https://researchdata.se/en/catalogue Search results en This dataset contains the fitted statistical models behind the ED Atlas. They are Bayesian logistic regression models that give the probability of an outcome after an emergency department (ED) visit, given the patient’s sex, age and presenting complaint. The models were trained on the Swedish SwED cohort, which covers ED visits in [regions/hospitals, years, approx. number of visits]. SwED is linked to the National Patient Register and the Cause of Death Register (Socialstyrelsen). The material is model parameters. It contains no individual-level data and no personal data. Each model is stored as a NetCDF file (.nc) with posterior samples of the regression coefficients, drawn with MCMC (PyMC 5.17.0, numpyro sampler, 8,000 draws per model). The files can be read with any NetCDF/HDF5 reader, for example ArviZ/xarray in Python. The dataset contains: One model for 30-day mortality, defined as death within 30 days of ED arrival, and one base model with sex and age only. 930 diagnosis models, one per ICD-10 chapter, block or three-character code. Each gives the probability of that diagnosis being registered [define: time window and care setting] after the ED visit. For each model, a CSV file listing the column order (the SNOMED CT complaint codes it includes). A README with step-by-step instructions, in pseudocode, for making a prediction from sex, age and a presenting complaint. Model inputs are sex, age (adults, 18 and over) and one presenting complaint coded in SNOMED CT. Age is modelled with a natural cubic spline, and sex, age and complaint interact. Presenting complaints use SNOMED CT concept identifiers. The SNOMED CT terms themselves are not included. The material can be reused to compute risk estimates for new patients, to compare with other models or cohorts, or to study how complaint, age and sex relate to outcomes. Fri, 13 Nov 2026 12:25:35 GMT https://researchdata.se/en/catalogue/dataset/2026-360 https://researchdata.se/en/catalogue/dataset/2026-360 Uppsala University Nicolas Pielawski Stefan Gustafsson Ludvig Hult Frida Blankenau Thomas Schön Susann Järhult Johan Sundström