A MODEL-BASED DECISION SUPPORT TOOL FOR HOSPITAL QUEUING AND WORKFORCE ANALYSIS USING MONTE CARLO SIMULATION TECHNIQUE

Abstract

ABSTRACT

Government hospitals worldwide face rising patient numbers and have started looking into health informatics to improve their services. This dissertation reports the outcome of a project with one of the hospitals in Saudi Arabia on enhancing patients' waiting time by using existing resources more effectively. The scenario begins with the hospital planning to extend the RD's operational time from 8 to 24 hours a day. In this case, without a decision support tool, the capacity of the hospital's radiology department (RD) services could be described as evolved rather than planned. This can be attributed to at least two challenges: varying perspectives regarding patient demand, service time and limited decision support when planning and estimating capacity. Therefore, this project aims to develop a decision support tool that the management can use in making quality decisions. In dealing with these challenges, this study has identified several key performance indicators: patient waiting time, staff overtime, and demand for technician services. This study proposed three models for the decision support tool to assist hospital management when making decisions related to human resources. The first model is a predictive model based on the Support Vector Machine (SVM) to predict the service time. The second model is a prescriptive model for staff planning. The model is proposed to plan the staff schedule automatically. The prescriptive model is developed based on the hospital's human resource policy and queuing theory. The model is processed using the branch and bound algorithm. The output of those models is used in a simulation model. The developed simulation model is the foundation of the spreadsheet-based decision support tool (DST) for planning capacity. The Excel spreadsheet-based DST is the delivery mechanism in understanding the impact of a decision on the overall waiting time. The DST is intended to analyze the effect of altering specific features of the system on the overall waiting time. The proposed model incorporated queuing theory principles and extended the discrete event simulation to account for time-based arrival and service time rates. The prediction model is used to assign patients into rooms. The simulation results determined that the main challenge is inefficient patient management instead of insufficient resources. The number of staff and the simulation results show that it has accurately predicted service time. The strategy of allocating patients to a separate room has lowered the overall median wait time in the RD. The proposed modelling technique and spreadsheet-based DST are easy to distribute and help decision-makers analyze the impact of implementing a fast track or comparable system on patient waiting times.