Over the last three decades, disasters worldwide claimed more than 3 million lives and adversely affected the lives of at least 1 billion people (Noji, 1997). Regarding the threats posed by these disasters, emergency disaster management has emerged as a vital tool to reduce the harm and alleviate the suffering these disasters can cause to their victims. A significant task of planners involved in emergency disaster management is planning for and satisfying the vital needs of the people located in emergency shelters such as the Superdome in New Orleans.This thesis proposes a novel and comprehensive framework for the development of a humanitarian emergency inventory management system based on the real-time tracking of emergency supplies and demands through the integration of emerging technologies such as Radio Frequency Identification Devices (RFID) for commodity tracking and logistics. The novelty of this thesis is that, for the first time in the emergency inventory management field, the proposed approach combines an offline planning strategy with online control techniques in a unified framework. Within this framework, the offline planning problem is solved by the stochastic humanitarian inventory iii management approach, whereas the online modeling strategies include the application of neural network-based functional approximation, simultaneous perturbation stochastic approximation (SPSA), and continuous time model predictive control (CMPC) techniques.Unlike previous studies, the flexibility of the proposed inventory management and control model allows the application of the developed mathematical model to extreme events making online real-time tracking possible. Realistic case studies built using information available from past disasters are used to examine the differences in inventory strategies for different types of disasters based on the impact area and duration of the extreme event. The proposed methodology is also capable of representing and understanding real-life cases where uncertainty and limitations on the inventory levels and flow of supplies can be modeled by introducing different levels of stochasticity and real-life constraints.The overall findings of this thesis have pointed out that the proposed integrated framework can be efficiently used for emergency inventory planning and inventory control during disaster relief operations without ignoring the real-world uncertainties, fluctuations, and constraints of disaster conditions. iv

No abstract

The impacts of disasters have recently attracted increased attention from researchers and policy makers. However, there has been little consensus about how an efficient inventory management model can be developed for postdisaster conditions. Victims of a disaster are generally gathered into shelters during and after a severe disaster to ensure their security. Many evacuees do not have the financial resources to leave the disaster area or to find food, drugs, and other necessities. Hence, their vital needs should be supplied efficiently throughout the disaster and postdisaster periods. Without an adequate stock of goods, satisfying the daily requirements of the victims without disruption might be problematic. To solve this problem, humanitarian inventory control models that can aid in adequately responding to a disaster or a humanitarian crisis are needed. In this context, response represents preparedness, planning, assessment, appeal, mobilization, procurement, transportation, warehousing, and distribution. This paper is concerned with the development of a subproblem of the general humanitarian supply chain problem: an efficient and quick-response humanitarian inventory management model able to determine the safety stock that will prevent disruptions at a minimal cost. The humanitarian inventory management problem is first mathematically formulated as a version of the Hungarian Inventory Control Model. A solution to this time-dependent stochastic model is then proposed by using the p-level efficient points algorithm. The single commodity case results are given, and a sensitivity analysis of the model vis-à-vis various model parameters that affect safe inventory levels is conducted.

No abstract

Healthcare resource availability is potentially associated with COVID-19 mortality, and the potentially uneven geographical distribution of resources is a looming concern in the global pandemic. Given that access to healthcare resources is important to overall population health, assessing COVID-19 patients' access to healthcare resources is needed. This paper aims to examine the temporal variations in the spatial accessibility of U.S. COVID-19 patients to medical facilities, identify areas that are likely to be overwhelmed by the COVID-19 pandemic, and explore associations of low access areas with their socioeconomic and demographic characteristics. We use a three-step floating catchment area method, spatial statistics, and logistic regression to achieve the goals. Findings of this research in the State of Florida revealed that North Florida, rural areas, and zip codes with more Latino or Hispanic populations are more likely to have lower access than other regions during the COVID-19 pandemic. Our approach can help policymakers identify potentially possible low access areas and establish appropriate policy intervention paying attention to those areas during a pandemic.