Abstract
Public Health Scotland (PHS) is refining its hospital bed forecasting model, initially focusing on heart diseases, dementia, and respiratory conditions to enhance data-driven decision-making. They seek to improve predictive accuracy for patient length of stay (LoS) by integrating patient diagnoses and various data sources, including open, disclosive, and synthetic data. However, effectively conveying model uncertainty to a diverse audience is a challenge. PHS envisions a collaborative framework that extends beyond a single model, promoting ongoing data challenges and innovation in public health. This initiative involves coordinating research, data sharing, and collaborations to enhance healthcare accuracy and bed availability understanding in Scotland. The report discusses efforts during ESGI 171 to enhance the PHS Whole System Model by improving predictions of hospital stay length for patients with specific characteristics. It includes an analysis of anonymised and depersonalised data, outlines methodological approaches, and explores results and potential future developments.