Linking Predictive and Prescriptive Analytics of Elderly and Frail Patient Hospital Services

Abstract

Predictive and prescriptive techniques are being evaluated to predict demand for inpatient services within South East Wales. This work is specifically focusing on multi-site hospital services for elderly and frail patients, using classification and regression trees to determine patient clusters with similar attributes, yielding results of up to 89.62% accuracy. By incorporating these results into mathematical models we aim to quantify the value of incorporating the clustering results in a deterministic and stochastic mathematical programme.