Hospital admission pattern analysis, bed resource requirements forecasting, allocation and management

Abstract

Healthcare professionals, resource planners and hospital staff need to ensure the optimum utilization and availability of the limited healthcare resources at an acceptable cost of care by developing the policies. The patients’ Length of Stay (LOS) determines resource requirements and the number of patients admitted daily. Many factors such as the covariates that represents the characteristics of a patient like; gender, age and district, or the changes in the weather, may affect these. The dataset used for this study includes all the patients admitted in a three-year period (2011-2013) at Mater Dei Hospital and a three-year (2011-2013) weather data from Free Meteo. By designing a new and innovative application that can be used for similar studies in the future, the data was grouped by the patient covariates gender, age, district, source of admissions and the weather data was grouped by the covariates such as maximum temperature, minimum temperature, average temperature and maximum variability in temperature. To acquire the results, this study uses Coxian phase-type distribution (C-PHD) to generate Phase-Type Survival (PTS) trees to analyse groups of patients through patient covariates along with weather covariates to verify which affect the admission and LOS distribution. From the results generated by the C-PHD the PTS trees were generated that work to minimizing the weighted-average information criterion (WIC). By starting from the root node, at each node a split that has the highest gain in WIC is selected and recursively portioned the node into child nodes to grow the tree. If at a node there is no improvement in the WIC, the node is said to be a terminal (leaf) node. Based on the patients’ characteristics and weather covariates, the PTS tree effectively clusters the patients based on their LOS bearing in mind the prognostic importance of these covariates. In relation to LOS the results were effective for covariates related to patient characteristics and weather. Similarly, based on the patients’ characteristics and weather covariates, the PTS tree clusters the patients based on admission rate bearing in mind the prognostic importance of these covariates. During this study some limitations arose during the construction of the admission tree.