Informs 2015

439 / 552

INFORMS Philadelphia – 2015
437
WC33
33-Room 410, Marriott
Health Care Operations Management
Sponsor: Health Applications
Sponsored Session
Chair: Amin Khademi, Assistant Professor, Clemson University,
130-D Freeman Hall, Clemson University, Clemson, SC, 29634,
United States of America,
1 - Improving Outpatient Scheduling through Patient Complexity and
Integer Programming
Eva Lee, Georgia Institute of Technology,
Prashant Tailor,
This work is joint with Emory Brain Center. We develop a clinical tool that
classifies and schedules patients based off a patient complexity metric. The goal is
to maximize the number of patients seen and increase providers and patient
satisfaction. A classification model is first used to predict patient complexity,
severity and type of follow-up appointment. This information is then used within
the MIP scheduling model.
2 - Classifying Heterogeneous Tumor Subtypes via Matrix
Factorization and Mixed-integer Programming
Andrew Trapp, Assistant Professor, Worcester Polytechnic
Institute, 100 Institute Rd., Worcester, MA, 01609,
United States of America, , Patrick Flaherty
We consider tumor subtype classification via regularized mixed-membership
matrix factorization, where one factor matrix has a limited number of non-zero
entries, and the other has simplex constraints. This provides a mixed-membership
representation for each column of the original matrix with sparse mixing
components. We transform the original and NP-hard biconvex optimization
problem into a mixed-integer linear program, and discuss exact and approximate
solution approaches.
3 - Prediction of Operating Room End Time using
Regression Modeling
Robert Allen, Clemson University, 130 Freeman Hall, Clemson,
SC, United States of America,
Kevin Taaffe
The decision to let staff go home or bring more staff in depends on the ability of
the nurse manager to predict when certain rooms will be finished for the day. We
explore several different methods of enhancing the nurse’s predictive capability
by using hospital process flow data to build several models aimed at predicting the
OR end time. We compare varying predictive models such as a simple offset and
regression modeling to better predict the room end offsets as they occur during
the day.
WC34
34-Room 411, Marriott
Joint Session HAS/Analytics: Unleashing the
Potential of Big Data using Visualization in
Health Care Delivery
Sponsor: Health Applications
Sponsored Session
Chair: Mustafa Ozkaynak, Assistant Professor, University of Colorado,
13120 E 19th Ave, Aurora, CO, 80045, United States of America,
1 - Understanding Adherence and Prescription Patterns using
Large Scale Claims Data
Margret Bjarnadottir, Assistant Professor of Management Science
and Statistics, Robert H. Smith School of Business, University of
Maryland, 4324 Van Munching Hall, College Park, MD, 20742,
United States of America, ,
Sana Malik, Catherine Plaisant, Tanisha Gooden,
Eberechukwu Onukwugha
Traditionally, studies have measured medication adherence using summary
statistics. However, advanced computing capabilities and novel visual analytics
tools now allow us to move beyond the traditional reporting of “average
adherence” to analyze longitudinal adherence patterns. Utilizing EventFlow, a
novel discrete event sequence visualization software, we investigates patterns of
prescription fills and illustrate the use of visual analytics tools in summarizing
large scale claims data.
2 - Visualizing Differences in Patient Use of an EHR Patient Portal
Informed by Clickstream Data
Sharon Johnson, Associate Professor, Worcester Polytechnic
Institute, Foisie School of Business, 100 Institute Road, Worcester,
MA, 01609, United States of America, ,
Farhan Mushtaq, Bengisu Tulu, Diane Strong, John Trudel,
Lawrence Garber
In this paper, we explore patient usage behavior of a patient portal by analyzing
patterns of use in clickstream data combined with data on demographics and
health system utilization. Directed and undirected mining techniques were used
to explore the data and to visualize specific patterns. This type of analysis can be
used to improve processes for engaging patients through a patient portal, as well
as to enhance the portal interface to support different user needs.
3 - Visualization of Care Delivery to Asthma Patients in Pediatric
Emergency Departments
Mustafa Ozkaynak, Assistant Professor, University of Colorado,
13120 E 19th Ave, Aurora, CO, 80045, United States of America,
Marion Sills
We used Eventflow (an interactive visualization tool) to examine the temporal
relationship between care delivery activities for asthma patients in an academic
pediatric emergency department and its four satellite clinics. Time-stamped event
logs from Electronic Health Records were processed and workflow patterns in
each of the five settings for different acuity level and arrival mode were
highlighted. Findings can inform systematic organizational interventions that will
improve quality of care.
4 - Healthcare Process Discovery and Visualization
Rahul Basole, Associate Professor, Georgia Institute of
Technology, 85 Fifth Street NW, Atlanta, GA, 30332,
United States of America, , Mayank Gupta,
Mark Braunstein, Polo Chau, Hyunwoo Park, Robert Pienta,
Brian Kahng, Vikas Kumar, Nicoleta Serban, Michael Thompson
Healthcare processes are complex activities that span organizational, spatial, and
temporal boundaries. Systemic insights are consequently difficult to achieve. Our
research develops a data-driven methodology, fusing systems modeling, data
mining, and visualization, to identify, describe, and visualize healthcare processes.
We illustrate our methodology with a case study in pediatric healthcare.
WC35
35-Room 412, Marriott
Global Issues I
Contributed Session
Chair: Satish Nargundkar, Associate Professor, Georgia State University,
35 Broad St., Suite 827, Atlanta, GA, 30302, United States of America,
1 - Creative Destruction through Online Education:
What Industry and Students Can Teach Academics
Andrei Villarroel, Professor, Swiss Entrepreneurship Institute,
4 Chemin du Musee, Fribourg, Switzerland,
Our research spans 54 countries, 44 industries, unveiling the value of a new
generation of online education organizations from HEI professors, industry
practitioners, and current students. Amongst respondents with first-hand
experience with MOOC education, they find it superior to traditional education in
14 out of 15 dimensions long-believed better served by the traditional campus-
based education model.
2 - Orientation Determination of Hotel Buildings using 3-d GIS for
Maximum Rental Revenue
Young-ji Byon, Assistant Professor, Khalifa University, Al Saada
St. and Muroor Rd., Abu Dhabi, 127788, United Arab Emirates,
Joonsang Baek, Chung-suk Cho
Hotel room rates are strongly related to the quality of scenery from the rooms. In
Dubai and Abu Dhabi, great number of new hotel constructions are being
planned. By utilizing the 3-D elevation model in GIS, it is possible to quantifiably
determine the optimal direction of the hotel buildings before the constructions
begin, that would maximize the scenic views of the hotels and hence also the
room rental revenue.
3 - Healthcare Analytics Data and Insights on Patient Satisfaction
Satish Nargundkar, Associate Professor, Georgia State University,
35 Broad St., Suite 827, Atlanta, GA, 30302, United States of
America, Subhashish Samaddar
The lack of patient satisfaction data across the hospital system limits the ability of
researchers to investigate the customer service elements of the patient experience.
In this paper we collect and organize hospital-level patient satisfaction data over a
six year period from multiple websites, and make it available to the research
community. We identify research opportunities in healthcare quality analytics.
We analyze the data to offer some insights on patient satisfaction across the U.S.
WC35