2015 Informs Annual Meeting

WE35

INFORMS Philadelphia – 2015

WE32 32-Room 409, Marriott Data Mining in Manufacturing Contributed Session

WE33 33-Room 410, Marriott Scheduling in Residency Rotational Medical Programs Sponsor: Health Applications Sponsored Session Chair: Ruben Proano, Associate Professor, Rochester Institute of Technology, 81 Lomb Memorial Drive, Rochester, NY, 14623, United States of America, rpmeie@rit.edu 1 - Constructing Annual Block Schedules for Family Medicine Residents Jonathan Bard, Professor, University of Texas, 204 E. Dean Keeton St., C2200, Austin, TX, 78712, United States of America, jbard@mail.utexas.edu, Zhichao Shu, Douglas Morrice, Ramin Poursani, Luci Leykum This paper presents a mathematical model that can be used to help construct annual block schedules for Family Medicine programs. Starting with the current rotation templates our goal was to redesign them to concurrently (i) balance the number of outpatients that can be seen in the clinic during each half-day session, and (ii) minimize the changes to the current template assignments. Computations are illustrated with data provided by the University of Texas School of Medicine in San Antonio. 2 - Decision Support for Duty- and Workstation Rostering Subject to Fairness and Preferences Jens Brunner, University of Augsburg, Universitätsstrafle 16, Universität Augsburg, WiWi, Augsburg, De, 86159, Germany, jens.brunner@unikat.uni-augsburg.de, Andreas Fögener We consider a practical physician scheduling problem at hospitals. We formulate two mixed integer linear programming models for duty- and workstation assignments. The duty-roster model assigns physicians to 24h- and late-duties whereas the workstations-roster model assigns physicians to actual workstations for each regular working day. To promote for job satisfaction we take into account fairness and preferences. We present the status of the software development and an evaluation of our models. 3 - A Multi-objective Optimization Approach to the Resident Rotation Scheduling Problem Akshit Agarwal, Rochester Institute of Technology, 81 Lomb Memorial Drive, Rochester, NY, 14623, United States of America, aa9425@rit.edu, Ruben Proano Resident rotation scheduling is a complex process, which makes it challenging for schedulers to develop yearlong rotation schedules that not only adhere to the various managerial and legal restrictions but also balance resident workload and the quality of their education experience. We propose a multi-objective mixed integer programming approach and evaluate its performance on a multi-criteria scorecard based on AHP. 4 - Models and Approaches for Residents’ Annual Rotation Schedules William Pozehl, Research Specialist, University of Michigan, 1075 Beal Avenue, Suite 3246, Ann Arbor, MI, 48109, United States of America, pozewil@umich.edu, Amy Cohn, Ed O’brien Building annual resident rotation schedules requires not only meeting resident education and service coverage requirements, but also fulfilling a variety of requests to improve training quality and resident satisfaction. Priorities and constraint complexity vary by program, but the underlying models and scheduling approaches share fundamental similarities. We present our experiences in building pediatric and surgical rotation schedules. WE35 35-Room 412, Marriott Public Health Contributed Session Chair: Gina Dumkrieger, ASU, 1151 S. Forest Avenue, Tempe, AZ, 85281, United States of America, gina.dumkrieger@asu.edu 1 - Rapid Vaccination Response to a Deadly Disease Outbreak Rebecca Scott, University of North Carolina Wilmington, 601 S. College Road, Wilmington, NC, 28403, United States of America, scottra@uncw.edu, Gayle Prybutok, Victor Prybutok Rapid vaccination response is developed for Ebola using a contextualized newsvendor model. A network model that allows evaluating the importance of decision making factors is both posited and tested. Implications are reported that provide insights for increasing the ability to respond in a populated urban area.

Chair: Weihong Guo, University of Michigan, 1205 Beal Avenue, Ann Arbor, MI, 48105, United States of America, graceguo@umich.edu 1 - Deterioration Monitoring of Ultrasonic Welding Tools Based on High-order Decomposition Yaser Zerehsaz, PhD Student, University of Michigan, Ann Arbor, MI, 48105, United States of America, yzereh@umich.edu, Chenhui Shao, Jionghua (Judy) Jin Ultrasonic welding has been used for joining lithium-ion battery cells in electric vehicles. The geometric change of tool surfaces severely affects the weld quality. A new monitoring scheme based on a high-order decomposition method is suggested for the purpose of detecting the tool wear. A classification approach is used to determine the wear level of tools, which is helpful in finding the assignable causes of the extraordinary tool wear. 2 - Predictive Analytics of Semiconductor Chip Quality under Data Imbalance Situations Jin Soo Park, Mr., Korea University, Anam-Dong, Seongbuk-Gu Korea University, Innovation Hall #816, Seoul, 136-713, Korea, Republic of, pjsoo78@gmail.com, Seoung Bum Kim The quality prediction of the semiconductor industry has been widely recognized as important and critical for quality improvement. The main objective of this paper is to predict the final quality of semiconductor chips by solving data imbalance problems. Our proposed method is generating more synthetic data near the decision boundary in order to specify the minority class region. We demonstrate the usefulness and applicability of the proposed method using real data from a semiconductor industry. 3 - Identifying Key Factors Affecting Knowledge Absorptive Capacity of Automobile Industry Knowledge absorptive capacity is essential for the R&D of automobile firms. Thus, identifying key factors affecting this capacity is important which, however, is ignored by much research. This paper firstly builds a two-level factor framework which contains four dimensions in the first level and 16 factors in the second level. Interval valued intuitionistic fuzzy and DEMATEL model are then constructed to identify key factors based on expert interview. Finally, some suggestions are proposed. 4 - Decision Support Information System Based on Rough Set to Telecom Churn Management Mihrimah Özmen, University of Erciyes, Muhendislik Fakultesi In this study a rough set based heuristic algorithm is developed enabling data selection, filtering and aggregation. And then, in order to identify the customers with churning tendency, a relatively novel and popular algorithm is proposed. This cost-based particle swarm optimization algorithm is particularly efficient in analyzing high dimensional data aggregate stacks with high accuracy. 5 - Multi-stream Profile Monitoring and Fault Diagnosis via Sensor Fusion Weihong Guo, University of Michigan, 1205 Beal Avenue, Ann Arbor, MI, 48105, United States of America, graceguo@umich.edu, Jionghua (judy) Jin, S. Jack Hu When multiple signals are acquired from different sources, sensor fusion and data dimension reduction are two major issues to achieve a better comprehension of the process. A method for analyzing multi-stream profiles based on uncorrelated multilinear discriminant analysis and ensemble learning is proposed in this research for the purpose of profile monitoring, fault detection, and fault diagnosis. The proposed method is compared with other solutions with both simulated and real data. Guohua Qiu, School of Economics and Management, Tongji Univesity, No.1239 Siping Road, Shanghai, China, 2298900836@qq.com, Yixi Xue, Jianxin You Endustri Muh Bolum, Kayseri, 38039, Turkey, mihrimah@erciyes.edu.tr, Emel Kizilkaya Aydogan

489

Made with