Informs Annual Meeting 2017

WB67

INFORMS Houston – 2017

3 - A Stochastic Program Approach for Path Reconstruction Oriented Sensor Location Model Ning Zhu, Tianjin University, Tianjin, Weijin Road no. 92, China, Tianjin, China, zhuning@tju.edu.cn, Fu Chenyi, Ma Shoufeng Path flow identification is of particular interest for a number of traffic applications. We propose a scenario based two stage stochastic programming framework which considers the uncertainty of the link-path matrix. The first stage model aims to minimize the total traffic sensor installation cost and the expected penalty for uncovered and undifferentiated paths. The second stage model attempts to minimize uncovered and undifferentiated paths for a given sensor location pattern and a specific scenario. A branch and bound based integer L-shaped algorithm is presented. Extensive numerical experiments are conducted to verify the effectiveness of the proposed models and algorithm. 4 - Investigating Significant Features Contributing on Safety of Pedestrians at Intersections Over the Years Hamidreza Ahady Dolatsara, Auburn University, 311 West Glenn Ave., Apt. 28, Auburn, AL, 36830, United States, hamid@auburn.edu The objective of this study is investigating factors affect safety of pedestrians over the years. For this purpose Machine learning algorithms will utilize to extract the important features and predicting crash frequencies at intersections. 5 - The Morning Commute Problem with Ridesharing and Dynamic Parking Charges Rui Ma, University of California, Davis, 2041 Academic Surge, University of California, One Shields Avenue, Davis, CA, 95616, United States, drma@ucdavis.edu, Michael Zhang We study the traffic flow patterns in a single bottleneck corridor with a dynamic ridesharing mode and dynamic parking charges. Besides the scheme with constant parking charges and ridesharing payments, dynamic parking charges and ridesharing payments are derived to achieve congestion-free traffic in the corridor. With the dynamic ridesharing ratios, it is found that genuinely nonlinear departure rates and travel time functions can be generated in certain ridesharing cases, which was not observed in the traditional ADL model (Arnott et al., 1990) for the morning commute problems without ridesharing or with constant ridesharing ratios. 371B Data Analytics to Advance the Science of Health Care Delivery Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Devashish Das, Mayo Clinic, Rochester, MN, 55901, United States, das.devashish@mayo.edu 1 - Data Analytics Framework for Online Chronic Disease Management Based on Remote Health Information Systems with a Case Study in Smart Asthma Management An array of advancements in information technology has resulted in a next- generation health information system based on remote patient monitoring technology and such system is especially emphasized in chronic disease management because of the system’s consistent monitoring capability. In this paper, we first discuss some pitfalls of analyzing data from the remote health systems that may mislead patients and their care providers. Then, we introduce a set of principles that enables more effective data analytics based on remotely acquired patient data. We believe the discussion in this paper is crucial for unleashing a new era of technology-enabled and data-driven chronic disease management. 2 - Impact of Resuscitation Activations on Patients Left Behind in an Emergency Department Xiaochen Xian, 936 Eagle Heights Apt C, Madison, WI, 53705, United States, xxian@wisc.edu, Mustafa Y Sir, Devashish Das, Kalyan Pasupathy In the Mayo Clinic Emergency Department, a level one trauma center, a resuscitation team is formed by pulling resources from different treatment areas on a rotational basis to respond to incoming trauma patients. This impacts the care delivery process of patients that are left behind by the staff members responding to the resuscitation activation. We use a functional regression model to study the impact of these activations and system recovery time. This model can help design a workload-based resuscitation model instead of a rotation-based model. WB67 Junbo Son, University of Delaware, 303 Alfred Lerner Hall, Newark, DE, 19716, United States, junboson@udel.edu, Shiyu Zhou, Patricia Brennan

3 - Personalizing Depression Treatment Follow-up via Rule-based Method Ying Lin, University of Houston, Houston, TX, United States, ylin58@uh.edu, Shuai Huang, Shan Liu Depression is a common mental illness with complex and heterogeneous progression dynamics. Risk grouping of depression treatment population based on their longitudinal patterns has the potential to enable cost-effective monitoring policy design. This study aimed at identifying a set of risk-predictive longitudinal patterns that predict depression progression and segment population into subgroups with different risk implications using rule-based method. We applied the rule-based method on an electronic health record dataset of depression treatment population and identified 12 rules that provide better individual prognostics and enable cost-effective monitoring policy design. 4 - Improving Accuracy of Non-invasive Hemoglobin Monitors: A Functional Regression Model for Streaming SPHB Data Devashish Das, Mayo Clinic, 2015, 41st Street NW, F47, Rochester,

MN, 55901, United States, das.devashish@mayo.edu, Kalyan Pasupathy, Nadeem N. Haddad, Susan Hallbeck, Martin Zielinski, Mustafa Y. Sir

In this article, we propose a method of improving accuracy of SpHb monitors, which are non-invasive hemoglobin monitoring tools with potential to improve critical care protocols in trauma care. The proposed method is based on fitting smooth spline functions to SpHb measurements collected over a time window and then use a functional regression model fit to predict the adjusted SpHb measurement for the end of the time window. The accuracy of the proposed method is compared to traditional methods.

WB68

371C Uncertainty Analysis in Modeling, Control, and Optimization Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Du Dongping, Texas Tech University, Lubbock, TX, 79409, United States, dongping.du@ttu.edu 1 - Generalized Polynomial Chaos-based Uncertainty Propagation in Multi-scale Modeling of Cardiac Electrophysiology Zhiyong Hu, Texas Tech University, Lubbock, TX, United States, zhiyong.hu@ttu.edu, Dongping Du Cardiac modeling has been widely used in physical experiments and clinical studies to uncover disease mechanism and investigate treatment strategies. Cardiac functions involve many uncertainties resulting from intrinsic variability. However, most of the existing cardiac models cannot efficiently quantify the effect of uncertainty on model prediction. This study develops a Generalized Polynomial Chaos based uncertainty propagation technique to propagate parametric uncertainties across multi-scale organizational level of cardiac ion channel, cell, and tissue. The method shows superior performance, compared with the Monte Carlo method, in computational efficiency and accuracy. 2 - Porosity Analysis for Additive Manufacturing by Augmented Spatiotemporal Point Process Chenang Liu, Virginia Tech, 837 Claytor Square, Blacksburg, VA, 24060, United States, lchenang@vt.edu, Jia Liu, Zhenyu Kong, Prahalad Rao Porosity is one major defect in additive manufacturing (AM) process. The research on modeling, forecasting, and control of porosity is critical to improve AM part quality. Using spatial and temporal correlations of pores due to layer- wise building, an augmented spatiotemporal point process is proposed to identify and forecast areas prone to pores on next layers. By this method, the distribution of pores captured from CT scan images is effectively represented by the proposed augmented point patterns incorporating morphological features of pores. This method is demonstrated by a real AM part with CT scan images. 3 - Physics-driven Spatiotemporal Regularization for High- dimensional Predictive Modeling Bing Yao, Pennsylvania State University, 801 B6 Southgate Dr, University Park, PA, 16802, United States, bzy111@psu.edu, Hui Yang Rapid advancement of distributed sensing and imaging bring a lot of high- dimensional spatiotemporal data in complex systems. Traditional regression fails to describe the spatial and temporal correlation of these data, and is not generally applicable for predictive modeling in these systems. This talk presents a physics- driven spatiotemporal regularization (STRE) method for high-dimensional predictive modeling in complex dynamic systems. This model is implemented to predict the electric potentials on the heart surface from the ECG data on the body surface. The model performance is evaluated and validated in both a simulated two-sphere geometry and a realistic torso-heart geometry.

498