2015 Informs Annual Meeting

MD33

INFORMS Philadelphia – 2015

MD29 29-Room 406, Marriott

talk, we will present a novel MIP formulation for automating this complex decision, which minimizes the total cost of the wheel tru operation, thereby enhancing locomotive wheel life.

Joint Session Analytics/CPMS: 2015 Innovative Applications in Analytics Award Winner Reprise Sponsor: Analytics Sponsored Session Chair: Pooja Dewan, BNSF Railway, Fort Worth, TX, 76092, United States of America, Pooja.Dewan@bnsf.com 1 - Intelligent Surgical Scheduling System Kalyan Pasupathy, Associate Professor, Mayo Clinic, 200 First Street SW, HA 2-43, Rochester, MN, 55905, United States of America, Pasupathy.Kalyan@mayo.edu, Narges Hosseini, Jeanne Huddleston, Paul Huddleston, Yariv Marmor, Thomas Rohleder Orthopedic Surgery was facing highly fluctuating utilization of their operating rooms due to inaccurate estimation and scheduling of procedures. The existing scheduling optimization problems in literature were insufficient with just a single “optimal” solution. The team conducted descriptive research of clinical and operational factors, developed predictive models for surgical durations, and a prescriptive scheduling algorithm. Implementation results exhibit improvement in key metrics. MD30 30-Room 407, Marriott Practice Presentations by INFORMS Roundtable Companies III Sponsor: INFORMS Practice Sponsored Session Chair: Stefan Karisch, Digital Aviation Optimization & Value Strategy, Boeing Commercial Aviation Services, 55 Inverness Drive East, Englewood, CO, 80112, United States of America, stefan.karisch@jeppesen.com 1 - Xpress-mosel: New Modeling Features for Distributed and Cloud Computing A major concern when deploying optimization models in distributed computing environments are questions related to security for the transmission and storing of data, and the protection of the model itself - we show examples how these are adressed by the new Mosel module mmssl. We further discuss the new concept of model annotations, metadata that can be used to configure optimization applications, and touch on new interfaces (HTTP, XML, JSON, Hadoop, R). 2 - Solving the Airline Pilot Manpower Planning Problem. Per Sjügren, Jeppesen systems AB, Odinsgatan 9, Gothenburg, 41311, Sweden, per.sjogren@jeppesen.com The pilot manpower planning problem consists of the long term planning of recruitment and promotion to meet the forecasted crew need. Complicating factors are strict seniority promotion rules and limited training resources. We further consider movable activities such as vacation and overtime distribution as well as required recurrent training. We will present a high level description of the mixed integer model, the heuristic solution process and successful applications. 3 - Prescriptive Analytics on the Cloud with Python Vincent Beraudier, Architect And Program Manager, IBM ILOG CPLEX, Porte Neuve, Bat A, 4 Av Alphonse Morel, Grasse, Al, 06130, France, vincent.beraudier@fr.ibm.com, Philippe Couronne Python’s scipy provides tools for large-scale predictive/ prescriptive analysis for manipulating, cleaning, and crunching data, and publication-quality graphics. The use of these web based tools with both state-of-the art OR solvers and cloud computing will allow new users to enter the world of OR. Users with few development skills can leverage state-of-the-art solvers to develop, tune and publish their results without installing software. Come and discover those scientific python pillars. 4 - A Mixed Interger Programming Model for Optimizing Wheel Tru Operations for a Locomotive Rajeev Namboothiri, GE Global Research, Susanne Heipcke, FICO Xpress Optimization, FICO House, Starley Way, Birmingham, B37 7GN, United Kingdom, susanneheipcke@fico.com

MD32 32-Room 409, Marriott Big Biological Data: Computational and Analytical Challenges Cluster: Big Data Analytics in Computational Biology/Medicine Invited Session Chair: Jian Peng, Assistant Professor, University of Illinois, 2118 Siebel Center, 201 N Goodwin Ave, Urbana, IL, 61801, United States of America, jianpeng@illinois.edu 1 - Reconstruction of Species Histories using Genomic Data Siavash Mirarab, UCSD, Jacobs Hall, EBU1, 2nd Floor, University of California, San Diego, San Diego, CA, 92093, United States of America, smirarab@gmail.com, Shamsuzzoha Bayzid, Bastien Boussau, Tandy Warnow Reconstructing phylogenies, trees that show evolutionary histories of species, can be now attempted using genomic data. Building these species trees is complicated by potential differences between evolutionary histories across the genome. In this talk, we introduce two new methods used to infer whole-genome phylogenies of 48 birds and 103 plants. These new algorithms can analyze datasets with thousands of genes and species with high accuracy, and can account for weak signal across the genome. 2 - Entropy-Scaling Search of Massive Biological Data Sets Noah Daniels, MIT, 32 Vassar St., 32G-572, Cambridge, MA, 02139, United States of America, ndaniels@csail.mit.edu Recently, we have seen an exponential increase in biological data, outpacing advances in computing power. Extracting new science from these massive datasets requires algorithms that scale sublinearly in the size of the datasets. We present a novel entropy-scaling data structure for similarity search. Applying this data structure provides massive acceleration of several standard tools in three biological domains: genomics, high-throughput drug screening, and protein structure search. Chair: Gino Lim, Department Chair, Hari And Anjali Agrawal Faculty Fellow, Associate Professor, University of Houston, E206, Engr. Bldg 2, Houston, TX, 77204, United States of America, ginolim@Central.UH.EDU 1 - Benders Decomposition and an LP-based Heuristic for Selecting IMRT Treatment Beam Angles Sifeng Lin, The University of Texas at Austin, 1 University Station C2200, Austin, TX, 78712, United States of America, sifenglin@utexas.edu, Jonathan Bard, Gino Lim This talk presents two Benders decomposition algorithms and a novel two-stage integer programming-based heuristic to optimize the beam angle and fluence map in Intensity Modulated Radiation Therapy planning. The results indicated that implementing Benders using the lazy constraint usually led to better feasible solutions than the traditional approach. Moreover, the LP rounding heuristic can generate good solutions quickly, with further improvement obtained with the local branching search. 2 - Robust Optimization for Craniospinal Irradiation using Intensity Modulated Proton Therapy Li Liao, Research Assistant, University of Houston, 4800 Calhoun Rd, Houston, TX, 77004, United States of America, lliao5@uh.edu, Gino Lim, Xiaodong Zhang Conventional passive scattering proton therapy (PSPT) is an extremely complex technique for craniospinal irradiation (CSI). In this study, we proposed a robust intensity modulated proton therapy (IMPT) for CSI. A small dose deviation can be achieved when ±3 mm mis-alignment errors were applied on field junction for the robust IMPT plans, whereas this index was more than 40% for PSPT plans. A simplified dose model was introduced to predict dose deviation in different field arrangement situation. MD33 33-Room 410, Marriott Radiation Therapy Optimization: Algorithms and Biological Effects Sponsor: Health Applications Sponsored Session

John F Welch Technology Centre, Bangalore, India, Rajeev.Namboothiri@ge.com, Srinivas Bollapragada, Reejo Mathew, Mark Smith

FRA regulations mandate tolerance limits on locomotive wheel measurements for safe locomotive operations. The wheel tru machine operator needs to decide the amount of material to be trued from each wheel, and the wheels that need to be replaced with inventory wheels, in order to comply with FRA regulations. In this

241