In Genome Medicine, a team from the University of Utah, Rady Children's Institute for Genomic Medicine, and elsewhere outline an automated prioritization pipeline for picking sick infants who are apt to benefit most from newborn whole-genome sequencing (WGS) or rapid WGS (rWGS) when time, personnel, and resources are limited. Their machine learning- and clinical natural language processing-based method, known as the "Mendelian Phenotype Search Engine" (MPSE), taps into electronic health record (EHR) databases found within a given institution to identify infants in the neonatal intensive care unit (NICU) who may have yet-to-be-diagnosed Mendelian conditions that might be detected with WGS or rWGS, the researchers write. They found that the approach accurately reproduced sequencing recommendations made by experts for more than 1,000 critically ill infants at Rady Children's Hospital and another nearly 3,000 additional NICU infants from the University of Utah. "We have demonstrated the feasibility of prioritizing individuals for WGS, using automated means, and that supplementing clinical review with this automated process could meet or exceed diagnostic yields obtained solely through manual review of clinical notes," the authors explain, noting that "[m]ore sophisticated machine learning techniques might further improve the accuracy of prioritization."