The purpose of this dissertation intends to discover as well as categorize safety concern reports in online reviews by using key terms prevalent in sub-categories of safety concerns. This dissertation extends the literature of semi-automatic text classification methodology in monitoring and classifying product quality and service concerns. We develop various text classification methods for finding key concerns across a diverse set of product and service categories. Additionally, we generalize our results by testing the performance of our methodologies on online reviews collected from two different data sources (Amazon product reviews and Facebook hospital service reviews). Stakeholders such as product designers and safety regulators can use the semi-automatic classification procedure to subcategorize safety concerns by injury type and narrative type (Chapter 1). We enhance the text classification approach by proposing a Risk Assessment Model for quality management (QM) professionals, safety regulators, and product designers to allow them to estimate overall risk level of specific products by analyzing consumer-generated content in online reviews (Chapter 2). Monitoring and prioritizing the hazard risk levels of products will help the stakeholders to make appropriate actions on mitigating the risk of product safety. Lastly, the text classification approach discovers and ranks aspects of services that predict overall user satisfaction (Chapter 3). The key service terms are beneficial for healthcare providers to rapidly trace specific service concerns for improving the hospital services. / Doctor of Philosophy / This dissertation extends past studies by examining safety surveillance of online reviews. We examine online reviews reporting specific categories of safety concerns and contrast them with reviews not reporting these specific safety concerns. Business and regulators are benefited in detecting, categorizing, and prioritizing safety concerns across product categories. We use key terms prevalent in domain-related safety concerns for granular analysis of consumer reviews. Secondly, beyond utilizing the key terms to discover specific hazard incidents, safety regulators and manufacturers may use the extended risk assessment framework to estimate the risk severity, risk likelihood, and overall risk level of a specific product. The model could be useful for product safety practitioners in product risk identification and mitigation. Finally, this dissertation identifies the aspects of service quality concerns present in online hospital reviews. This study uses text analytics method by using key terms to detect these specific service concerns and hence determine primary rationales for patient feedback on hospital services. Managerially, this information helps to prioritize the areas in greatest need of improvement of hospital services. Additionally, generating key terms for a particular service attribute aids health care policy makers and providers in rapidly monitoring specific concerns and adjusting policies or resources to better serve patient

Electrical cabinet fires occur frequently in commercial and industrial facilities. However, the severity of these fire events varies widely, making it difficult to estimate the fire growth and size with certainty. This study aims to identify the significant parameters that affect electrical cabinet fires, which are quantified as the heat release rate (HRR), and properly categorize them. With this knowledge, optimal parameter-response relationships can be developed to predict the electrical cabinet fire behavior. Statistical analysis conducted in this study on historical fire incident data revealed that the fires in Nuclear Power Plants (NPP) were primarily associated with electrical cabinets. The database used in this research was an electronic version of the publicly available Updated Fire Event Database developed by Electric Power Research Institute, including 2,111 fire events. 540 of these events were labeled as being challenging fires with 74.2% of these challenging fire events being due to eleven selected fire types. Electrical cabinets were found to represent a majority (40.7%) of all the challenging fire events. Although historically conducted electrical cabinet fire experiments sought to explore the influence of parameters on HRR, the parameters were not systematically varied to statistically quantify which parameters were most important/relevant. Research in this study used statistical analysis on a series of simulation results on electrical cabinet fires from the computational fluid dynamics code Fire Dynamic Simulator (FDS). Simulation matrices were developed and evaluated using fractional factorial Design of Experiments (DOE) to screen the importance of different parameters on the electric cabinet HRR. Based on statistical analysis of the results, the combustible material surface area was found to be the most significant parameter followed by cabinet volume, combustible configuration, burning duration, and combustible material heat release rate per unit area. Material ignition temperature was found to not be statistically significant. The last phase of this research assessed the robustness of the electrical cabinet parameters on the predicted HRR with more detailed simulations. Two investigations were undertaken. To identify the nonlinear effects of parameters on the electrical cabinet fire HRR, a Response Surface Methodology (RSM) based Central Composite Design (CCD) was used to create a simulation matrix that would allow statistical analysis of important parameters as well as their effects on the fire heat release rate while keeping the combustible configuration inside the cabinet constant. A series of simulations were conducted to explore the impact of combustible configuration and ignition source location while keeping all other variables consistent. The analysis revealed that all variables had a statistically significant effect on peak HRR. For the average HRR, both the ventilation area into the cabinet and the ignition source HRR were found to be statistically insignificant. For both output variables, the cabinet volume, material heat release rate per unit area, and material surface area were the most significant parameters. Combustible configuration and ignition source location were also found to be statistically significant. / Master of Science / Electrical cabinet fires are a significant concern for industries, commercial electric plants, telecommunication buildings, and nuclear power plant (NPP) facilities. These cabinets typically represent a metallic enclosure of varying sizes. Additionally, several different electronic components of heterogenous composition and configuration are included within this cabinet. As a result, the fires within the cabinet can propagate to several other nearby components, resulting in large fires that are difficult to suppress. Thus, it becomes necessary to understand the fire behavior of electrical cabinets and the factors influencing fire propagation. Knowing the factors influencing the electrical cabinet fires will enable facilities to have better fire resilience and further prevent multiple components and structures from being damaged by these fires. Statistical analysis of historic fire events validated that the most frequently challenging fires in NPP involve electrical cabinets.Therefore, aA detailed study was conducted to investigate what parameters most significantly affect the size of the electrical cabinet fire, which is quantified as the heat release rate (HRR). The parameters in the study included cabinet volume, ventilation area, combustible fuel detail (ignition temperature, heat release rate per unit area (HRRPUA), burning duration), fuel configuration inside the cabinet, and size of the ignition source. To determine which of these factors significantly impacted the electrical cabinet HRR, a computational fluid dynamics code Fire Dynamic Simulator (FDS), was used to predict the fire growth of electrical cabinet fires. After employing a rigorous statistical analysis of the FDS results, the combustible material surface area was found to be the most significant parameter, followed by cabinet volume, combustible configuration, burning duration, and flammable material HRRPUA. The last phase of the research sought to explore the significance of the parameters while developing a nonlinear expression to predict the fire HRR based on cabinet parameters. Given the wide range of electrical cabinet parameters, especially combustible configuration, two studies were conducted where the configuration was fixed or varying with respect to other parameters. For fixed combustible configuration, simulations were conducted with FDS systematically varying the other parameters so their importance could be ranked. Simulations were also performed with all parameters fixed except the combustible configuration and ignition source location. The analysis revealed that all variables had a statistically significant impact on peak HRR. For the average HRR, both the ventilation area into the cabinet and the ignition source HRR were found to be statistically insignificant. For both output variables, the cabinet volume, material heat release rate per unit area, and material surface area were found to be the most significant parameters. Combustible configuration and ignition source location were also found to be statistically significant.

京都大学 / 新制・論文博士 / 博士(工学) / 乙第13616号 / 論工博第4213号 / 新制||工||2002(附属図書館) / (主査)教授 高木 郁二, 教授 佐々木 隆之, 教授 米田 稔 / 学位規則第4条第2項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Risk indicators are metrics that are widely used in risk management to indicate how risky an activity is. Among different types of risk indicators, early warning systems are designed to help decision makers predict and be prepared for catastrophic events. Especially, in complex systems where outcomes are often difficult to predict, early warnings can help decision makers manage possible risks and take a proactive approach. Early prediction of catastrophic events and outcomes are at the heart of risk management, and help decision makers take appropriate actions in order to mitigate possible effects of such events. For example, physicians would like to prevent any adverse events for their patients and like to use all pieces of information that help accurate early diagnosis and interventions. In this research, first we study risk assessment for occupational injuries using accident severity grade as an early warning indicator. We develop a new severity scoring system which considers multiple injury severity factors, and can be used as a part of a novel three-dimensional risk assessment matrix which includes an incident's severity, frequency, and preventability. Then we study the predictability of health outcome based on early risk indicators. A systems model of patient health outcomes and hospital length of stay is presented based on initial health risk and physician assessment of risk. The model elaborates on the interdependent effects of hospital service and a physician's subjective risk assessment on length of stay and mortality. Finally, we extend our research to study the predictive power of early warning systems and prognostic risk indicators in predicting different outcomes in health such as mortality, disease diagnosis, adverse outcomes, care intensity, and survival. This study provides a theoretical framework on why risk indicators can or cannot predict healthcare outcomes, and how better predictors can be designed. Overall, these three essays shed light on complexities of risk assessments, especially in health domain, and in the contexts where individuals continuously observe and react to the risk indicators. Furthermore, our multi-method research approach provides new insights into improving the design and use of the risk measures. / Ph. D.

There is no single methodology toward freshwater conservation planning, and few analytical tools exist for summarizing ecological risks at a landscape scale. I constructed a relative risk model, the Ecological Risk Index (ERI), to combine the frequency and severity of human-induced stressors with mappable land and water use data to evaluate impacts to five major biotic drivers: energy sources, physical habitat, flow regime, water quality, and biotic interactions. It assigns 3 final risk rankings based on a user-specified spatial grain. In a case study of the 5 major drainages within the upper Tennessee River basin (UTRB), U.S.A, differences in risk patterns among drainages reflected dominant land uses, such as mining and agriculture. A principal components analysis showed that localized, moderately severe threats accounted for most of the threat composition differences among watersheds. Also, the relative importance of threats is sensitive to the spatial grain of the analysis. An evaluation of the ERI procedures showed that the protocol is sensitive to how extent and severity of risk are defined, and threat frequency-class criteria strongly influenced final risk rankings. Multivariate analysis tested for model robustness and assessed the influence of expert judgment by comparing my original approach to a quantile-based approach. Results suggest that experts were less likely to assign catchments to high-risk categories than was the quantile approach, and that 3 final risk rankings were appropriate. I evaluated the influence of land use on freshwater ecosystems by studying the relationship between land cover changes and the persistence of freshwater mussels. First, historical species data were collected and the Upper Tennessee River Mussel Database (UTRMD) was constructed. The UTRMD contains >47,400 species records from 1963-2008 distributed across nearly 2,100 sampling sites. My study suggests that 30 years of land cover change does not explain observed freshwater mussel declines. Quantitative surveys are recommended basin-wide to provide more accurate information about mussel distribution and abundance. Lastly, results suggest that streams with repeated mussel surveys have increasing populations, including active recruitment in several beds. Additional quantitative surveys since 2004 have probably provided more accurate species and population counts, although actual population sizes are still uncertain. / Ph. D.

Quantitative population dynamics modeling is needed to evaluate the stock status and fisheries management plans to provide robust model and management strategies. Atlantic Weakfish (Cynoscion regalis), one important commercial and recreational fish species along the west coast of Atlantic Ocean that was found to be declining in recent years, was selected as an example species. My study aimed to explore the possible spatial heterogeneity of CPUE (catch per unit effort) year trend based on three fishery independent surveys and explore the influence of nonstationary natural mortality on the fisheries management through a MSE (Management Strategy Evaluation) algorithm based on the Weakfish stock assessment results. Five models for catch rate standardization were constructed based on the NEAMAP (NorthEast Area Monitoring and Assessment Program) survey data and the ‘best' two models were selected based on the ability to capture nonlinearity and spatial autocorrelation. The selected models were then used to fit the other two survey data to compare the CPUE year trend of Weakfish. Obvious differences in distribution pattern of Weakfish along latitude and longitude were detected from these three surveys as well as the CPUE year trend. To test the influence of the model selection on the MSE, five stock-recruitment models and two forms of statistical catch-at-age models were used to evaluate the fishery management strategies. The current biomass-based reference point tends to be high if the true population dynamics have nonstationary natural mortality. A flexible biomass based reference point to match the nonstationary process is recommended for future fisheries management. / Master of Science

High levels of fecal indicator bacteria (e.g. E. coli) are the leading cause of identified surface water impairments in the United States. The US Clean Water Act of 1972 requires that jurisdictions establish priority rankings for impaired waterways and develop a Total Maximum Daily Load (TMDL) plan for each. Although past research indicates that the risk of illness to humans varies by source of fecal contamination, current watershed assessments are developed according to total concentration of indicator bacteria, with all sources weighed equally. A stochastic model using Quantitative Microbial Risk assessment (QMRA) principles to translate source-specific (e.g. human, livestock) daily average concentrations of E.coli into a daily average risk of gastroenteritis infection was developed and applied to Pigg River, an impaired watershed in southern Virginia. Exposure was calculated by multiplying a ratio of source related reference pathogens to predicted concentrations of E.coli and a series of qualifying scalars. Risk of infection was then determined using appropriate dose response relationships. Overall, human and goose sources resulted in the greatest human health risk, despite larger overall E.coli loading associated with cattle. Bacterial load reductions specified in the Pigg River TMDL were applied using Hydrological Simulation Program- FORTRAN (HSPF) to assess the effect these reductions would have on the risk of infection attributed to each modeled bacterial source. Although individual risk sources (neglecting geese) were reduced below the EPA limit of 8 illnesses per 1000 exposures, the combined risk of illness varied between 0.006 and 64 illnesses per 1000 exposures. / Master of Science

This thesis presents a comparison of correlation methods in risk analysis. A theoretical solution is given to the correlation problem along with a discussion of each method. Each method is compared to a developed test case and two other cost projects. Restrictions on correlation coefficients are also given followed by the advantages and disadvantages of each method. / Master of Science

Freshwater mussels have played an important role in the history of human culture and also in ecosystem functioning. But during the past several decades, the abundance and diversity of mussel species has declined all over the world. To address the urgent need to maintain and restore populations of endangered freshwater mussels, quantitative population dynamics modeling is needed to evaluate population status and guide the management of endangered freshwater mussels. One endangered mussel species, the oyster mussel (Epioblasma capsaeformis), was selected to study its population dynamics for my research. The analysis was based on two datasets, length frequency data from annual surveys conducted at three sites in Clinch River: Wallen Bend (Clinch River Mile 192) from 2004-2010, Frost Ford (CRM 182) from 2005 to 2010 and Swan Island (CRM 172) from 2005 to 2010, and age-length data based on shell thin-sections. Three hypothetical scenarios were assumed in model estimations: (1) constant natural mortality; (2) one constant natural mortality rate for young mussels and another one for adult mussels; (3) age-specific natural mortality. A Bayesian approach was used to analyze the age-structured models and a Bayesian model averaging approach was applied to average the results by weighting each model using the deviance information criterion (DIC). A risk assessment was conducted to evaluate alternative restoration strategies for E. capsaeformis. The results indicated that releasing adult mussels was the quickest way to increase mussel population size and increasing survival and fertility of young mussels was a suitable way to restore mussel populations in the long term. The population of E. capsaeformis at Frost Ford had a lower risk of decline compared with the populations at Wallen Bend and Swan Island. Passive integrated transponder (PIT) tags were applied in my fieldwork to monitor the translocation efficiency of E. capsaeformis and Actinonaias pectorosa at Cleveland Islands (CRM 270.8). Hierarchical Bayesian models were developed to address the individual variability and sex-related differences in growth. In model selection, the model considering individual variability and sex-related differences (if a species has sexual dimorphism) yielded the lowest DIC value. The results from the best model showed that the mean asymptotic length and mean growth rate of female E. capsaeformis were 45.34 mm and 0.279, which were higher than values estimated for males (42.09 mm and 0.216). The mean asymptotic length and mean growth rate for A. pectorosa were 104.2 mm and 0.063, respectively. To test for the existence of individual and sex-related variability in survival and recapture rates, Bayesian models were developed to address the variability in the analysis of the mark-recapture data of E. capsaeformis and A. pectorosa. DIC was used to compare different models. The median survival rates of male E. capsaeformis, female E. capsaeformis and A. pectorosa were high (>87%, >74% and >91%), indicating that the habitat at Cleveland Islands was suitable for these two mussel species within this survey duration. In addition, the median recapture rates for E. capsaeformis and A. pectorosa were >93% and >96%, indicating that the PIT tag technique provided an efficient monitoring approach. According to model comparison results, the non-hierarchical model or the model with sex--related differences (if a species is sexually dimorphic) in survival rate was suggested for analyzing mark-recapture data when sample sizes are small. / Master of Science

Risk assessment of organised crime groups can assess different types of risk at a group- or individual-level. Operational definitions of key concepts, units of analysis and type of data are a few issues prevalent in risk assessment (RA) instruments when assessing risk in gangs on a group-level. Do these methodological issues also affect risk assessment in OCG members on an individual level? This review will also focus on how risk assessment instruments are used when assessing individual risk in gang members. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Five databases were included in the review. The abstract screening process yielded 47 studies suitable for full-text screening. Only 8 studies qualified for inclusion after applying the eligibility criteria. A narrative synthesis revealed that risk assessment in OCG members were focused on an adolescent population within a mainly western cultural setting. The RA tools were mostly used in custody settings but also took place in the community and in schools. Variations on RA instruments based on Level of Service Inventory (Andrews & Bonta, 2010) were the most common ones, but other frameworks and models assessing risk were also used. Similar methodological issues were observed on an individual-risk assessment level as on a group level, resulting in a negative impact on validity of RA instruments when used on OCG members.