Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MA58

4 - Patient Admission Decision at Emergency Department under Mass Casualty Incident Taesik Lee, KAIST, 291 Daehak-ro, Yuseong-gu, Daejeon, Korea, Republic of, Hyun-Rok Lee In the aftermath of a mass casualty incident, an emergency department (ED) may divert less critical patients to preserve its resources for possible future arrivals of more critical ones. Justification for the diversion is to increase the overall expected survival of the entire casualties. When this decision making is a challenge for a single ED case, multiple ED environment makes it even more difficult. We formulate the problem as a decentralized partially observable MDP model, where individual EDs make a diversion decision with only partial information on other EDs’ state. Heuristic solution techniques to approximately solve the model are presented and compared for their performance. n MA58 West Bldg 101C Operations Research for Medical Decision Making Sponsored: Health Applications Sponsored Session Chair: Brian T. Denton, University of Michigan, Ann Arbor, MI, 48109- 2117, United States Co-Chair: Lauren N. Steimle, University of Michigan, Ann Arbor, MI, 48105, United States 1 - Robust Personalized Treatment Selection for Partially Observable Chronic Conditions Jue Gong, University of Washington, Seattle, WA, 98115, United States, Shan Liu For many chronic diseases, an individual patient may experience many progression pathways. Developing a personalized treatment plan is a difficult sequential decision-making problem. The natural stochasticity of the disease progression makes the estimation of the disease model difficult, and inaccurate model reduces the performance of the treatment selection strategy. We will develop novel approaches for modeling the uncertainties in individual disease progression and finding the robust POMDP policy optimal for individual patients. A data-driven method to construct or calibrate the uncertainty set is required to Julia L. Higle, University of Southern California, University Park Campus, GER 240B, Los Angeles, CA, 90089-0193, United States, Jing Voon Chen Recent studies indicate the existence of fallopian tube lesions as a precursor to serous ovarian cancer, the most aggressive form of ovarian cancer. This suggests an opportunity for prevention of serous ovarian cancer through surgical removal of the fallopian tubes (salpingectomy). Using a model of ovarian cancer that is differentiated by patient risk (i.e., genetic mutation status) and ovarian cancer sub-type, we examine the potential impact of prophylactic salpingectomy. 3 - Behavioral Intervention Design using Precision Analytics Yonatan Mintz, UC Berkeley, Berkeley, CA, United States, Anil Aswani, Philip Kaminsky, Elena Flowers, Yoshimi Fukuoka In this talk, we describe a precision analytics framework that uses patient data to effectively design personalized weight loss interventions. Our framework utilizes a utility maximization model for patient behavior which combined with integer programming and Bayesian prediction allows us to create several personalized interventions, as well as aggregate these interventions into a cohort weight loss program. We then present clinical trial and simulation results which show that our method maintains efficacy while potentially reducing the associated person hours and cost of the intervention. 4 - Two-step Markov Process Methodology for Parameterizing Cancer State Transitions in Limited-data Settings Chaitra Gopalappa, University of Massachusetts Amherst, 160 Governors Dr, Amherst, MA, 01003, United States, Guo Jiachen, Prashant Meckoni, Buyannemekh Munkhbat, Carel Pretorius, Jeremy Lauer, Andre Ilbawi, Melanie Bertram Among all premature deaths from non-communicable diseases (NCDs) reported globally, 90% occurred in low and middle income countries. To address this burden, the WHO-CHOICE analyses developed a map of the æBest Buy’ interventions to update the Global NCD Action Plan for 2013-2020. The evidence for æBest Buys’ were based on model predictions of cost-effectiveness of alternative intervention scenarios. Our team developed models and predictions for 3 types of cancers, breast cancer, colorectal cancer, and cervical cancer, including a new methodology for parameterizing a natural history model for countries where longitudinal cancer registry databases are not available. We will present this work. guarantee the effectiveness of the individualized robust policy. 2 - Ovarian Cancer Prevention and Risk Reduction: A Model-based Analysis

n MA59 West Bldg 102A Teams in Healthcare Delivery Sponsored: Health Applications Sponsored Session Chair: Song-Hee Kim, CA, United States 1 - Collaboration, Interruptions and Changeover Times: Model and Empirical Study of Hospitalist Processing Times Itai Gurvich, Cornell University, 2 W. Loop Rd, New York, NY, 10044, United States, Jan A. Van Mieghem, Kevin O’Leary, Lu Wang Collaboration may lead to interruptions. Task switching can introduce changeover times when resuming the preempted task and thus can increase total processing time. We analyze and quantify how collaboration, through interruptions and discretionary changeovers, affects total processing time. We introduce an episodal workflow model that captures the interruption and discretionary changeovers dynamics—-each switch and the episode of work it preempts—-present in settings where collaboration and multitasking is paramount. We then deploy the model in a field study of hospital medicine physicians—-``hospitalists’’ to estimate the total changeover time they incur. 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, 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 greater prior partner exposure directly improves performance for an unstandardized process. For a more standardized process, this effect is moderated by a new recruit’s total experience. For both processes the effect is more pronounced at times of high workload. We explore the implications of our results for team formation strategies by balancing the benefits of partner diversity with those of team familiarity. 3 - Cross-disciplinary Temporary Teams in Health Care Organizations: The Role of Partner Variety and Shared Experience Hummy Song, The Wharton School, University of Pennsylvania, 3730 Walnut Street, 560 Jon M. Huntsman Hall, Philadelphia, PA, 19104, United States, Song-Hee Kim, Melissa Valentine We examine how managers can compose temporary teams of cross-disciplinary health care professionals to facilitate team performance. We leverage random team and task assignment in an emergency department to identify the performance effects of shared experience and partner variety as they are accumulated within a shift versus over the longer term among physicians and nurses. We propose easily implementable managerial recommendations. Using simulation, we illustrate that the gains in patient throughput would be substantial, even without increasing staffing levels. 4 - Toward an Effective Design of Preventive Health Care Delivery: Collaboration with Primary Care Providers

Yingchao Lan, Ohio State University, 2100 Neil Avenue, Fisher Hall 252C, Columbus, OH, 43210, United States, Aravind Chandrasekaran

Chronic diseases are responsible for 70% deaths among Americans each year and around 75% of the nation’s health spending. While both practitioners and researchers have recognized the importance to design an effective preventive care delivery system to improve population health and health delivery efficiency, it’s still unclear how. This paper addresses this gap by studying how collaboration with primary care providers can improve population health and healthcare delivery efficiency.

140