Decision support system for appointment scheduling and overbooking under patient no-show behavior

• Published in: Annals of operations research Vol. 342; no. 1; pp. 845 - 873
• Main Authors: Topuz, Kazim, Urban, Timothy L., Russell, Robert A., Yildirim, Mehmet B.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.11.2024; Springer Nature B.V
• Subjects: Analytics; Business and Management; Combinatorics; Computer aided scheduling; Data analysis; Decision support systems; Machine learning; Operations Research/Decision Theory; Original Research; Scheduling; Sensitivity analysis; Theory of Computation; Decision support systems; Appointment scheduling; Overbooking; Predictive analytics; Probabilistic graphical modeling
• ISSN: 0254-5330; 1572-9338
• DOI: 10.1007/s10479-023-05799-0

Data availability enables clinics to use predictive analytics to improve appointment scheduling and overbooking decisions based on the predicted likelihood of patients missing their appointment (no-shows). Analyzing data using machine learning can uncover hidden patterns and provide valuable business insights to devise new business models to better meet consumers’ needs and seek a competitive advantage in healthcare. The innovative application of machine learning and analytics can significantly increase the operational efficiency of online scheduling. This study offers an intelligent, yet explainable, analytics framework in scheduling systems for primary-care clinics considering individual patients’ no-show rates that may vary for each appointment day and time while generating appointment and overbooking decisions. We use the predicted individual no-show rates in two ways: (1) a probability-based greedy approach to schedule patients in time slots with the lowest no-show likelihood, and (2) marginal analysis to identify the number of overbookings based on the no-show probabilities of the regularly-scheduled patients. We find that the summary measures of profit and cost are considerably improved with the proposed scheduling approach as well as an increase in the number of patients served due to a substantial decrease in the no-show rate. Sensitivity analysis confirms the effectiveness of the proposed dynamic scheduling framework even further.