Informs Annual Meeting 2017

WB14

INFORMS Houston – 2017

WB14

chain efficiency. The presentation will focus on technologies to help manage forecasting inaccuracies, to improve responsiveness, and also find opportunities to improve the forecast itself. The presentation will also cover some of the opportunities and challenges for further improvements and hopefully encourage a dialog on research directions. 2 - Automation of Strategy with Respect to Supply Chain using Neural Networks Kumail Selani, Arizona State University, Mesa, AZ, United States, kselani@asu.edu We will discuss and demonstrate a working model that exemplifies how automation of strategy with respect to supply chain can capitalize on the available data. The proposed technique makes use of neural networks embedded in an IDEF0 environment. It is founded on the IDEF0 strategy model as presented in the paper “Automation of strategy using IDEF0 — A proof of concept” (Operations Research Perspectives (ORP) Dec. 2015 by Waissi et.al.). We will look at the unique advantage of neural networks in their ability to map variable data points to show trends and relationships, their high flexibility, adaptability and explore how supply chain strategy can be automated using this novel machine learning approach. 3 - A Three-dimensional Modelling Approach for Short-term Demand Forecasting Jay Patel, Arizona State University, Tempe, AZ, United States, jdpatel2@asu.edu Forecasting methods are either quantitative or qualitative, or combination of both. Any practical forecasting model is a combination of both. A novel approach is being introduced in this paper. We will construct a three-dimensional model for short-term demand forecasting by treating the supply chain as a “hybrid three- dimensional ecosystem”. The first two dimension include quantitative and qualitative techniques, and the third dimension is a weighting matrix to combine the two dimensions into one model. We will first analyze supply chain forecasting models using the first two dimensions, and then develop the weighting matrix. Vishwakant Malladi, Doctoral Student, University of Texas-Austin, McCombs School of Business, CBA 5.202, IROM, Austin, TX, 78712, United States, vishwakant@gmail.com Classical facility location problems do not incorporate the possibility of correlated disruptions amongst facilities and often result in optimal solutions that are too conservative. We propose using subordinated Markov chains to model the probability of the correlated risk of disruptions which offers a more realistic disruptions scenarios. We also present a new methodology to solve the facility location problem with correlated disruptions. Optimal solutions considering correlated disruptions show improved performance over traditional disruption models Eylem Koca, Assistant Professor, Ozyegin University, Nisantasi Mah. Orman Sk. 34-36, Cekmekoy Campus, Istanbul, 34794, Turkey, eylem.koca@ozyegin.edu.tr We investigate the two crucial decisions when introducing new experience goods: 1) preannouncements, and 2) availability of return policies. In our model, the main role of preannouncements is that of making product information available in order to reduce the product fit uncertainty for the consumers; the preannouncement timing decides on the available information. Coupled with this, we study the possibility of the firm selling the products in two forms simultaneously - with and without a return policy - and we characterize the conditions for which doing so is optimal. 2 - When to Offer Free Extended Warranty? Customer Segmentation, Repairing Cost, and Product Reliability Wen-Hsing Ting, National Taiwan University, Taipei, Taiwan, d04724017@ntu.edu.tw, Ling-Chieh Kung We study a firm’s optimal design of base and extended warranty plans when product reliability is exogenous or endogenous. We consider customers with different needs of repairing a damaged product and find that the firm’s decision depends two critical factors, the number of frequent users and customers’ repairing cost without warranty production. When the reliability is exogenous and the number of frequent users is large, the firm should sell extended warranty when customer’s repairing cost is either high or low but should not if moderate. With endogenous reliability, free extended warranty should be offered if customers’ repairing cost is low. The warranty should be life-long under some conditions. We will then combine the dimensions into an integrated model. 4 - Facility Location Problem with Joint Disruptions using Subordinated Markov Chains WB16 332F Operations Management Contributed Session Chair: Jie Yang, Tianjin University, Tianjin, China, nemoyj1989@126.com 1 - New Experience Good Introduction Decisions

332D Applications in Healthcare Sponsored: Manufacturing & Service Oper Mgmt, Healthcare Operations Sponsored Session Chair: Sandeep Rath, Sandeep_Rath@kenan-flagler.unc.edu 1 - IMPT Treatment Planning Optimization Considering RBE and LET Gino J.Lim, University of Houston, Houston, TX, United States, ginolim@uh.edu, Guven Kaya, Jiming Peng We present optimization models and algorithms to obtain optimal beamlet intensities considering relative biological effectiveness (RBE) in intensity modulated radiation therapy. Based on three RBE biophysical models, quadratic programming optimization models (RBEMs) are developed with the use of linear energy transfer (LET) values. We implement a functional parallelism approach to achieve substantial speed gains. We also compare results of RBEMs with the traditional dose-based fluence map optimization model. Two cancer cases; prostate and head-and-neck (including LET data) are tested. 2 - Learning from Many: Partner Exposure and Team Familiarity in Fluid Teams Jonas Oddur Jonasson, MIT.Sloan School of Management, 30 Memorial Drive, E62-588, Cambridge, MA, 02142, United States, joj@mit.edu, Zeynep O.Aksin Karaesmen, Sarang Deo, Kamalini Ramdas We use data from London Ambulance Service to study the impact of partner exposure of new paramedics on their operational performance. We find that the greater diversity in prior partners directly improves performance for an unstandardized process. For a more standardized process, this effect is moderated by a new recruit’s total experience. We explore the implications of our results for team formation strategies by balancing the benefits of partner diversity with those of team familiarity. 3 - Modeling Emotional Health and Quality of Therapeutic Alliance using Data from Chat Conversations Shishir Dash, Ginger.io, SanFrancisco, CA, United States, shishir.dash@gmail.com, Sai Moturu Each year, one in five people suffer from stress, anxiety or depression. Apps providing chat-based talk therapy, video sessions with professionals and self-care tools are crucial to addressing the huge access and cost hurdles in this space. They also enable the continuous modeling of the therapeutic alliance and patient progress. We present initial analysis of a large trove of anonymized conversation data collected via the Ginger.io app to (a) identify indicators of depression and anxiety, (b) triage users to the right level of care, and (c) identify successful talk- therapeutic strategies. We aim to show how these technologies can help build a more scalable and efficient mental health care system. 4 - Kidney Transplant Wait List Management: Patient Assessment and Prioritization Joseph Milner, University of Toronto, Toronto, ON, Canada, milner@rotman.utoronto.ca, Adam Diamant We consider the admission control problem for the kidney transplant wait list for a major research and teaching hospital located in Toronto, Ont. With increasing demand and limited diagnostic capacity, the transplant program seeks to ensure sufficient patients of each type are maintained on the list. Uncertain supply leads to significant delays in obtaining an organ. Because patient states are dynamic, there are significant costs to being maintained on the list. We develop an evidence-based policy to lower patient costs while maintaining hospital service levels. 332E Forecasting within a Supply Chain Sponsored: Manufacturing & Service Oper Mgmt, Supply Chain Sponsored Session Chair: Gary R Waissi, PhD, Arizona State University, 6049 S Backus Mall, Sutton Hall #340J, 6049 S Backus Mall, Sutton Hall #340J, Mesa, AZ, 85212, United States, gary.waissi@asu.edu 1 - Tactical Demand Forecasting from an Industry Perspective – Opportunities and Challenges Divya Prakash, Intel, 5000 W. Chandler Blvd, Chandler, AZ, 85226, United States, divya.prakash@intel.com, Nital S. Patel Changing business landscape is driving changes in supply chain design resulting in the need to be more responsive to market needs. Given the manufacturing lead times along with the changing demand landscape it is crucial to closely examine the horizon and fidelity of tactical forecasting for purposes of optimizing supply WB15

478