This document discusses predictive modelling to identify patients at high risk of unplanned hospital admissions. It notes that a small percentage of patients account for a large share of emergency bed days and unplanned admissions are expensive, lead to poor patient experiences, and may indicate suboptimal care. The document presents data showing predictive models can accurately identify patients at high risk of readmission and describes how the models have been used to develop interventions targeted at high-risk patients, leading to reductions in emergency admissions. It concludes by discussing expanding the use of Hospital Episode Statistics data to improve care coordination and integration across health systems.