1 August 2020 in Management Science
This paper studies how to schedule medical appointments with time-dependent patient no-show behavior and random service times. The problem is motivated by our studies of independent datasets from countries in two continents that unanimously identify a significant time-of-day effect on patient show-up probabilities. We deploy a distributionally robust model, which minimizes the worst-case total expected costs of patient waiting and service provider’s idling and overtime, by optimizing the scheduled arrival times of patients. This model is challenging because evaluating the total cost for a given schedule involves a linear program with uncertainties present in both the objective function and the right-hand side of the constraints. In addition, the ambiguity set considered contains discrete uncertainties and complementary functional relationships among these uncertainties (namely, patient no-shows and service durations). We show that when patient no-shows are exogenous (i.e., time-independent), the problem can be reformulated as a copositive program and then be approximated by semidefinite programs. When patient no-shows are endogenous on time (and hence on the schedule), the problem becomes a bilinear copositive program. We construct a set of dual prices to guide the search for a good schedule and use the technique iteratively to obtain a near-optimal solution. Our computational studies reveal a significant reduction in total expected cost by taking into account the time-of-day variation in patient show-up probabilities as opposed to ignoring it. This paper was accepted by David Simchi-Levi, optimization.
1 February 2020 in Management Science
Despite the prevalence and significance of walk-ins in healthcare, we know relatively little about how to plan and manage the daily operations of a healthcare facility that accepts both scheduled and walk-in patients. In this paper, we take a data-analytics approach and develop an optimization model to determine the optimal appointment schedule in the presence of potential walk-ins. Our model is the first known approach that can jointly handle general walk-in processes and heterogeneous, time-dependent no-show behaviors. We demonstrate that, with walk-ins, the optimal schedules are fundamentally different from those without. Our numerical study reveals that walk-ins introduce a new source of uncertainties to the system and cannot be viewed as a simple solution to compensate for patient no-shows. Scheduling, however, is an effective way to counter some of the negative impact from uncertain patient behaviors. Using data from practice, we predict a significant cost reduction (42%–73% on average) if the providers were to switch from current practice (which tends to overlook walk-ins in planning) to our proposed schedules. Although our work is motivated by healthcare, our models and insights can also be applied to general appointment-based services with walk-ins. This paper was accepted by Gad Allon, operations management.
1 July 2020 in INFORMS Journal on Computing
We consider assortment optimization problems, where the choice process of a customer takes place in multiple stages. There is a finite number of stages. In each stage, we offer an assortment of products that does not overlap with the assortments offered in the earlier stages. If the customer makes a purchase within the offered assortment, then the customer leaves the system with the purchase. Otherwise, the customer proceeds to the next stage, where we offer another assortment. If the customer reaches the end of the last stage without a purchase, then the customer leaves the system without a purchase. The choice of the customer in each stage is governed by a multinomial logit model. The goal is to find an assortment to offer in each stage to maximize the expected revenue obtained from a customer. For this assortment optimization problem, it turns out that the union of the optimal assortments to offer in each stage is nested by revenue in the sense that this union includes a certain number of products with the largest revenues. However, it is still difficult to figure out the stage in which a certain product should be offered. In particular, the problem of finding an assortment to offer in each stage to maximize the expected revenue obtained from a customer is NP hard. We give a fully polynomial time approximation scheme for the problem when the number of stages is fixed.
21 February 2019 in Production and Operations Management
1 September 2019 in Management Science
Motivated by the increasing use of online appointment booking platforms, we study how to offer appointment slots to customers to maximize the total number of slots booked. We develop two models, nonsequential offering and sequential offering, to capture different types of interactions between customers and the scheduling system. In these two models, the scheduler offers either a single set of appointment slots for the arriving customer to choose from or multiple sets in sequence, respectively. For the nonsequential model, we identify a static randomized policy, which is asymptotically optimal when the system demand and capacity increase simultaneously, and we further show that offering all available slots at all times has a constant factor of two performance guarantee. For the sequential model, we derive a closed form optimal policy for a large class of instances and develop a simple, effective heuristic for those instances without an explicit optimal policy. By comparing these two models, our study generates useful operational insights for improving the current appointment booking processes. In particular, our analysis reveals an interesting equivalence between the sequential offering model and the nonsequential offering model with perfect customer preference information. This equivalence allows us to apply sequential offering in a wide range of interactive scheduling contexts. Our extensive numerical study shows that sequential offering can significantly improve the slot fill rate (6%–8% on average and up to 18% in our testing cases) compared with nonsequential offering. Given the recent and ongoing growth of online and mobile appointment booking platforms, our research findings can be particularly useful to inform user interface design of these booking platforms. This paper was accepted by Gad Allon, operations management.
1 October 2023 in Management Science
Outpatient care providers usually allow patients to access service via scheduling appointments or direct walk-in. Patients choose strategically between these two access channels (and otherwise balking) based on the trade-off of appointment delay and in-clinic waiting. How to manage outpatient care with such dual access channels, taking into account patient strategic choice behavior, is a challenge faced by providers. We study three operational levers to address this management challenge: service capacity allocation between these two channels, appointment delay information revelation via the choice and design of online scheduling systems, and a walk-in triage system that restricts the use of walk-in hours only for acute care. By studying a stylized queueing model, we find that neither a real-time online scheduling system (which offers instant access to appointment delay information at time of booking) nor an asynchronous online system (which does not directly provide delay information) can be universally more efficient. Although real-time systems appear more popular in practice, asynchronous systems sometimes can result in higher operational efficiency. Under the provider’s optimal capacity allocation, which scheduling system is more efficient hinges on two key factors: the patient demand–provider capacity relationship and patient willingness to wait. For the walk-in triage system, we find that it may or may not be beneficial to adopt; the provider’s own cost trade-off between lost demand and overtime work is the key determinant. Our research highlights that there is no one-size-fits-all model for outpatient care management, and the best use of operational levers critically depends on the practice environment. This paper was accepted by Jayashankar Swaminathan, operations management. Funding: S. Wang’s work was supported in part by the National Natural Science Foundation of China [Grants 72001220 and 71931008]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/mnsc.2023.4676 .