Decision-tree analysis

This dataset contains state-level decision tree classification models developed to predict severe norovirus outbreaks in U.S. long-term care facilities using data from the National Outbreak Reporting System (NORS), 2000–2023. Separate decision tree models were constructed for 23 U.S. states with sufficient outbreak data (n ≥ 30 outbreaks), using routinely collected surveillance variables including outbreak year, month of onset, suspected etiology status, and mode of transmission as predictors. Outbreak severity was defined as illness count in the top quartile (≥39 cases). Each state-level model provides an interpretable, locally tailored classification tree identifying the threshold conditions and predictor combinations most strongly associated with severe outbreaks in that jurisdiction. These files supplement the manuscript "Spatiotemporal Prediction of Severe Norovirus Outbreaks in U.S. Long-Term Care Facilities Using Explainable Machine Learning" submitted to Health and Place.