The overall goal of this project is to develop national and local data sources that allow us to model and mitigate the unintended impact of law enforcement efforts to disrupt the illicit drug supply network. Our analysis will look both at county and neighborhood level effects of the illicit supply network and distinguish across multiple illicit substances (e.g., fentanyl, heroin, methamphetamine, cocaine, benzodiazepines, prescription medications, and cannabis). We will study the effect of network disruptions on the transition to more potent substances and subsequent changes in rates of drug overdoses.

Legitimate massage businesses across the United States use online platforms, such as advertisement services, job recruitment ads, and review boards. Illicit massage businesses (IMBs), which are often associated with human trafficking activity, do so as well. To better triage, pivot through, and identify illicit businesses through online reviews, we will examine coexisting features present in business level, spatial, and socio-demographic data. This project will create a robust, diverse data-enrichment methodology applicable by cross-domain researchers in adversarial online environments through two primary methods: 1) developing data collection, cleaning, and joining methods and 2) utilizing external geospatial and socio-demographic data to prioritize and target features most likely present in illicit massage business reviews online.

Effective management of product transitions, the introduction of new products into volume manufacturing is a critical competitive advantage in high-technology industries.The literature has treated product transitions in isolation from each other and from products not involved in the transition. However, the frequent product transitions in firms with short product life cycles renders product transitions an ongoing, operational issue. The technical information required to implement product transitions resides in the product engineering (ENG) and manufacturing (MFG) units, while long-term corporate goals are the domain of corporate management (CORP), who sets targets for which product lines to prioritize to maximize long-term profit. In practice, ENG and MFG units must negotiate with CORP for financial resources, and with each other for shared resources. Specifically, ENG units require access to factory capacity to debug new designs, reducing revenue in the short term; but MFG must allocate ENG sufficient capacity for timely introduction of new products to ensure future revenue. The effective structuring and support of this decentralized decision process is important to all firms managing multiple products over short life cycles, and requires careful modelling of engineering issues. Intellectual Merit. In contrast to centralized approaches that formulate the problem as large-scale mathematical programs or stylized, aggregated models with little engineering content, we study two alternative game-theoretic formulations of the problem. In the first of these, bilevel programming, CORP takes the role of leader allocating financial resources to the MFG and ENG units, who must then optimize their performance under this budget. CORP must make its decisions considering the reactions of ENG and MFG units. Classical bilevel approaches impose a strict hierarchy among the DMs, with the decisions of the leader constraining those of independent followers. We extend this framework by considering interactions among the followers (the ENG and MFG units) providing novel formulations and solution approaches.

Human trafficking is the world������������������s fastest growing criminal economy and the third largest overall. Law enforcement, government agencies, nonprofit organizations, technology companies and commercial enterprises all have relevant data and skills that could drastically increase the number of victims identified and rescued as well provide evidence to secure traffickers������������������ convictions. In the digital age, human trafficking has evolved into a complex system that leverages the power of technology to build supply chains, create sophisticated communication techniques, and protect criminals from scrutiny. There are many forms of human trafficking, including sex trafficking and labor trafficking, among others. This proposal focuses on countering human trafficking at illegal massage businesses by enabling the identification of victims and traffickers via statistical learning.

The U.S. Centers for Disease Control and Prevention������������������s (CDC) ILINet surveillance system plays an important role in our understanding of influenza dynamics and is used to identify seasonal influenza disease burden, severity, epidemic onset and seasonality, but it suffers from reporting delays and limited, opportunistic sampling of the population. FluView for Fall 2020 shows relatively few visits for ILI. To date, approaches based on EHRs or medical claims have been found to be cumbersome for implementation on a dynamic, rolling horizon basis. Moreover, they usually require deep technical know-how to update when diagnosis codes change or new billing is added such as for telemedicine. We propose to use a data factory approach that provides dynamic, easy to update reporting, based on the Fast Healthcare Interoperability Resources (FHIR) standards for EHR. We will demonstrate the capabilities of the system using data from MedStar Health, a 10-hospital system serving Maryland, Virginia, and District of Columbia.

The COVID-19 pandemic has intensified interest in health systems������������������ ability to prepare for, and cope with, surges (sudden large-scale escalations) in the treatment needs of critically ill patients. Several studies in the literature and the US Department of Health and Human Services provide recommendations and standard operating procedures for intensive care unit (ICU) and hospital preparations for a mass disaster or pandemic. However, due to the rarity of the events and the lack of a mechanism to gather data in real-time, the implementation of these recommendations in a healthcare organization usually depends on the experience of hospital emergency managers rather than evidence-based approaches. Furthermore, the scale, velocity, and global impact of the COVID-19 pandemic is beyond what any plans had anticipated. In this proposal, we aim to document medical surge capacity preparations and real-time adjustments for the COVID-19 pandemic in two major hospital systems: (1) Prisma Health, the largest not-for-profit health organization in South Carolina, serving more than 1.2 million patients annually; and (2) MedStar, a not-for-profit healthcare organization operating more than 120 entities, including ten hospitals in the Baltimore������������������Washington, DC area.

Every year approximately 700,000 people die in US hospitals. In 16% of them, the first diagnosis at death was septicemia ������������������ one of the most common delayed diagnoses associated with inpatient death. Sepsis is one of the ten leading causes of death. While it is difficult to estimate how many of these deaths could have been averted or postponed if a better system of care was in place, it is widely recognized that sepsis patients have better outcomes if treated earlier. Sepsis is one of the most common of these diagnoses with delayed effective treatment interventions. As opposed to wrong diagnoses, delayed diagnoses have historically not been considered adverse events as there is no change in patient condition as a result of care delivered. However, patients with delayed diagnoses do have worse outcomes than those who receive timely treatment. These diagnostic and/or treatment delays associated with inpatient mortality and long term morbidity consequences represent a significant and modifiable patient safety issue. Awareness of sepsis is low; many septic patients are under-diagnosed at an early stage when aggressive treatment could still reverse the course of the infection. Early recognition and implementation of early goal directed therapy improves outcomes and decreases mortality. For every one hour delay in treatment of severe sepsis or severe shock with antibiotics, there is an incremental decrease in patient survival. For example, a delay in antibiotics of five hours decreases survival to 50%. We propose to take a date-driven evidence-based approach that integrates computer science and industrial engineering to develop personalized sepsis diagnosis and treatment plans. The goal of this research is to integrate electronic health records (EHR) and clinical expertise to provide an evidence-based framework for diagnosing sepsis patients, accurately risk-stratifying patients within the sepsis spectrum, and developing intervention policies that inform sepsis treatment decisions. We will achieve this research goal through accomplishing three specific aims based on a longitudinal data set of EHRs of Mayo Clinic Rochester hospital patients discharged and Christiana Care Health System hospital patients.

The Southeastern Regional Medical Center (SRMC) in Lumberton, NC, would like to improve throughput of patients within the Emergency Department (ED). The goal is to have 60% of patients exit the ED within 180 minutes; the current performance is that 40% of patients do so. Ultimately, measures such as this one drive patient satisfaction with the hospital, and thus the measure is a priority of the hospital leadership. Based on benchmarking and discussions with other hospitals, SRMC has asked for analytics and systems engineering approaches to assist in meeting their performance goal. The ISE team proposes engineering analytics and modeling to assist in understanding what actions will improve patient throughput in the ED to meet the performance goal of the hospital.

Many dynamic decision problems involving uncertainty can be appropriately modeled as multi-stage stochastic programs. Examples include problems in capacity planning, financial management, production planning, transportation and logistics, and healthcare management. Despite their appeal as a natural modeling framework, stochastic programs suffer exponential growth in their size with the number of decision stages. The size of such a model can easily exceed the available memory on a state-of-art desktop computer. To overcome this barrier, typical applications in practice content with reducing the size of the problem, and therefore, losing some information. Even then, the modern algorithms employed in solving such optimization problems can hit the memory barrier during the solution process, or the solution times can be excessive. Therefore, practical applications have generally been restricted to two-stage linear models with limited modeling of parameter uncertainty. Two-stage models, however, lack an ability to dynamically respond to information/uncertainties that become available to the decision makers in stages. The objective of the proposed research involves the development, evaluation, and implementation of a comprehensive methodology for solving large-scale multi-stage stochastic mixed-integer programs using a distributed computing environment

Every year approximately 700,000 people die in US hospitals. In 16% of them, the first diagnosis at death was septicemia ������������������ one of the most common delayed diagnoses associated with inpatient death. Sepsis is one of the ten leading causes of death. While it is difficult to estimate how many of these deaths could have been averted or postponed if a better system of care was in place, it is widely recognized that sepsis patients have better outcomes if treated earlier. We propose to take a date-driven evidence-based approach that integrates computer science and industrial engineering to develop personalized sepsis diagnosis and treatment plans. The goal of this research is to integrate electronic health records (EHR) and clinical expertise to provide an evidence-based framework for diagnosing sepsis patients, accurately risk-stratifying patients within the sepsis spectrum, and developing intervention policies that inform sepsis treatment decisions. We will achieve this research goal through accomplishing three specific aims based on a longitudinal data set of EHRs of Mayo Clinic Rochester hospital patients discharged and Christiana Care Health System hospital patients. The project will provide graduate students and health services researchers with cross-disciplinary educational experience. This is formalized via the rotational internships of engineering and computer science students that immerse them into the healthcare systems.