Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TC68

3 - Go to YouTube and See Me Tomorrow: The Role of Social Media in Managing Chronic Conditions Xiao Liu, University of Utah, Salt lake City, UT, United States, Bin Zhang, Anjana Susarla, Rema Padman To assess the medical knowledge in YouTube videos, we propose an interdisciplinary lens that synthesizes deep learning methods with themes emphasized in Information Systems (IS) research and research on healthcare informatics. We extract medical terminology from videos. We annotate videos using inputs from domain experts and build a logistic regression based classifier to categorize videos based on whether they encode a high degree of medical knowledge or not. We find that medical terminology embedded in textual data is more salient to an assessment of medical knowledge encoded in a video, rather than image analytics. 4 - Predicting Hepatocellular Carcinoma Recurrences: A Data-Driven Multiclass Classification Method by Incorporating Hidden Risk Factors Qiuhua Sheng, University of Utah, Salt Lake City, UT, 84102, United States, Da Xu, Paul Hu, Tingshuo Huang, Wen-Chen Lee Hepatocellular carcinoma (HCC), a malignant disease, is normally treated with surgical resections that however often associated with high cancer recurrence rates. We propose a Bayesian network-based method to infer HCC recurrences by incorporating distinctive pathogenesis that differs between early and late recurrences. The proposed method considers the underlying mechanisms control the clinical endpoint in the learning process and offers interpretability and flexibility to support HCC prognosis predictions. n TC68 West Bldg 105C Joint Session QSR/DM: Developments in Additive Manufacturing Systems Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Arman Sabbaghi, Purdue University, West Lafayette, IN, 47907, United States 1 - An Adaptive Data Augmentation Strategy for Fitting Gaussian Gaussian process (GP) models are useful tools for fitting models that predict in- plane deformation of 3D-printed products. However, products are often produced sequentially (either individually or in batches), making it necessary to develop computational procedures that allow sequential augmentation of data obtained from manufactured products. We propose a novel approach for adaptive estimation of GP model parameters using an inferential tool called confidence distribution (CD) that also helps capture the uncertainty associated with the estimators. The method can also help reduce the computational burden associated with GP modeling using the “divide and conquer strategy. 2 - Data and Compute Challenges in Enabling 3D Printing for Mass Customization Jun Zeng, HP Labs, HP Inc., 1501 Page Mill Road, Palo Alto, CA, United States Industry 4.0 will disrupt the 12 Trillion-dollar manufacturing business by enabling producing customized products with mass manufacturing cost efficiency. 3D printing holds the promise to realize this Industry 4.0 vision however there are multiple technology challenges from design to production covering a broad spectrum of hardware, software and material sciences. This paper will describe generative design thinking (aka. design-for-additive manufacturing), interoperability and file formats (e.g., 3MF), high-fidelity manufacturing including process, sensing, and data-driven compensation and control drawing from our learning from commercializing HP’s Multi Jet Fusion technology. 3 - Understanding the Requisite Ecosystem to Qualify Aerospace Additive Manufactured Parts William Bihlman, Aerolytics LLC, West Lafayette, IN, United States Additive manufacturing (AM) shows promise to be highly disruptive in many industrial markets. Aerospace, however, lags. This presentation will address the primary challenges that are rather unique to commercial aerospace. It explores questions such as: Why is aerospace slow to adopting this technology? What are the main targets/components? How does this differ between engine vs airframe markets? In short, it will discuss the requisite ecosystem to necessitate adoption of AM for one of society’s highest risk-adverse industries. Indeed, AM is a powerful new paradigm. But is aerospace able to engage? Process Models with Application to 3D Printing Tirthankar Dasgupta, Rutgers University, Minge Xie

4 - A Data Fusion Framework for Historical and Prototype Experimental Data with Applications to Personalized Heart Surgery Planning and Optimization Chuck Zhang, Georgia Institute of Technology, Atlanta, GA, United States, Jialei Chen, Kan Wang, Simon Mak, Roshen Vengazhiyil, Ben Wang This paper presents a study of augmenting historical patient data with experimental data based on 3D printed patient prototypes to develop more reliable predictive models for physicians and surgeons to use to make more informed decisions for surgery planning and optimization. This research work involves innovative materials design for creating tissue-mimicking structures, multi-material 3D printing, machine learning for generating “virtual patients, and data fusion. This method is demonstrated through an application case of outcome prediction of transcatheter aortic valve replacement (TAVR) surgery. n TC69 West Bldg 106A Joint Session QSR/Practice Curated: Reliability and Quality for Industry Applications Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Zhimin Xi, University of Tennessee - Knoxville, 851 Neyland Dr, 517 JDT, Knoxville, TN, 37996, United States 1 - Lithium-ion Battery Pack SOC Estimation Considering Cell-to-Cell Variability Accurate estimation of lithium-ion battery pack SOC is technically challenging because of the cell-to-cell variability caused by the manufacturing tolerance. In addition, there is no unanimous definition of the pack SOC since each cell has its own SOC and the pack can be configured in different ways. This study firstly investigates different pack SOC definitions and adopts the one best suitable in our study. Next, uncertainty modeling and propagation analysis are conducted for accurate pack SOC estimations considering the cell-to-cell variability. Both analytical and simulation solutions are reported under different scenarios. 2 - A New Collision Avoidance Algorithm for Autonomous Vehicles Based on the Velocity Obstacle Algorithm Elnaz Torkamani, Rutgers, The State University of New Jersey, Piscataway, NJ, United States, Zhimin Xi Vehicle navigation autonomously in a dynamic environment is a challenging task because it should not be pre-programmed under a given handful situations and the vehicle must be able to avoid collisions under numerous unforeseeable but reasonable situations to a human being. From the practical perspective in autonomous cars, the algorithm must be efficient and reliable. This study proposes a new vehicle collision avoidance algorithm for multiple obstacles by eliminating the need of running sampling or optimization approaches, so that the algorithm could be practical useful in real-time collision avoidance under high vehicle speed. 3 - Reliability-Based Optimal Design of a Micro-Grid System under Natural Disasters Zhetao Chen, Rutgers, The State University of New Jersey, Piscataway, NJ, United States, Zhimin Xi This study proposes reliability-based optimal design of a micro-grid system under service disruptions due to natural disasters. The objective is to determine the minimum number of generators and their distributions in the micro-grid so that the system’s recoverability can be guaranteed under random failure scenarios of the power transmission lines. Power flow analysis combing with the Monte Carlo simulation (MCS) are used for uncertainty propagation analysis to quantify the system’s recoverability distribution under random failure scenarios of the transmission lines. The proposed work is demonstrated through a 12-bus power system. 4 - Finite Element Modeling of the Selective Laser Melting Process for Ti-6Al-4V Alaa Olleak, Rutgers, The State University of New Jersey, Piscataway, NJ, United States, Zhimin Xi Physics-based modeling of the selective laser melting (SLM) process is critical for better understanding the influence of the parts quality with respect to various process parameters and scanning strategies. The challenge is to balance model validity, domain size, and computational efficiency so that the model can be practically useful for improving reliability and quality of the printed products. In this study, a transient thermal finite element model of a SLM process for Ti-6Al- 4V is developed using ANSYS for predicting the melt pool size and its temperature history. The thermal model is then integrated with mechanics-based modeling process for predicting the residual stress of the products. Modjtaba Dahmardeh, Rutgers, The State University of New Jersey, Piscataway, NJ, United States, Zhimin Xi

332