Machine Learning Model Sets Global Standard for Newborn Genomic Screening
Mass General Brigham researchers publish a data-driven framework to harmonize gene selection across 27 international programs, marking a milestone in newborn genomic screening initiatives.
Overview
- The machine learning framework analyzes 4,390 genes across 27 newborn genomic screening (NBSeq) programs, addressing significant variability in gene selection worldwide.
- Only 74 genes, or 1.7% of those studied, are consistently included in over 80% of global NBSeq programs, highlighting the need for standardization.
- Key predictors for gene inclusion include U.S. Recommended Uniform Screening Panel (RUSP) listing, natural history evidence, and treatment efficacy.
- The model uses 13 predictors to achieve high accuracy in ranking genes, offering a scalable, adaptable tool for evidence-based decision-making.
- This is the inaugural publication from the International Consortium of Newborn Sequencing (ICoNS), founded in 2021 to promote global collaboration in NBSeq efforts.