2015 Informs Annual Meeting

MA43

INFORMS Philadelphia – 2015

2 - Dynamic Assignment of Emergency Department Patients to Primary and Secondary Inpatient Units Derya Kilinc, Arizona State University, 699 S. Mill Ave., Tempe, AZ, 85281, United States of America, dkilinc@asu.edu, Soroush Saghafian, Stephen J. Traub One of the main reasons for Emergency Department (ED) crowding is the long boarding time, of patients who are waiting for admission to inpatient wards. We study suitable mechanisms to overflow such patients to alternative wards. An overflow policy can improve ED waiting times and Length of Stay (LOS), but may reduce the quality of care. We study an MDP-based approach to gain insights into the impact of overflow policies on waiting times and quality of care. 3 - Data Driven Staffing of Hospital Support Staff Cassandra Hall, PhD Student, Northwestern University, IEMS Department, Evanston, IL, CassandraHall2017@u.northwestern.edu, Sanjay Mehrotra, Seyed Iravani We study hospital patient transport requests as a multiclass queue with server specialization and construct an approximation for the minimum number of servers required to achieve a probabilistic bound on the patient waiting time. We also explore the effects of different specialization and routing policies on performance at a given staffing level. Output is then compared with theoretical worst case waiting time bounds derived from the literature. 4 - A Clinical Decision Support System for Treatment-resistant Depression: A Pilot Study Martin Cousineau, Desaultels Faculty of Management, McGill University, 1001 Sherbrooke Street West, Montreal, QC, H3A 1G5, Canada, martin.cousineau@mail.mcgill.ca, Gustavo Turecki, Vedat Verter, Joelle Pineau This research project aims to develop a clinical decision support system to assist psychiatrists seeking to achieve remission in treatment-resistant depression patients. This system is based on a longitudinal dataset of outpatient mental health clinic patients, and consists of (1) a predictive model of the patient outcome depending on the selected treatment, and (2) a decision-theoretic module for recommending treatment strategies based on similar patient files. 5 - An Empirical Study of Patient Prioritization in Emergency Department Triage Systems We analyze patient choice behavior of the ED personnel who decides which patient waiting in the ED will be seen by the next available physician. We use a discrete choice framework consistent with random utility theory. The choice maker’s valuation of each patient depends on both the patient’s medical and operational characteristics including wait time and ED congestion. We study over 270,000 patient choices in five EDs using the Canadian Triage and Acuity Scale (CTAS). MA42 42-Room 102B, CC Modeling Healthcare Provider and Processes Interaction Sponsor: Manufacturing & Service Oper Mgmt/Healthcare Operations Sponsored Session Chair: Vikram Tiwari, Vanderbilt University Medical Center, Nashville, TN, United States of America, vikram.tiwari@vanderbilt.edu 1 - Quantifying Case Cancellations and Add-On Case Scheduling among Patients Inpatient Preoperatively Among 24,735 inpatient scheduled cases, 22.6 ±0.5% (SE) of scheduled minutes cancelled after 7 AM the day before or day of surgery (14.0% ±0.3% day of). Most (83.1% ±0.6%) cases performed were evaluated the day before surgery, 62.3% ±1.5% before 6:00 PM. The cancelled procedures were very diverse (Herfindahl index 0.021 ±0.001). When cancelled, often no procedure was subsequently performed (50.6% ±0.9%), showing the surgical indication no longer existed or patient/family decided no surgery. 2 - The Role of Wearable Devices Data in Physician-patient Relationship Zafar Özdemir, Professor, Miami University, Farmer School of Business, Oxford, OH, 45056, United States of America, ozdemir@miamioh.edu, Shailesh Kulkarni, Hakan Tarakci We investigate a medical provider’s optimal level of investment in remote monitoring where, unlike the current norm, the provider can initiate a face-to- Mahesh Nagarajan, University of British Columbia, 2053 Main Mall, Vancouver, BC, V6T1Z2, Canada, mahesh.nagarajan@sauder.ubc.ca, Eric Park, Yichuan Ding Franklin Dexter, Professor, University of Iowa, Anesthesia, 200 Ha, Iowa City, IA, 52242, United States of America, franklin-dexter@uiowa.edu, Richard Epstein

face visit or a remote treatment depending on the information streamed from the patient’s wearable device. Our model provides valuable insights across a variety of payment models. 3 - Evaluating Peer-to-peer Performance of Anesthesiology Fellows using Data Envelopment Analysis Vikram Tiwari, Vanderbilt University Medical Center, Nashville, TN, United States of America, vikram.tiwari@vanderbilt.edu, Avinash Kumar Factors that contribute to the success of individuals at a critical care fellowship have not been well studied. We explore what aspects of the educational program and work characteristics contribute the most to an individual fellow’s success as determined by year end Multidisciplinary Critical Care Knowledge Assessment Program scores and summative evaluations. We show the feasibility of using data envelopment analysis to evaluate the academic performance of fellows compared to their peers. MA43 43-Room 103A, CC Game Theoretic Models in Revenue Management I Sponsor: Revenue Management and Pricing Sponsored Session Chair: Ozan Candogan, University of Chicago, Booth School of Business, Chicago, IL, United States of America, ozan.candogan@chicagobooth.edu Co-Chair: Santiago Balseiro, Assistant Professor, Duke University, 100 Fuqua Drive, Durham, NC, 27708, United States of America, srb43@duke.edu 1 - Customer Referral Incentives and Social Media Ilan Lobel, NYU, 44 W 4th St, New York, NY, 10012, United States of America, ilobel@stern.nyu.edu, Evan Sadler, Lav Varshney We study how to optimally attract new customers using a referral program. Whenever a consumer makes a purchase, the firm gives her a link to share with friends, and every purchase coming through that link generates a referral payment. The firm chooses the referral payment function and consumers play an equilibrium in response. We show that the optimal payment function is nonlinear and complex, and prove revenue properties of simple approximate solutions such as linear and threshold policies. 2 - Dynamic Reserve Prices for Repeated Auctions: Learning from Bids Yash Kanoria, Assistant Professor, Columbia University, New York, NY, United States of America, ykanoria@columbia.edu, Hamid Nazerzadeh A large fraction of online advertisements are sold via repeated second price auctions. In these auctions, the reserve price is the main tool for the auctioneer to boost revenues. We present a simple approximately incentive-compatible and optimal dynamic reserve mechanism that can significantly improve the revenue over the best static reserve when there is uncertainty in the distribution of valuations. 3 - A Dynamic Model of Crowdfunding Mohamed Mostagir, Assistant Professor, University of Michigan Ross School of Business, 701 Tappan Ave, R5316, Ann Arbor, MI, 48109, United States of America, mosta@umich.edu, Saeed Alaei, Azarakhsh Malekian Crowdfunding has emerged as an alternative to traditional methods of funding new products. Backers arrive over time and decide whether to pledge money to a crowdfunding campaign. If the total contribution reaches a certain threshold, the campaign is successful and production takes place. We identify a fundamental tension in these environments that leads to a sharp characterization of empirical outcomes, and we show how to determine the optimal duration and price to maximize campaign success. 4 - Effect of Network Perturbation on Aggregate Performance Azarakhsh Malekian, Rotman School of Management, University of Toronto, Toronto, Canada, azarakhsh.malekian@rotman.utoronto.ca, Opher Baron, Ming Hu In this work, we characterize the role of perturbing network interactions in macroeconomic aggregate performance. We then provide a fairly tight characterization of the aggregate performance under worst-case network perturbation as well as average-case random network perturbation. Finally, we identify robust networks under perturbation.

159