<p> Continuing population growth and increased urbanization within disaster-prone areas have led to greater numbers of mass casualties and economic losses caused by natural or human-made disasters. Efficient decision-making processes are crucial in all phases of a disaster life cycle, from mitigation and preparedness to response and recovery. The overarching goal of this dissertation is to contribute to region-wide disaster operation management capabilities by creating a set of tools to facilitate fast, life-saving decision-making. The dissertation begins with initial first responders&rsquo; assignments to affected structures and spans health care and infrastructure preparation and response. In mass casualty incident (MCI) circumstances in particular, situations are complicated, networks are often large, and conditions are transient and time-dependent. Thus, models developed in this thesis evaluate and update decisions based on available information at each point in time to the system. </p><p> The functioning of various response networks, whether in the disaster scene or at the health care facilities, is conceptualized mathematically. Each model can be viewed as a type of queueing network in which MCI victims are customers and responders or health care facilities are servers. Each queueing network is employed to: (1) test developed protocols, acting as queueing system operational policies to support disaster response, (2) assess tactics developed otherwise, or (3) optimize regional resiliency of the health care system given its dependence on set of interdependent supporting lifelines in disasters through preparedness and response actions. Resource-constrained patient flow models of hospitals are presented for routine and emergency operations for the purpose of the study. Using queueing network conceptualizations, discrete event simulation and simulation-based optimization techniques are developed to propose and evaluate protocols that guide responses and for assessing performance and resilience of these systems.</p><p>

<p> In the U.S. medical sector, software failures in safety-critical systems in healthcare have led to serious adverse health problems, including patient deaths and recalls of medical systems. Despite the efforts in developing techniques to build resilient systems, there is a lack of consensus regarding the definition of resilience metrics and a limited number of quantitative analysis approaches. In addition, there is insufficient guidance on evaluating resilience design patterns and the value they can bring to safety-critical systems. </p><p> This research employed the interdependency analysis framework to evaluate the static resilience of safety-critical systems used in the healthcare field and identified software subsystems that are vulnerable to failures. Resilience design patterns were first implemented to these subsystems to improve their ability to withstand failures. This implementation was followed by an evaluation to determine the overall impacts on system&rsquo;s static resilience. </p><p> The methodology used a common medical system structure that collects common attributes from various medical devices and reflects major functionalities offered by multiple medical systems. Fault tree analysis and Bayesian analysis were used to evaluate the static resilience aspects of medical safety-critical systems, and two design patterns were evaluated within the praxis context: <i> Monitoring</i> and <i>N-modular redundancy</i> resilience patterns. </p><p> The results ultimately showed that resilience design patterns improve the static resilience of safety-critical systems significantly. While this research suggests the importance of resilience design patterns, this study was limited to explore the impact of structural resilience patterns on static resilience. Thus, to evaluate the overall resilience of the system, more research is needed to evaluate dynamic resilience in addition to studying the impact of different types of resilience design patterns. </p><p>