ADONIS Cohort: High-Resolution Clinical Data for Neonatal Sepsis Detection

The SPHN ADONIS dataset is a high-resolution, multi-center clinical dataset designed to support research on the early detection of neonatal sepsis using machine learning. It includes data from over 15'000 neonates admitted to the neonatal intensive care units of three major Swiss hospitals—CHUV, USZ, and KiSpi—between 2007 and 2022. The dataset comprises >400 confirmed sepsis cases, >90 necrotizing enterocolitis cases, and a large control group. Data are collected at one-minute intervals and cover more than 800 variables, including vital signs, laboratory results, treatments, and device outputs. All data are semantically harmonized using SPHN standards and mapped to international ontologies such as SNOMED CT, LOINC, and ICD-10. Stored in RDF format and hosted on the secure BioMedIT infrastructure, the dataset is interoperable, pseudonymized, and ethically approved for research use.