2016 INFORMS Annual Meeting Program

413 / 561

INFORMS Nashville – 2016
413
WB52
214-MCC
New Advances in Operating Room (OR) Scheduling
Sponsored: Public Sector OR
Sponsored Session
Chair: Gino J Lim, University of Houston, E206 Engineering Building 2,
Houston, TX, 77204, United States,
1 - Integrated Anesthesiologist And Room Scheduling For Surgeries
Sandeep Rath, UCLA Anderson School of Management,
, Kumar Rajaram
At large hospitals the assignment and scheduling of anesthesiologists and
operating rooms is a complex resource allocation decision undertaken daily by the
managers of operating room suites. We validate and implement a data-driven
decision support system at the UCLA Ronald Reagan Medical Center. We also
conduct analyses related to capacity expansion and process improvement efforts.
2 - A Discrete Event Simulation Evaluation Of Distributed Operating
Room Scheduling
Vahid Roshanaei, PhD Candidate, University of Toronto,
5 King’s College Road, Toronto, ON, M5S3G8, Canada,
, Shuo Wang, Dionne Aleman,
David Urbach
We use discrete event simulation to assess the performance of deterministically
optimized OR schedules in a network of collaborating hospitals with shared
resources, called distributed OR scheduling (DORS), in the face of uncertain
surgical durations, emergency arrivals, and limited downstream resources. We
quantify the individual and combined disruptive impact of these stochastic factors
on the DORS schedule, using real data obtained from the University Health
Network (UHN) in Toronto, Canada. We show that the schedule constructed by
DORS results in higher OR utilization and lower average surgery cost compared to
UHN’s current schedule.
3 - Scheduling Of Multi-priority Patients With Preference,
Cancellation, No Show And Capacity Uncertainty
Deepak Agrawal, Penn State University, ,
Guodong Pang, Priyantha Devapriya, Soundar Kumara
Reasons of No-shows and how to stop them, has been focus of the research since
more than a decade. No-show adds up to the increases healthcare waste.
Therefore, motivated by this we aim to develop advanced scheduling models
which can reduce No-shows by scheduling appointments at patients’ preferred
day and time with their preferred physicians while simultaneously maximizing
the profit for clinics. Patient choices are modeled as mixed-logit model. The
dynamic scheduling process is modeled as a Markov decision process. We conduct
numerical experiments to test the performance of the dynamic model in
comparison to several heuristics proposed in the literature.
4 - A Lagrangian Relaxation Algorithm For Solving Surgery
Scheduling Problem Under Uncertain Durations
Amirhossein Najjarbashi, University of Houston,
, Gino J Lim
We studied a surgery scheduling and resource allocation problem by considering
uncertain durations. A mixed integer programming formulation is developed in
order to maximize throughput and minimize overtime. A Lagrangian Relaxation
algorithm is applied to solve large-scale problems efficiently. Having a risk-averse
attitude, we applied Conditional Value-at-Risk to tackle uncertainty.
WB53
Music Row 1- Omni
Opt, Stochastic VI
Contributed Session
Chair: Tugce Isik, Clemson University, 277D Freeman Hall,
Dept. of Industrial Engineering, Clemson, SC, 29634, United States,
1 - Valuing A Portfolio Of Systemic Urban Infrastructure Investments
Using Approximate Dynamic Programming With Decision
Dependent Uncertainties
Sebastian Maier, Imperial College London, Imperial College Road,
Skempton Building, London, SW7 2AZ, United Kingdom,
, John W Polak, David M Gann
We present a new portfolio-based framework for the application of approximate
dynamic programming to the valuation and risk-management of urban
infrastructure investments with decision dependent uncertainties. We use this
framework to formulate a multistage stochastic optimisation model in which the
value function is approximated by linear regression using both simulation and the
modelling of decision dependent uncertainties. Using the real-world case of
district heating network investments in London, we investigate the effects of the
consideration of decision dependent uncertainties on both the optimal portfolio
value and the underlying optimal strategic and operational decisions.
2 - A Stochastic Program For Debris Collection Problem
Derya Ipek Eroglu, Research and Teaching Assistant, Middle East
Technical University, Dumlupinar Street N:1, Ankara, 06400,
Turkey, , Duygu Pamukcu
Our aim is to develop a stochastic program for debris collection problem, which
enables us to make decisions regarding different disaster types or different
scenarios that can take place for a disaster type. Debris collection is important for
human health since uncollected debris may lead to pollution. We develop
stochastic programs using appropriate decomposition methods which will be
compared in terms of computational efficiency, test the program with different
variants, compare the model with pre-developed deterministic model(Celik et al.)
and analyse related performance measures. We present results of our
computational studies and analysis.
3 - Management Of Scarce Blood Supplies Accounting For
Cross-matching Characteristics
Nooshin Valibeig, Northeastern University, 360 Huntington Ave,
Boston, MA, 02115, United States, ,
Jacqueline Griffin
For blood transfusion, availability of blood of a compatible type is crucial to
patient treatment and reductions in mortality rate. In isolated environments, such
as in the aftermath of a disaster or combat environment, demand for blood is
unpredictable and timing and quantity of stock replenishment is unreliable.
Correspondingly, the risk and the cost of blood shortages is high. We develop a
stochastic optimization model to develop threshold policies to prevent from blood
shortages considering proactive allocation, accounting for cross-matching criteria,
to satisfy the requests for various blood types. We assess the effectiveness of these
real-time allocation policies via simulation method.
4 - Multistage Power Generation Capacity Expansion Models With
Different Risk Measures
Shu Tu, lehigh university, 200 West Packer Avenue, Bethlehem,
PA, 18015, United States, , Boris Defourny
When it comes to the multistage problems, the stochastic programming models
rely on the convexity property of the problems and the solution approaches
usually rely on the stage independence assumption, with Markov decision
processes being advantageous from these aspects. Therefore, we adapts the “good-
deal” generation capacity expansion model to the form of Markov decision
processes and will implement it with C++ which can make use of parallel
computing and ILOG Concert Technology.
5 - Control Policies For Queueing Systems With Time Sensitive Jobs
Tugce Isik, Clemson University, 277D Freeman Hall, Dept. of
Industrial Engineering, Clemson, SC, 29634, United States,
Bahar Cavdar
We consider a queueing system where each job has a preferred time window for
service. We assume that all jobs must be served and the jobs are outsourced when
the capacity is insufficient. Costs are incurred for both outsourced jobs and early
service. For systems with short time windows and a single class of jobs, we show
that a class of threshold service policies are optimal. For general systems, we
devise heuristic policies based on similar threshold structures.
WB54
Music Row 2- Omni
Service Science: Uncertainty in Business Processes
Sponsored: Service Science
Sponsored Session
Chair: Genady Grabarnik, St. John’s University, St. John’s University,
Queens, NY, United States,
1 - Continuity Of The Lyapunov Exponenets Under Continuity Of
Measures In Sl (2, R)
Genady Grabarnik, St. John’s University,
In the business processes composition of processes results in special type of
product of the appropriate distributions. In this case it corresponds to product of
random matrices. It is well know that behavior of random matrices controlled by
its Lyapunov exponents. We are investigating stability of the top Lyapunov
exponent under perturbation of defining measure in weak topology.
WB54