Assessment of patient waiting time and queue length using simulation model and machine learning techniques

Purpose An accurate estimation of patient waiting time and queue length are essential for defining patient satisfaction, optimum resource allocation, healthcare response and acceptability of healthcare facilities. The conventional statistical prediction approach lacks precision, leading to patient and personnel difficulties. Methods To address this, a simulation model that optimises the different queuing configurations and machine learning approaches that facilitate the dynamic prediction of waiting times based on many criteria, including appointment scheduling, patient demographics, and facility workload, has been applied in this study to predict the wait time of patients and queue length before receiving the healthcare service. The TORA model has been applied to study the probabilistic nature of queue formation and associated waiting time. Linear Regression (LR), Random Forest (RF), and Support Vector Regression (SVR) were applied to model the correlation between significant factors and waiting or delay periods. The RF was applied to capture the intricate, nonlinear interactions, the SVR to record the high-dimensional environments, and LR to identify the most appropriate method for this application. Results The results suggested that machine learning techniques outperformed the simulation model. Among machine learning techniques, RF performed over SVR and LR to manage the intricacies of real-world data, although SVR offered significant insights owing to its interpretability. Conclusion Applying machine learning techniques to predict waiting time could improve patient satisfaction and experience and assist in optimising healthcare service provision. This study emphasises the significance of data-driven decision-making in minimising delays and enhancing patient flow in OPDs.