Social Network based sensitivity analysis for patient flow using computer simulation

Prediction of patient flow, an essential element of any healthcare system, is challenging due to uncertainties in patient volume. One such source of uncertainty is a propagated outbreak of epidemic diseases as they spread through networks of human populations. Till date, no study exists that studied the effect of social network structure on patient flow. In this study, we aim to examine the relationship between patient flow in a social network and the corresponding network characteristics. For this purpose, we developed a simulation model in which an epidemic spreads through a social network and then the generated patients are directed to a healthcare system. To quantify the patient flow, we considered the conditional expected value of the length of stay (LoS) as our performance measure. The network characteristics considered were average distance, closeness centralization, betweenness centralization, and eigenvector centralization. Results from this study indicate that the patient flow has a direct relationship with closeness centralization and an inverse relationship with average distance. Betweenness and eigenvector centralization did not provide any meaningful information in patient flow prediction. Overall, patient flow is more sensitive to average distance. This work helps healthcare planners and decision makers in better prediction of the patient flow during a propagated outbreak of an epidemic.