Energy: Policies to Support Renewable Energy and Carbon Emissions Control
Faculty: Tapas K. Das
Economic models are needed to analyze the impact of policies adopted for controlling carbon emissions and increasing distributed renewable generation in microgrids (green penetration). The impacts are manifested in performance measures like emissions, electricity prices, and demand for electricity. We develop economic models comprising bi-level optimization and Pareto analysis. In the bi-level framework, the upper level models the operation of the microgrids and the lower level deals with electricity dispatch in the grid. The model yields guidelines for setting policies for emissions reduction and green penetration without adversely impacting electricity prices and demand.
Federal, state, and local governments and the utility companies offer various incentive programs promoting net zero energy buildings (NZEB), a key carbon emissions reduction strategy adopted in the U.S. and in the EU countries. Via mixed integer programming (MIP) models, we examine the threshold levels of these incentives that would spur growth in NZEBs. The models consider building type, RE technology, desired return on investment, weather, time of use pricing (TOU), and the portfolio of all incentives. The thresholds are influenced by the levelized cost of electricity from RE.
NZEB Schematic A Pareto front for ER,GP,and MD
Healthcare Systems: Pandemic Influenza Outbreak Prevention and Mitigation
Faculty: Tapas K. Das & Alex Savachkin
The threat of a highly pathogenic pandemic influenza continues to prove its existence through sporadic occurrences of human infections and deaths around the world. The United Nations System Influenza Coordination (UNSIC) team in collaboration with The World Bank and the governments and industry around the world have remained active in supporting the pandemic mitigation priorities: capacity for prevention, surveillance and testing, and impact mitigation. Our research is aimed at prevention and mitigation of pandemic influenza outbreaks using both pharmaceutical and non-pharmaceutical intervention (PHI and NPI) strategies. We use an agent-based simulation modeling approach that incorporates detailed population demographics and dynamics, variety of mixing groups and their contact processes, infection transmission process, and pharmaceutical and non-pharmaceutical interventions. We use both tools of optimization and statistical designed experiments to seek best strategies for use of limited pharmaceutical and non-pharmaceutical intervention resources with goals to 1) prevent an influenza outbreak to become a pandemic and 2) mitigate the impact of pandemic influenza by reducing the infection attack rate.
Influenza Disease Natural History
Healthcare: Predicting Presence of BRCA Mutations
Faculty: Tapas K. Das
Accurate prediction of the presence of BRCA mutations is critically important as these significantly increase the probability of developing breast and/or ovarian cancers. Existing Mendelian and empirical prediction models do not possess a satisfactory level of predictive power. Scientists believe that the low power is partly due to the lack of availability and use of diverse and comprehensive data on medical and family history of cancer. A recent study (ABOUT) attempted to address the issue of diversity and availability of data by collecting personal medical history and family history of cancer for participant from a wide population across the U.S. through a survey in collaboration with the third largest health insurer. We examine the power of predicting BRCA mutations using machine learning models for classification. Though machine learning models have received much attention in medical decision making in recent years, they have not been applied to BRCA prediction. Using survey data from the ABOUT study, we compare performance of some of the most effective machine learning methods with those from the Mendalian and empirical models, which are widely used by the genetic consultants.
ROC Curves for Predictive Models
Transportation Systems: Hazardous Materials Transportation Regulation
Faculty: Changhyun Kwon
It is desirable to avoid extremely high hazmat accident consequences at any cost regardless of how the risk is measured. In hazmat transportation, a risk scenario corresponds to each arc, and specific scenarios can be completely avoided by not traveling the corresponding road arcs. This avoidance can be done by prohibiting hazmat trucks from traveling through those arcs. From this perspective, governments often enforces interventions to control hazmat travel. Curfews, which are time-based bans on road segments, is its first kind. To pose curfews as an optimization problem we allow a percentage of the road segments to be impacted for certain time periods. The goal, then, is to determine which road segments should be selected for curfew imposition at what specific time periods, so that the solution of the multi-trip hazmat problem, in which each hazmat shipment is individually optimized by the shippers, is also one that satisfies an acceptable level of equity. The second intervention method of our interest is that of using road bans. This is an extreme case of a curfew, in which a percentage of road segments are shut off to hazmat traffic at all times.
A more sophisticated method of controlling hazmat traffic flow is to guide the behavior of hazmat shippers through the use of tolls. The fundamental premise of this approach is that shippers will optimize a certain combination of the travel cost, toll cost, and risk. Thus, setting tolls judiciously on road segments can create the desired flow of hazmat traffic. It is possible that a different set of toll is imposed for regular traffic; we call such a toll system a dual-toll system. A dual-toll system may create a safer hazmat network flow than a single-toll system. The dual-toll project has been supported by NSF.
Transportation Systems: Risk Measures for Hazardous Materials Routing
Faculty: Changhyun Kwon
When we determine routes for transporting hazardous materials (hazmat), we must consider risk of accidents and hazmat release. Various measures of such risk have been suggested and used; but many of them are inappropriate for risk-averse routing and lack flexibility. In this focus area, we seek development of risk-averse, flexible, and robust routing methods based on advanced risk measures such as Value-at-Risk (VaR), Conditional Value-at-Risk (CVaR), and Spectral Risk Measures.
Healthcare: Image-based Modeling and Analysis for Medical Diagnosis
Faculty: Susana Lai-Yuen
Medical imaging has become important in assessing medical cases that may not be evident on clinical examination. Given the growing number of these images, computer algorithms for the automated identification and analysis of quantitative information from the images are increasingly needed for medical diagnosis and treatment planning. We are working on the design of image-based and computational geometry models to automatically identify anatomical features and obtain useful quantitative data from medical images. We are also working on data mining techniques for the analysis of image-based and clinical data for decision support in medical diagnosis. Current application is on the automated identification and analysis of pelvimetry features on magnetic resonance imaging (MRI) for the evaluation of pelvic organ prolapse. The ability to automate the process of data identification and analysis on radiologic studies will enable the use of imaging to predict the development of diseases in genetically predisposed patients, and potentially lead towards preventive strategies.
Analytics: Information Fusion-based System Informatics
Faculty: Mingyang Li
With the advancement of sensing technology and information storing systems, a data-rich environment has been created in many complex systems with multiple data sources available at different levels. For instance, in the reliability assessment of a typical hierarchical engineering system, multi-source reliability data (e.g., success/failure data, failure-time data, degradation data, etc.) is available at different levels (e.g., system level, component level, etc.). In the modern crowd surveillance system equipped with unmanned ground and aerial vehicles, the multi-fidelity crowd sensing data is continuously generated. To fully utilize and integrate such multi-source data for improving the system performance is a challenging task. This line of research is to develop a generic, coherent, flexible and scalable data fusion framework to improve the system performance. Some successful outcomes include the improvement of system reliability prediction accuracy and precision, the cost reduction of system reliability demonstration tests and the improvement of crowd tracking performance.
Network Optimization: Innovative Networks
Faculty: Kingsley Reeves
We investigate the impact of an organization's collaboration networking decisions on its innovation performance. As industrial innovation requires a novel recombination or reconfiguration of technological knowledge, an innovative organization seeks to maximize its opportunities using collaborations as conduits for inflow of technological knowledge. Collaboration can take the form of an alliance, a supplier-customer relationship, common ownership or social connections among organizations. Research shows that collaborations are increasingly necessary, yet the challenge is determine with whom and how to collaborate so that innovation is increased.
Our survey of organizations in the high-tech industry in Florida (red nodes in the picture) revealed the collaboration network map of high-tech companies and universities. The purpose was to test the hypothesis that being central in a collaboration network results in higher innovative performance (as measured by the number of patents issued). Results suggest that an organization does not necessarily have to be central in its surrounding network but has to be in collaboration with other central actors to benefit from the positive impact of its placement. The results also indicate that as opposed to the other collaboration ties, customer and alliance connections have a stronger positive impact on new product and service development.
Healthcare: Sensor-Based, Response-Oriented Technology for In-Home Monitoring of Senior Health & Well Being
Faculty: Ali Yalcin and Carla VanderWeerd
Ninety one percent of seniors live in individual residences under the care of family caregivers without requisite monitoring of their daily activities. The number of seniors whose daily activities need monitoring will grow as our nation ages and the life expectancy of individuals increases. AlwaysNear is a wireless sensor system that provides comprehensive 24/7 monitoring of daily living activities of seniors that live in individual residences. It detects and issues alerts in response to deviations in daily activities. The alerts are responded to in a number of different ways including calls to the individual and, when warranted, the deployment of emergency responders. The goal of this project is to validate the AlwaysNear home monitoring sensory system's functionality, assess the alert response system's performance, and lay the ground work for an extended health outcomes study via a multi-phase evaluation.