Poster sessions were held during the conference and below is the list of posters that were presented at the designated session times. Click on the title of a poster to jump to its abstract. Posters

Poster Listing

An Institutional Framework for Disaster Recovery

David Swanson (University of Arkansas)


Analytic Solution to the Susceptible-Infective Disease Spread Model with Varying Contact Rate

Hamed Yarmand (North Carolina State University), Julie Ivy (North Carolina State University)

A common class of epidemiological models developed for the spread of infectious diseases is the Kermack-McKendrick model and its variations. These models are represented as systems of ordinary differential equations which in most cases are strongly nonlinear and cannot be solved analytically. In this research, we consider SEIR and SI models, two variations of Kermack-McKendrick model. The model names are based on the epidemiological classes included in the model: S for susceptible, E for exposed, I for infective, and R for recovered. One important parameter which affects the infection rate is the contact rate, the rate at which individuals make effective contacts (contacts which result in the disease transmission). We let contact rate be a function of number of infective individuals, which is an indicator of the disease spread during the course of the outbreak. We use the Markov process approach to represent SEIR and SI models as continuous-time Markov chains. The result would be a pure death (or birth) process with state-dependent rates for the SI model, for which we find the transient as well as the steady-state probability distribution of the associated continuous-time Markov chain by solving the Kolmogorov forward equations. Finally we use the solution to the Markov chain to find the analytic solution to the original SI model.



Hiral Modi (Georgia Institute of Technology), Ruban Monu (Georgia Institute of Technology), Santosh S. Vempala (Georgia Institute of Technology)

We describe the C4G Basic Laboratory Information System (BLIS), a joint initiative of Computing for Good(C4G) at Georgia Institute of Technology, the Centers for Disease Control and Prevention (CDC) and Ministries of Health in several countries in Africa. BLIS focuses on addressing two major areas of public health systems in developing regions: (1) the need to efficiently manage and maintain data about patients, specimens and test results generated within a laboratory facility and (2) the need for an efficient way of dissemination of aggregate laboratory data to public health officials. The system is designed to work in low-resource laboratories and across sites with intermittent or no internet availability. To overcome challenges of varied practices, workflow and terminology being utilized across laboratories in PEPFAR countries, the system has been developed to enable each laboratory to customize and configure the system in a way that suits them best. We outline other challenges that developing the system entailed.

C4G BLIS in its current version is a low-cost, easy-to-configure solution that enables the laboratories to manage clinical data and disseminate aggregate trends in real time. A pilot phase is ongoing in several laboratories in Cameroon and Uganda with similar efforts to begin in Ghana and Tanzania. Throughout this pilot phase, the emphasis has been on incorporating feedback as it is received and sending out regular, incremental updates to the pilot laboratories. By the end of this period of testing with the volume of clinical laboratory data being accumulated, we aim to obtain a stable and reliable version of the system. If quantitative measurements of benefits, usability and sustainability of C4G BLIS indicate that it is an effective tool for laboratory information management, then it be scaled up to all other laboratories within the participating countries.


Capacity Factor Analysis: A Decision Support Model for Selecting & Designing Basic Needs Infrastructure in Developing Countries

Justin Henriques (University of Virginia), Garrick Louis (University of Virginia)


Case Study of a Multi-disciplinary Engineers Without Borders Project in Njinikom, Cameroon

Hannah Kates (Georgia Institute of Technology), Courtney Pare (Georgia Institute of Technology), Christian Weil (Georgia Institute of Technology), Chris Quintero (Georgia Institute of Technology), Andrew Foote (Emory University/Georgia Institute of Technology)


Catch-up Scheduling for Childhood and Adult Vaccination

Hannah Smalley (Georgia Institute of Technology), Faramroze Engineer (University of Newcastle), Pinar Keskinocak (Georgia Institute of Technology), Larry Pickering (Centers for Disease Control and Prevention)


Decisions in Disaster Recovery Operations: A Game Theory Perspective on Actor Cooperation, Communication, and Resource Utilization

John Coles (State University of New York at Buffalo), Jun Zhuang (State University of New York at Buffalo)

Using perspectives from game theory in the problem of cooperative interaction between international and local agencies, we discuss the potential for improvement in humanitarian logistics using cooperative strategies in the developing world. The Indian Ocean tsunami which struck on December 26th, 2004 killed over 160,000 people and destroyed much of the infrastructure in the region. As a result, one of the largest international relief efforts in modern history was mounted to save both life and property and stabilize the devastated region. With the recent disasters in Haiti and Chile in January and February 2010 respectively, the need for a more holistic approach to interagency cooperation has become increasingly clear in order to increase the efficacy of these partnerships. A lack of sensitivity to these critical issues could render the desired positive long-lasting changes or recovery of the region impossible due to the failure of aid to address economic and social issues in a culturally acceptable manner. Although sensitivity to cultural issues is challenging in the initial response phase, identifying and employing sustainable initiatives throughout the recovery phase is critical to the acceptance of aid and long-term recovery of the region.


Designing Optimal Water Quality Monitoring Network for River Systems and Application to Altamaha River

Chuljin Park (Georgia Institute of Technology), Seong-Hee Kim (Georgia Institute of Technology), Ilker Telci (Georgia Institute of Technology), Mustafa Aral (Georgia Institute of Technology)

The problem of designing a water quality monitoring network for river systems is to find the optimal location of a finite number of monitoring devices that minimizes the expected detection time of a contaminant spill event with good detection reliability. We formulate this problem as a stochastic optimization problem with a stochastic constraint on detection reliability where both detection time and reliability need to be estimated by simulation. Existing Optimization via Simulation (OvS) algorithms with global or local convergence are not directly applicable to our problem because they are developed without consideration of stochastic constraints. We propose a method called Penalized Objective (PO) that integrates general stochastic constraints into the original objective function and thus converts an optimization problem with stochastic constraints into an unconstrained problem. Then we apply PO to the water quality monitoring problem for the Altamaha River and solves it with an OvS algorithm called the nested partition method.


Development of a Virtual Training System for Transitional Shelter Management in Second Life

Fuminori Toyasaki (York University), Ali Asgary (York University), John Reid (TRP360), Albert Kong (York University)

Post disaster shelter management is an important element of disaster response and recovery operations that requires well trained personnel and volunteers. Virtual learning systems (VLS) can be an effective means of enhancing, motivating, stimulating, as well as for reducing educational costs and have great impacts on the modernization of learning. The paper presents the outcome of an ongoing project aiming to develop a virtual learning system in form of a virtual transitional shelter management (VTSMS). The proposed system simulates the agents who are involved in management of transitional shelter for the purpose of training and education of students, practitioners, and volunteers. In this VTSMS, there will be two types of agents: agents which are controlled by the system using Agent Based Modeling (ABM); avatar agents which are controlled by the real persons (i.e. students, or volunteer). We simulate the VTSMS agents and environment and their interactions and allow students to be part of the simulation by playing the role of transitional shelter management agents in a typical emergency scenario. The VTSMS will be a significant teaching, learning, and research tool in transitional shelter management field because: 1) access to and use of real transitional emergency shelter sites by students is very limited; 2) simulation of post disaster/emergency transitional shelter sites in real world for training purposes are very costly and beyond the budget limits for many educational institutions, Humanitarian NGOs; 3) unlimited number of simulations can be developed and tested. This paper explains various aspects of this system and its usage by students.


Disaster Management: Repositioning Disaster Centers and Determining Stock Levels

Nur Timurlenk (Bilkent University), Okan Dukkanci (Bilkent University), Oner Kosak (Bilkent University), Ali Irfan Mahmutogullari (Bilkent University), Hasim Ozlu (Bilkent University)

In our project, the aim is to reposition disaster centers of Turkish Red Crescent (TRC) and determine stock levels. In disaster management, TRC is responsible for interventions aiming to rescue as many people as possible by efficient techniques. Disaster victims are provided with relief services such as sheltering and food. As their organizational goal, TRC wants to reach disaster places in 2 hours to supply disaster victims with urgent sheltering and psychological support. Moreover, TRC has to keep reasonable amount of inventory for an emergency so that they can supply enough relief material for victims in a short time. To observe whether they are successful with current applications, we examined current strategy of TRC in disaster interventions. We evaluated centers by modeling current locations via ILOG OPL optimization-tool and we found out that with these locations, TRC cannot cover whole country in 2 hours. Moreover, when stock levels are examined after an abroad relief, we observed that they are under critical levels. In order to effectively solve these problems, a mathematical model -including both locations and inventory problems- is constructed. In this model, we give risk points for each city/district so that a place having higher risk point becomes a powerful candidate for being a disaster center or model lessens distances between disaster centers and risky settlements, in addition to 2 hours constraint. Moreover, model allows multi-sourcing so that relief materials can be supplied from different disaster centers in an emergency case. We coded the model in ILOG OPL to determine the locations of TRC centers and their stock levels. In order to observe the implementation of proposed system, we developed a simulation model. For sustainability of model, we provided TRC with a user interface to open or close a disaster center and to determine stock levels for future needs.


Disaster Relief Routing: Integrating Research and Practice

Luis de la Torre (Northwestern University), Irina Dolinskaya (Northwestern University), Karen Smilowitz (Northwestern University)


Disease Spread Model to Evaluate the Effectiveness of Home Confinement Strategy during Pandemic Influenza

Arsalan Paleshi (University of Louisville), Gerald W. Evans (University of Louisville), Sunderesh S. Heragu (University of Louisville), Kamran S. Moghaddam (University of Louisville)

Pandemic influenza has caused large-scale casualties and billions of dollars of loss in human history. The ever changing characteristics of the influenza virus makes new pandemic attacks unavoidable. Intervention strategies to mitigate the transmission of the disease are of great importance for authorities as an alternative to reduce the ill effects of a pandemic attack. The aim of this study is to evaluate the effectiveness of a home confinement intervention strategy on the reduction of the number of infected people during the course of a pandemic influenza. According to this strategy infected people with disease symptoms are confined to their homes until they recover. An agent based simulation model is developed to depict the progress of the disease within the body, and the interaction of agents (i.e. individuals) in various environments such as households, schools, workplaces, and communities in a US metropolitan area. The model is run before and after application of the intervention strategy. The results show that home confinement reduces the number of infected people by 31% when 50% of the infected individuals comply with the rules.


Engineering a Specimen Transport Carrier: Collaborative Humanitarian Research and Design for the Public Good

Victoria M. Gammino (Global Immunization Division, National Center for Immunization and Respiratory Diseases, US Centers for Disease Control and Prevention), Michael F. C. Moreland (SEEDR L3C), Olen Kew (Division of Viral Diseases, US Centers for Disease Control and Prevention), Sue Gerber (Global Immunization Division, National Center for Immunization and Respiratory Diseases, US Centers for Disease Control and Prevention)


Evaluation of Reverse Cold Chain for Wild Polio Virus Surveillance

Allison Taylor (US Centers for Disease Control and Prevention, Johns Hopkins Bloomberg School of Public Health), A.J. Williams (US Centers for Disease Control and Prevention), Steve Oberste (US Centers for Disease Control and Prevention), Steve Wassilak (US Centers for Disease Control and Prevention), Mark Pallansch (US Centers for Disease Control and Prevention)


Field Investigation on the Comparative Performance of Alternative Humanitarian Logistic Structures after the Port au Prince Earthquake: Preliminary Findings and Suggestions

(Rensselaer Polytechnic Institute), Miguel Jaller Martelo (Rensselaer Polytechnic Institute), Tricia Wachtendorf (Disaster Research Center, University of Delaware)


GIS as a Decision Support Tool in Healthcare (Hospital and Bed Availability Analysis for Jefferson County, KY)

Trivikram Rao (University of Louisville)

The existence of hospitals within any community is critical to ensure the functioning of its people in a healthy, uninterrupted manner. This study analyzes the availability of hospitals and beds to the population of Jefferson County, Kentucky in an attempt to gauge the medical preparedness of the county under normal and emergency circumstances. The following Hypotheses are proposed for the purposes of this project:
1. A significant portion of the population in Jefferson County does not have immediate access to hospitals.
2. The number of hospital beds (and other proportional resources) available to the people of Jefferson County is insufficient, especially during emergencies.

Hypothesis #1 was proven to be right, especially for emergency scenarios as 54.16 % of Jefferson county hospitals do not have a hospital within a two mile radius. This value increases to more than 80% without access when only West Jefferson is considered. The number of hospital beds (and other proportional resources) per 1000 people for Jefferson county is a healthy 6.02, well above the state and national averages. However, when we consider individual districts, the west region has a ‘zero’ hospital bed per 1000 people index. Thus the author concludes that the new and current hospitals, beds and other medical resources need to be more evenly distributed throughout the county to enable greater access to the population and facilitate better emergency preparedness.


Health and Humanitarian Logistics: What Beyond "Relief"?

Shibu Mani (Tata Institute of Social Sciences), Mohammed Irshad (Tata Institute of Social Sciences), Saswata Sanyal (Tata Institute of Social Sciences), Faisel Illiyas (Mahatma Gandhi University), Amalraj M (Ernakulam District Collectorate)

The humanitarian logistics in developing countries have a focus on the initial stage/period of the relief phase. The review of such an approach has increasingly been a felt need at the academic side of disaster management. We have aimed at finding the operational limits, opportunities and challenges within the frame work of humanitarian logistics by reviewing case studies/study (original research) from India on two aspects: (a)solar disinfection of contaminated drinking water (b)logistics operations in a cyclone affected region.

The introduction of custom made household level solar water treatment system (during 2001-2006 period)in different physiographic settings (arid, alluvial plains and coastal)but had prevalence of gastro-enteritis in common has shown effectiveness of the method by many well being indicators. The acceptance level of the technology for the long term has been found to be satisfactory with some people whereas with others practicing on a daily basis was found to be very low. It is important to consider the customer based needs in the innovation research for the effective technology transfer.

The field based review of logistic operations in the cyclone Aila affected region in India has shown that: majority of the organizations maintained a steady logistics flow for the first three to four months. The withdrawal of organizations (due to the limitations in operational capacity/time) put the community in a crisis situation almost similar to that of the initial days after the cyclone. How to resolve such an issue where thrust on social infrastructure development is below minimum standards is a question to be addressed.

The lack of social overhead capitals such as health and ware house infrastructures would critically challenge the estimated impact of humanitarian logistics. To go further up (beyond relief), there is a need for linking humanitarian logistics with the developmental paradigm which encompasses all sections of the society.


Hub Location Model for Determining Relief Center Locations During Disaster

Ornurai Sangsawang (Clemson University), Mary Elizabeth Kurz (Clemson University)

After a natural disaster, victims need fast assistance as well as emergency supplies through an efficient distribution system. In this paper, we present the hub location model applied to the decision of locations for relief centers, which will provide medical care, basic human services, public information and transportation. The centers function as hub facilities which interact with affected areas and other centers. The distance is considered as an important factor since it directly affects the time required to get services to the public. The uncapacitated single allocation p - hub median model is considered with a speed-rate factor for flows which are directly transferred between hubs. By designating hubs before the natural disaster occurs, we can influence the order in which the transportation system itself is repaired, so that the routes between hubs can be prioritized. Well - recognized metaheuristics (Tabu Search and Simulated Annealing) are investigated with respect to solution and CPU time.


Identifying Performance Measures in Inventory Management for Disaster Relief Operations

Indraneel Dabhade (Clemson University ), Selina Begum (Clemson University )

The aim of this study is to identify key performance measures for inventory management for disaster relief operations. The results are presented in a format of a balanced scorecard. The indicators have been divided into six major categories covering the horizon from value-based measures to keeping a track of the different utilization factors contributing towards effective and responsive inventory management strategies. A scoring system model has been developed to track the performance of the measures at all periodic reviews. Implementing the scorecard with the scoring model would help the different stake holders be informed of the inventory positions as well as the on field volunteers to optimize the ordering process.


Improving Influenza Pandemic Mitigation Policy through Agent-based Modeling and Simulation Experiments

Michael Beeler (University of Toronto), Dionne Aleman (University of Toronto), Michael Carter (University of Toronto)


Improving the Performance of a Surgical Department

Pinar Keskinocak (Georgia Institute of Technology), Pengyi Shi (Georgia Institute of Technology), Monica Villarreal (Georgia Institute of Technology)


Increasing Survival Chances of EMS Patients While Equitably Locating Facilities

Sunarin Chanta (Clemson University), Maria Mayorga (Clemson University), Laura McLay (Virginia Commonwealth University)

For public sector services, especially basic services, such as Emergency Medical Service (EMS), providing equity of service is an important factor to consider when allocating resources. In contrast to private sector services, the objective here is not to maximize profit, but to save more lives while minimizing inequity of access to service. We present a facility location model which considers the envy associated with p-serving facilities. The minimum p-envy model we present emphasizes survivability by incorporating a survival function which depends on the distribution of operating facilities. The inequity of the system is reduced by minimizing the differences in quality of service between all possible pairs of demand zones with respect to their ordered priority serving facilities weighted by the number of emergency calls in the zones. The chance of vehicles being busy is captured in order to reflect the real operating system using the hypercube model. Since the objective function is complex, a heuristic approach is developed to solve the problem. The model was tested on a real world data set from the EMS system at Hanover County, VA, and also compared to other location models. The results indicate that placing facilities more effectively, following the solution suggested by the proposed model, can save more lives without adding extra facilities. The solution of the proposed model yielded higher number of lives saved than other location models selected to be compared. Sensitivity analysis revealed that the advantage of the proposed model increases as the number of vehicles decreases. Furthermore, the model performs well in terms of coverage, a traditional EMS performance measure. These results provide guidance that is useful and actionable to the field of EMS response planning.


Influence of Framing on Inventory Prepositioning Decisions

Jaime (University of Lugano), Paulo Goncalves (University of Lugano)

Some Humanitarian Relief Organizations (HROs) must preposition emergency items to prepare to serve beneficiaries in the aftermath of a humanitarian emergency. These prepositioning decisions are suitable of being modeled by the Newsvendor Model given that the demand of emergency items coming from beneficiaries is not known beforehand.

Laboratory experiments on the Newsvendor Model have shown that prescribed inventory prepositioning is lower than the optimum when a high stock is required, whereas the prescribed inventory prepositioning is higher than the optimum when a low stock is required; this is known as the anchor and insufficient adjustment bias.

In an experimental design based on the cognitive dissonance theory, we argue that the perceived importance of an item in joint decisions may de-bias the anchor and insufficient adjustment bias. According to this theory, when two simultaneously held cognitions (pieces of knowledge) are inconsistent, the decision maker will experience a state of cognitive dissonance. The theory states that dissonance, being unpleasant, motivates the decision maker to change his/her cognitions.

The design considers two manipulations: a cognitively consonant manipulation, where we bundle a high safety stock decision with a high-importance item and a low safety stock decision with a low-importance item, and a cognitively dissonant manipulation, where the safety stock conditions are reversed. We compare their results against the results of a baseline treatment where such decisions are made independently. Analyses show that bundling two consonant decisions improves results, while bundling two dissonant decisions worsens them. Moreover, the consonant decisions outperform the dissonant ones.

These preliminary results suggest that framing may influence newsvendor-type decisions. In particular, bundling newsvendor-type decisions that differ in their perceived importance may motivate decision makers to change his/her cognitions and, hence, may help to de-bias ordering behavior in newsvendor-type environments.


Integration of Field Test Data for Validation and Analysis of a Cold Chain Simulation Model

Trustin Clear (Georgia Institute of Technology), Michael F. C. Moreland (SEEDR L3C), William Rouse (Georgia Institute of Technology)

This work continues an effort to understand the role of insulated containers in the vaccine and specimen transport cold chains (CC), vital to public health activities worldwide. To facilitate this, we have constructed a discrete event simulation to model container performance in realistic distribution scenarios; our collaborators at SEEDR L3C, an engineering design firm, developed prototype containers designed to exceed WHO Performance Quality Standards for this type of equipment. Using data collected in field tests, we aim to validate the model, refine assumptions, and analyze CC performance with different container types.

A strategy is presented to use operational details and container performance data to construct and validate simulations of equipment field tests, and to make quantitative estimates of system performance using these simulations. These are critical steps in connecting improved container performance with value-creation in the CC, and represent progress toward the goal of building tools to aid public health activities, such as immunization and disease surveillance.

Field tests, designed by CDC and SEEDR, will measure internal and external container temperature versus time during each transport leg; covariates will include refrigerant volume and condition, and transport mode. Data will be analyzed for each link in the distribution network, and model parameters will be adjusted to fit the calibration sample; model fit will be checked for remaining data. Individual link estimates will be combined to yield best-fit parameters for each container type, which will be used to construct a simulation of the complete distribution network.

Field tests begin in February 2011. The analysis of a theoretical data set is included to illustrate the methods under discussion, pending the availability of field test data to model the impact of containers on CC performance.


Locating Facilities for the Strategic National Stockpile

Hugh Medal (University of Arkansas), Ed Pohl (University of Arkansas), Manuel Rossetti (University of Arkansas)

The United States Government has observed that a large-scale bioterror attack on a large city would necessitate such a large amount of medicine and pharmaceuticals that local inventories would quickly be depleted. To address this, the Strategic National Stockpile (SNS) was implemented to provide supplies that are ready to be deployed in a medical emergency. The SNS system consists of strategically placed warehouses around the country and reportedly has the capability to concurrently provide supplies for several emergencies in large cities within 12 hours.

The goal of this research is to recommend how to efficiently design the SNS system and give insight into how conflicting objectives trade off against each other. Specifically, we recommend where to locate warehouses and response vehicles in order to minimize the worst case response time. We use an integer programming approach to solve this problem; in particular we model the problem as a generalization of the set cover problem. We report computational results for our solution method and perform various trade off analyses.


Raha Akhavan-Tabatabaei (Universidad de los Andes), Diomar (Universidad de los Andes), Luis E. Yamin (Universidad de los Andes), Wilfredo Ospina (Universidad de los Andes), Raha Akhavan-Tabatabaei Raha Akhavan-Tabatabaei (Universidad de los Andes)


Managing Debris Operations

Kael Stilp (Georgia Institute of Technology), Ozlem Ergun (Georgia Institute of Technology), Pinar Keskinocak (Georgia Institute of Technology), Antonio Carbajal (Georgia Institute of Technology), Monica Villarreal (Georgia Institute of Technology)

Debris removal is costly, long and complicated process requiring the careful consideration of both short term and long term effects on people’s health and safety, and the environment. In the short term, the main consideration is the clearance of debris to allow for the transportation of relief resources and access to disaster areas or critical facilities for lifesaving activities. Given that the debris may contain toxic or hazardous waste, one needs to weigh the benefits of rapid clearing with the long term impact to ensure that their management would not pose a future threat to human health or the environment.

The first stage of operations that we model, clearance of debris, is a computationally difficult network expansion problem over multiple periods. In this model there are big M constraints, which are a common and well known family of constraints which often make computation difficult. We use the idea of over-restricting these M values, using estimates to set them at much smaller values than is actually valid. These over-restrictions have shown an ability to both speed up heuristics but also give significant improvement in solution quality. Furthermore, we introduce a secondary heuristic which we can use to prove quality of the heuristic solutions generated.

For the second stage that we model, collection of debris, we consider the recent earthquake in Haiti. For Port-au-Prince we create a road network suitable for our needs, estimate debris across the network based on openly available data, and focus the model on minimizing time to completion. We then use various strategies to observe the impact of considering the road network and disposal site openings when designing a collection plan for Port-au-Prince.


Mobilizing Disaster Communication with LifeNet

Hrushikesh Mehendale (Georgia Institute of Technology), Amit Prakash (Jamshetji Tata Centre for Disaster Management, Tata Institute of Social Sciences), Soma Sinha (Jamshetji Tata Centre for Disaster Management, Tata Institute of Social Sciences), Shibu Mani (Jamshetji Tata Centre for Disaster Management, Tata Institute of Social Sciences), Santosh S. Vempala (Georgia Institute of Technology)


Modeling Debris Cleanup Operations

Gary Fetter (Western Carolina University), Mauro Falasca (East Carolina University)

Within the area of disaster recovery, debris collection and disposal represents a major task that requires significant operational planning and control and can severely impact local, state, and federal financial resources. The unique nature of disaster debris and the extreme amounts generated as a result of the disaster event create challenges for decision makers that are not typically encountered during everyday solid-waste disposal operations. The collection and disposal of disaster debris can be quite challenging because the amount of debris is usually extremely significant and is generated very quickly (in a matter of hours or minutes, depending on the type of disaster), far exceeding typical amounts of solid-waste generated on an annual basis. In addition, in the case of large-scale disasters, debris is often spatially scattered throughout a large area encompassing several regions, counties, or states. This research is aimed at identifying the unique aspects of disaster debris disposal and developing a series of decision support tools to assist emergency management coordinators with allocating resources during debris cleanup operations. The nature and importance of debris cleanup to the success of recovery operations has primarily been discussed qualitatively in the relatively few articles published in the academic literature. This research represents an addition to the few quantitative research studies addressing debris cleanup, one of the most important and costly aspects of disaster recovery and management. We present a decision support system framework, discuss aspects of the knowledge base, model base, as well as the user interface, and show how an emergency management coordinator would use the system during ongoing debris cleanup operations. Finally, we demonstrate the usefulness of the system using real-world data from a past Atlantic hurricane.


Modeling of Antibiotic Distribution in Response to Anthrax Attack

Adam Montjoy (University of Maryland), Jeffrey Herrmann (University of Maryland)

The release of anthrax spores as part of a bioterrorist attack into a highly-populated area will require a quick and efficient response from federal, state, and local public health officials to reduce illness and death. Deaths from anthrax can be prevented by the timely consumption of antibiotics. The distribution of antibiotics will occur at the county or state level. Public health emergency preparedness planners will need to organize various aspects of distribution including preparing Points of Dispensing (PODs) and routing vehicles from a central depot to deliver the antibiotics to the PODs. Timely delivery is important in reducing the likelihood that PODs run out of medication, disrupting operation. This problem is formulated as a capacitated vehicle routing problem over a short time frame. Deliveries to PODs will likely occur while dispensing of medication to exposed persons has already begun. Thus, planners seek to make even deliveries to all PODs with respect to time and quantity based on expected demand. This goal is captured in the objective function rather than having a traditional shortest time or least cost scheme. A solution to this problem is a schedule specifying a route for each vehicle, starting times for each delivery, and a quantity to deliver to each POD. Techniques explored include heuristics that separate the problem into routing and scheduling and an adaptive large neighborhood search for finding routes, which provides better schedules than routing by heuristic only. The problem is also formulated as a mixed integer program and solved with a column generation approach that attempts to solve the routing and scheduling simultaneously. The goal of this project includes implementing the formulation and techniques into freely-available software for public health emergency preparedness planners. This research is funded by the Montgomery County, Maryland, Advanced Practice Center for Public Health Emergency Preparedness and Response.


Motivation for Health Information Exchanges: A Patient Crossover Study

Jacqueline Griffin (Georgia Institute of Technology), Hannah Smalley (Georgia Institute of Technology), Pinar Keskinocak (Georgia Institute of Technology), David Laborde (Emory University School of Medicine), George Mathew (Emory University School of Medicine)

To demonstrate the importance of a Health Information Exchange (HIE) between Atlanta area hospitals, we examine the rate of crossover among neurosurgical inpatients treated at Emory University Hospital (EUH) and Grady Memorial Hospital (GMH). We also identify the impact of diagnoses on crossover rates and movement patterns to determine where initial investments in HIE should be made. Using electronic medical record data from EUH and GMH, unique patients who visited both hospitals were identified through classification by name and age at time of visit. A study of the frequency of flow patterns by crossover patients, including time between visits, was conducted. The significant crossover, especially for patients with specific diagnoses demonstrate the importance of implementing a health information exchange between the two hospitals studied. An HIE could prevent duplicate testing and has the potential for improving patient care.


Patient Allocation Problem during Pandemic Influenza Outbreak

Li Sun (University of Louisville), Gail DePuy (University of Louisville)


Policies for Blood Allocation in Developing Countries

Melih Celik (Georgia Institute of Technology), Ozlem Ergun (Georgia Institute of Technology), Mallory Soldner (Georgia Institute of Technology), Julie Swann (Georgia Institute of Technology)

Blood is a scarce resource, and it is especially so for developing countries. Adding to this the fact that it is also a perishable product, it becomes important to effectively allocate blood units. In this study, we consider the allocation of blood units to hospitals from a single collection center in a developing country. Demand is classified under two types: urgent and regular. Hospitals apply different policies to satisfy the incoming demand, and the collection center, observing the policies at each hospital, has to allocate the on-hand units before demand is actually observed. We first characterize the optimal allocation under two different usage policies: prioritization of urgent demand and first-come, first-served. We derive bounds on the relative performance of the two policies. We also consider the cases where allocation assumes a certain usage policy, but the actual policy is different than assumed, and find out the cost of not incorporating behavior. Furthermore, we assume a centralized system where demand is handled by a single unit, and derive how much improvement can be achieved over the decentralized system. Our tools are tested on a realistic example based on the US vaccination campaign in 2009.


Prepositioning Supplies for Improving Efficiency Response under Predictable Natural Disaster Settings

Gina Galindo Pacheco (SUNY at Buffalo), Rajan Batta (SUNY at Buffalo)

In a natural disaster, demand for essential items like food, medicine and water arises from affected areas. For certain types of natural disasters, like hurricanes, it is possible to plan for prepositioning of supplies so as to improve the efficiency of the post-disaster relief effort. In this work, a preliminary model for prepositioning supplies in such a setting is developed. Our model includes a facility location analysis, since supply points are to be selected from a set of candidate nodes. Then, relief units to be prepositioned, will be sent from a Main Distribution Center (MDC) to the selected supply points in order to be used to satisfy promptly and efficiently the resulting demand once the disaster has occurred. Two decision are then to be made: the location of supply points and the storage level at selected places. The objective pursued is to minimize the total expected logistic cost. We consider the following components of the cost: the total distribution cost, the fixed cost of opening supply points and the cost associated with losing units at destroyed supply points. Supply units stored at destroyed supply points become useless during the emergency. Then if supply point i is destroyed, all the units planned to be sent from such a supply point are thought to be delivered from the main distribution center. Uncertainty related to forecasted demand is considered. A constraint limiting the maximum expected number of units allowed to get destroyed is included. A solution for a hypothetical case study is offered along with a cluster approach for larger problems. Then we will present our ongoing work related to an improved scenario-based model.


Priority Dispatching Strategies for EMS Systems

Damitha Bandara (Clemson University), Maria Mayorga (Clemson University)


Quantitative Models for Humanitarian Logistics

Vitoriano (Complutense University of Madrid), M. Teresa (Complutense University of Madrid), Gregorio Tirado (Complutense University of Madrid), J. Tinguaro (Complutense University of Madrid), Javier Montero (Complutense University of Madrid)


Real-Time Decision Support System for Healthcare and Public Health Sectors Protection - Gaps Identified in HPH and ESS Sectors and the Proposed Research Work

Aman Gupta (University of Louisville), Sunderesh S. Heragu (University of Louisville), Trivikram Rao (University of Louisville), Robert Kelley (University of Louisville)


Reducing the Environmental Impact of Public Health Supply Chains

Michael F. C. Moreland (SEEDR L3C), Victoria M. Gammino (Global Immunization Division, National Center for Immunization and Respiratory Diseases, US Centers for Disease Control and Prevention), Sue Gerber (Global Immunization Division, National Center for Immunization and Respiratory Diseases, US Centers for Disease Control and Prevention)


Reengineering Transport Containers for an Improved Vaccine Cold Chain

Victoria M. Gammino (Global Immunization Division, US Centers for Disease Control and Prevention), Michael F. C. Moreland (SEEDR L3C), Olen Kew (Division of Viral Diseases, US Centers for Disease Control and Prevention), Sue Gerber (Global Immunization Division, US Centers for Disease Control and Prevention)


Resource Allocation Problems during Disasters: The Cases of Points of Distribution Planning and Material Convergence Handling

Miguel Jaller Martelo (Rensselaer Polytechnic Institute), (Rensselaer Polytechnic Institute)

As recent disaster experiences have shown, there is an urgent need for significant improvements in the efficiency of humanitarian logistics. These improvements are not only to avoid logistical failures ensuring an efficient and reliable flow of critical resources to the disaster area, but to guarantee a delivery process that takes into account economic (i.e. transportation, inventory, location, material convergence costs) and social considerations (i.e. human suffering). This research attempts to contribute to the improvement efforts by developing analytical formulations to analyze the resource allocation problem in the planning of distribution systems from two different perspectives. On one hand, considering the specific planning of points of distribution (PODs) in terms of their locations and resources required to expedite the flow of supplies to the population in need, while minimizing the social impacts brought about by their serving capacity. Specifically, the formulation developed provides an indication of the optimal number of PODs considering the cost associated with the walking distance to the POD and the social cost of waiting time for service. Results show the importance of considering waiting time or deprivation time social costs, usually overlooked by disaster planning formulations found in the literature.

On the other hand, the research developed strategies to optimally allocate resources to handle the material convergence problem; that is, to maximize the flow of high priority goods. The analyses performed provide insights about the negative impacts it has in humanitarian logistics: great amount o resources need to be allocated to handle this flow (composed of different types of commodities, ranging from urgent high priority to non-priority items); if no control strategy is developed, all components reach the impacted area creating logistical nightmares. With this in mind, a control/processing strategy is developed and sensitivity analyses are performed to different parameters.


Robust Decision Making with Limited Information: An Application of Info-Gap Theory

Selina Begum (Clemson University), William Ferrell (Clemson University)

Humanitarian logisticians face challenges of unknown demand and supply due to characteristics inherent to any disaster; time, location and impact, damaged infrastructure, nature of funding, politically volatile environment of the region and the role various stakeholders play. Decision making in such a dynamic environment is a challenge; unfortunately logisticians are often ill prepared to confront this challenge. Decisions driven by reliable information and quantitative model will yield better result. This research applies information gap decision making frame work to help decision makers to conduct relief work during the immediate response phase of a disaster relief work. Building on three constructs- system model, uncertainty model and performance requirement, info gap theory focuses on quantification of information gap to predict the system behavior. Info-gap theory revolves around what is already known and what happens when these known parameters are varied, how the solution space behaves under the circumstance. Using a simple case study we show how info gap theory help make a robust decision on the face of severe uncertainty.


S2H: Monitoring, Validating and Analyzing Homeless Shelter Occupancy

Supraja Narasimhan (Georgia Institute of Technology), Santosh S. Vempala (Georgia Institute of Technology)


SAFE Water Now: Scaling up access to sustainable safe water solutions

Swetha Krishnakumar (SAFE Water Now), Ashley Cleveland (SAFE Water Now), Tracy Hawkins (SAFE Water Now)


SLiCE Applied to Water Treatment Systems

Molly Nelson (GTRI), Emily Woods (GTRI), Joseph Goodman (GTRI), Kevin Caravati (GTRI), Laura Kovalchick (GTRI)


The Spatial Distribution of Aid Recipients in Kenya

Michael Veatch (Gordon College), Matthew Forsstrom (Gordon College), Hang Yang (Gordon College)


Vehicle Routing for the Last Mile of Power System Restoration

Carleton Coffrin (Brown University), Pascal Van Hentenryck (Brown University), Russell Bent (Los Alamos National Laboratory)

This work considers last-mile disaster recovery for power restoration, that is, how to schedule and route a fleet of repair crews to restore the power network as fast as possible after a disaster. To overcome the computational difficulties raised by this joint repair and restoration problem, this work proposes a four-stage approach based on the idea of constraint injection, which decouples the power-restoration and vehicle-routing optimization problems, while still capturing the restoration aspect in the routing component. The practical benefits of this approach are demonstrated on hurricane disaster scenarios generated by Los Alamos National Laboratory using state-of-the-art disaster simulation tools and the infrastructure of the United States.