Transient performance & availability modeling in high volume outpatient clinics

High volume outpatient clinics such as eye care centers cannot afford excessive delays, especially when due to limited resources, time, or overhead. Modeling tools from reliability & maintainability practice may provide the means to better assess where improvements may be possible. The models developed in this demonstration build on previous efforts to address challenges in quantifiable performance evaluation in health care. Discrete event simulation solutions of non-Markovian Stochastic Reward Nets provide insight into operations from data collected at an academic medical center. The produced model of patient flow at a glaucoma practice reflects clinic process flow diagrams and electronic health record data collected from January to March 2016. In this report we describe the parameterization process, as well as how the collected data and direct observation of a clinic influenced model formation. We then employ maximum likelihood estimation to fit the distribution of times spent at different treatment phases to parametric models, distinguishing between groups in data where appropriate. Combining data fitting in a Stochastic Reward Net, we solve the resulting non-Markovian model using discrete event simulation. The model is then simulated to predict transient high patient load periods experienced by the clinic. These interim results indicate that for the moderately sized clinic presented here, patient volume is manageable even with minimal resource support. Finally, future directions in clinic model validation and operations optimization are presented.