Pittsburgh-based UPMC combined machine learning and whole genome sequencing to develop an artificial intelligence-powered system to prevent hospital-based disease outbreaks, according to a study published Nov. 17 in Clinical Infectious Diseases.
The system, called EDS-HAT, uses genomic sequencing surveillance to identify hospitalized patients who share the same strains of an infection. UPMC collaborated with researchers at Pittsburgh-based Carnegie Mellon University to examine patients’ EHR to determine what such patients have in common, such as a shared provider or procedures that use the same equipment.
The system uses data mining and machine learning to produce a list of transmission routes, such as tracing a group of infections to the same X-ray room, study author Lee Harrison, MD, told the Pittsburgh Post-Gazette.
UPMC tested the system with a six-month lag time from November 2016 to November 2018. It identified 99 groups of similar infections, finding at least one potential transmission route in 65.7 percent of those groups.
The research team estimates as many as 63 infectious disease transmissions could have been prevented if the system had been operating in real time. The team also estimated that extinguishing these outbreaks would have saved the health system as much as $692,500.