs to predict likelihood of developing diseases.
The study, published in Cell Patterns, details how researchers used machine learning to track patients' medical records over time and proposed an algorithm for exploiting the temporal information in EHRs for predictive analytics. The algorithm connects EHR information about patients' medications and diagnoses over time instead of based on independent health records to compute the likelihood that patients could have an underlying disease.
"Our study doesn't rely on single diagnostic codes but instead relies on sequences of codes with the expectation that a sequence of relevant characteristics over time is more likely to represent reality than a single element," said Hossein Estiri, PhD, of the Mass General Laboratory of Computer Science and lead author of the study. "Additionally, the computer sorts through thousands of patients and can find sequences that a physician would likely never identify on their own as relevant, but actually are associated with the disease."
For example, information in the medical record indicating patients had coronary artery disease followed by chest pain could predict heart failure development. Clinicians can also use the algorithm to identify disease markers, which could be a step toward developing computational models for identifying and validating new disease markers.
More articles on EHRs:
AdventHealth moves Epic EHR to cloud: 3 notes
Geisinger to implement EHR-agnostic tool for real-time COVID-19 surveillance
16 hospitals, health systems seeking Allscripts, Epic, Cerner, Meditech talent