Surgery scheduling is a complex administrative problem for hospital managers. New research led by USC Viterbi aims to streamline scheduling. Image/Pixabay.
If you’ve ever undergone elective surgery, you know the feeling of uncertainty. The anxious waiting and the phone calls with your care team and insurance are often compounded by a healthcare system struggling with logistical hurdles. What if an operating room is not free at your scheduled time? What if your procedure takes longer than anticipated? Will there be a bed available for you to recover after surgery? These complex variables often lead to frustrating, last-minute delays or cancellations. Hospital administrators must constantly try to avoid a system-wide gridlock caused by this high-stakes scheduling puzzle.
A new study led by Karmel Shehadeh, WiSE Gabilan Assistant Professor in the Daniel J. Epstein Department of Industrial and Systems Engineering at USC Viterbi School of Engineering, offers a powerful solution. The research team has created an integrated framework for planning surgeries to help hospitals untangle this gridlock, leading to better outcomes for patients and providers, fewer delays and idle time and reduced costs.
The research, published in the European Journal of Operational Research, offers an approach to solve the elective surgery assignment, sequencing, and scheduling problem (ESASSP).
“Prior studies have tackled isolated components of the ESASSP,” said Shehadeh. “Ours introduces the first models that account for uncertainties and ambiguities in surgery durations and post-operative lengths of stay in recovery units, while also addressing the challenges of optimizing surgery schedules under the limited capacities of the ICU and other wards.”
Shehadeh said that the key challenge for hospital managers is that scheduling an elective surgery sets off a complex domino effect. It requires coordinating not just an operating room (OR), but also a bed in the downstream recovery units (e.g., ICU and/or surgical ward). These resources are expensive and have tight capacity. The problem is magnified by the unpredictable nature of medicine, where a surgery’s duration or a patient’s recovery time can vary significantly, even for the same procedure.
Karmel Shehadeh, WiSE Gabilan Assistant Professor in the Daniel J. Epstein Department of Industrial and Systems Engineering
“Say you have an open-heart surgery planned to take three hours. But then you go in, and it takes longer, which delays the next surgery, delays the next physician, and so on,” Shehadeh said. “Or you could finish much earlier than expected–well before the subsequent surgery is scheduled—leaving the operating room, a very expensive resource, sitting idle.”
Shehadeh added that surgery recovery can also look vastly different depending on the patient’s characteristics, with some requiring longer ICU stays, others recovering first in the ICU and then in the ward, or requiring follow-up stays in the ICU.
“If the ICU is congested, then you either prematurely discharge someone to free a bed, or you could cancel a surgery. Both can cause negative healthcare outcomes. It’s such a complex problem for hospital managers,” Shehadeh said.
Shehadeh and her collaborators from Carnegie Mellon University, Texas Tech University, and the Medical University of South Carolina worked closely with a hospital, harnessing surgery data and case studies, to develop sophisticated mathematical models that can plan all these surgery scheduling steps simultaneously. Their models use a distributionally robust optimization (DRO) approach, which is particularly powerful because it doesn’t assume a perfect forecast. Instead, the models hedge against a range of possible distributions of surgery duration and post-operative length-of-stay, finding schedules that are robust even when things don’t go exactly as planned.
“It’s a mathematical model that can make all of these decisions, like how many surgeries to schedule, when to schedule them, and to which operating room, taking into consideration the capacity of the OR, the availability of the ICU and the ward, and all the variabilities that happen around surgery duration and length-of-stay,” Shehadeh said.
The real-world impact demonstrated by the model is potentially profound. By implementing these integrated models, hospitals could reduce total operational costs by between 24% and 60%.
“Our findings offer valuable insights into the ESASSP and demonstrate the practical impact of our integrated approaches,” said co-author Rema Padman, Trustees Professor of Management Science and Healthcare Informatics at Carnegie Mellon University’s Heinz College.
For patients, the benefits could be even more tangible. The study’s models create more realistic and reliable schedules, significantly reducing the likelihood of last-minute surgery cancellations due to a lack of recovery beds. The models also mitigate long delays in the OR, a common source of stress for patients and their families. Moreover, the integrated approach has the potential to enhance access to surgery.
The research also reveals a critical trade-off between maximizing the number of surgeries and maintaining smooth operational performance. While scheduling more procedures seems like a win for patient access, it can increase the risk of overwhelming recovery units, ultimately leading to more disruptions. The team’s models provide hospital administrators with the data to find the right balance for their specific resources.
While the models provide a powerful new guideline, the authors noted that the next step is to refine the models further and develop user-friendly software tools that can be implemented in hospitals.
“Implementation is a big challenge, especially in healthcare, where there are a lot of moving parts and many decision makers. So, our next step is to continue working with our collaborators and other health systems to try to make this more accessible for them and translate it into a decision support tool where they can input their data and get a schedule,” Shehadeh said.
Published on September 11th, 2025
Last updated on September 11th, 2025
