Current Projects

Optimal Policies for Arteriovenous Fistula Creation for Patients on Hemodialysis (Skandari, Shechter, Zalunardo)

To deliver hermodialysis (HD), patients need to have a vascular access. The gold standard for delivering HD is via arteriovenous fistula (AVF). Due to AVF creation failure, limited lifetime of a functional AVF, and limited places to create an AVF, patients may use an inferior access substitute called central venous catheter (CVC). We aim to characterize conditions under which the patient should or should not be referred for AVF creation, under total lifetime and quality adjusted lifetime metrics. 

Design of cadaver kidney allocation policy (Ding)

Since 2008, the Canadian federal and provincial governments have been actively looking for recommendations for an integrated organ and tissue donation and transplantation (OTDT) systemt o improve organ transplantations in Canada. A key issue is the design of a nationally coordinated policy for allocating cadaver organs, such as kidneys. The policy designer must carefully balance two possibly conflicting objectives; i.e., optimizing transplant outcomes, and providing equitable access to patients on the organ waitlist. The challenge is to have a concrete policy that achieves the desired tradeoff queuing systems by using a scoring-based ranking mechanism. In particular, I aim to address the primary challenge, mentioned above, by identifying the connection between a scoring formula and the allocation outcome. Specifically, when a social planner specifies a performance target in terms of efficiency and fairness, my analytical model will determine whether or not the target can be realized. Moreover, my model specifies the scoring formula to be implemented. This is joint work with Dr. Ata, Northwestern University, and Dr. Zenios, Stanford University.

Prediction of post-transplant survival probability (Ding)

Abstract: We introduced a non-parametric survival model for censored data, and applied our method to the setting of kidney transplantation to generate patient-specific survival distributions at the time of kidney transplantation. We tested performance and found that our method outperforms the standard Cox model when the proportional hazard assumptions is violated. This is a joint work with Dr. Lowsky, RAND Corporation.

Appointment scheduling for primary care clinics with follow-up visits (Ding)

A clinic had consulted us about efficient methods for scheduling their doctor appointments. After analyzing their data, we found that more than 40% of the visits to the clinic were follow-up visits, which have to be accommodated by the same doctor as in the initial visits. We formulated a multi-period news-vendor model to efficiently accommodate the follow-up visits, and developed practical scheduling policy. This is joint work with Dr. Gupta, University of Minnesota, and IBM employees Dr. Chen and Dr. Sharma. 

Allocation of operating room time between surgical specialties (Carew, Nagarajan, Shechter, Arneja, Skarsgard)

Excessive waiting for surgery can adversely impact patient health outcomes and consume additional system resources. As a result, hospitals face considerable pressure to appropriately manage limited operating room resources. In this study, we examine how to allocate operating room hours to different surgical specialties at the British Columbia Children's Hospital. This is a longer-run planning decision (made every six months), which has major effects on the wait time experience of the patient population. While there has been significant research on appointment scheduling, to our knowledge there has been little consideration of this upstream decision. To evaluate policies, we construct and validate a simulation model of the patient arrival and appointment processes that occur between six month decision periods. We then apply optimization and dynamic programming techniques to recommend improved allocation policies.