A Framework for Multi-response Optimization of Healthcare Systems Using Discrete Event Simulation and Response Surface Methodology

A framework is proposed that combines the use of discrete event simulation (DES), analysis of variance technique along with response surface methodology (RSM) in multi-response optimization of healthcare systems. Previous DES studies have studied effects of multiple factors and their interactions on several responses in such systems, but none have optimized these responses simultaneously. In this research, we not only take the effect of multiple factors and their interactions on several responses into consideration but also the responses are also optimized simultaneously using DES and RSM. A case study is presented in an Endocrine-Diabetes clinic in which patients' lengths of stay, staff utilizations, and throughputs are optimized simultaneously using this methodology with respect to several factors; number of nurses, number of delegates, working hours, appointment levels, and patient flows. This represents the first effort in using such an approach in healthcare systems analysis. Desirability functions were used in RSM in order to optimize the three responses. Results reveal that appointment levels had the maximum effect and number of nurses had the minimal effect. The overall desirability value is 0.667 in which throughput improved 70% with this response being positively proportional to the number of working hours and appointment levels and their interaction. Lengths of stay decreased around 19% and were negatively proportional to all factors except appointment levels, and around 64% is the enhancement in delegate utilizations, that is positively proportional to all factors except for the number of nurses and delegates which negatively affected this response.

Automated summary

A framework combining discrete event simulation, analysis of variance, and response surface methodology is proposed for multi-response optimization of healthcare systems. A case study in an Endocrine-Diabetes clinic is presented, optimizing patients' lengths of stay, staff utilizations, and throughputs simultaneously. Results show that appointment levels have the maximum effect, while number of nurses has the minimal effect.