NeoGx - Predicting Genetic Need in Level IV NICUs: analysis dataset

This resource contains data for reproducibility of the results in the manuscript NeoGx: Machine Learning to Predict Genetic Evaluation Need in the Level IV NICU. Included files: Phenotype vocabulary resources phecodeX_R_labels.csv: mapping of ICD-9, ICD-10 codes to Phecodes (available at https://github.com/PheWAS/PhecodeX) hp.obo: Human Phenotype Ontology (HPO) .obo file containing HPO term relationships, definitions, etc. phenotype_to_genes.txt: HPO annotation file mapping phenotypes to associated genes genes_to_phenotype.txt: HPO annotation file mapping genes to associated phenotypes genes_to_disease.txt: HPO annotation file mapping genes to associated diseases phenotype.hpoa: HPO annotation file containing relationships between phenotypes, genes, diseases, with frequencies and relationship qualifiers Model optimization data search-results.csv.gz: results of the Random Search carried out to optimize hyperparameters and feature sets Cohort data cohort-dataframe-calibration-validation.csv.gz: individual level demographics and outcome data for bias analyses and clinical impact projections cohort-split.csv.gz: individual membership in development/calibration/validation cohorts Feature matrices static-features.csv.gz: features calculated per individual (sex, gestational age, birth weight Z score, Level IV NICU admission age) pheno-features-vocab=phecode*.csv.gz: phenotype feature matrices derived from PhecodeX phenotype terms, for days 1...7, 14, 21, 28 in the Level IV NICU pheno-features-vocab=hpo*.csv.gz: phenotype feature matrices derived from HPO phenotype terms, for days 1...7, 14, 21, 28 in the Level IV NICU